Design and performance investigation of embedded dual-substrate fiber Bragg grating strain sensor for asphalt pavement

1 Introduction

As the dominant form of highway infrastructure, the structural health and service performance of asphalt pavements are directly related to road traffic safety, driving comfort, and life-cycle maintenance costs (Liu et al., 2023; Xin JY. et al., 2025). In recent years, with the rapid development of the transportation sector, the continuous improvement of the highway network, the sustained growth of traffic volume, and the increasing vehicle axle loads, asphalt pavement structures are facing increasingly severe challenges (Liang et al., 2023; Lu et al., 2024). Under the coupled effects of long-term repeated traffic loads and environmental factors such as temperature and humidity, asphalt pavements are prone to structural distresses including fatigue cracking, rutting, and reflection cracking. Such distresses not only accelerate pavement performance degradation and shorten service life but also pose potential threats to driving safety (Tian et al., 2025; Guo et al., 2022; Zhang et al., 2023). Therefore, achieving real-time, accurate, and long-term monitoring of the internal mechanical state of asphalt pavement structures is of great engineering significance for revealing damage mechanisms, evaluating structural health conditions, early-warning potential risks, and guiding scientific maintenance decision-making.

The reliability of pavement structure health monitoring relies heavily on sensing technologies (Xin X. et al., 2025). Traditional health monitoring methods for asphalt pavements are mainly classified into two categories: non-embedded and embedded sensor monitoring (Hou et al., 2021). Non-embedded monitoring (i.e., non-destructive testing (NDT) techniques) acquires response data via external equipment without damaging the pavement structure (Wang S. et al., 2022). For instance, Alani et al. systematically verified the effectiveness of 3D ground-penetrating radar (3D-GPR) in identifying internal distresses within pavement structural layers (Abdelsamei et al., 2024). Zhu et al. combined discrete element simulation and 3D-GPR to accurately identify the propagation mechanism of reflection cracks and the hydromechanical coupling degradation process (Zhu et al., 2025); Ozden et al. evaluated pavement deformation rates using SAR/InSAR technology (Ozden et al., 2016); Hao et al. proposed an acoustic NDT method based on A-weighted sound pressure level, realizing the accurate identification of thin-layer debonding defects in semi-rigid base asphalt pavements (Hao et al., 2025). Such techniques are characterized by non-destructiveness, high efficiency and easy implementation, making them suitable for large-scale rapid screening (Kamruzaman, 2021; Liu and Gu, 2023). However, their data collection is easily constrained by weather conditions, making long-term continuous monitoring difficult, and they cannot directly reflect the internal mechanical behavior and distress mechanism of the structure (Kaartinen et al., 2022). Embedded monitoring embeds sensors directly into the pavement structure to capture key mechanical responses (e.g., strain, stress, temperature) in real time (Wang P. et al., 2022). Commonly used sensors include resistance strain gauges, vibrating wire sensors, fiber Bragg grating (FBG) sensors, etc. (Di Graziano et al., 2020). Alessandro Di Graziano et al. assessed the application performance of smart sensing devices such as vibrating wire sensors and FBG sensors in asphalt pavement structure health monitoring (Di Graziano et al., 2020); Fahad et al. systematically investigated the application status and engineering applicability of pavement sensing systems incorporating resistance strain gauges and vibrating wire sensors (Moham et al., 2022); Wang et al. analyzed the strain-deformation-temperature transfer law of cement concrete pavements using improved FBG sensors (Wang et al., 2024); Liu et al. confirmed that FBG sensors are significantly superior to resistance sensors in measuring horizontal strain (Liu Z. et al., 2022). Embedded technologies have the advantage of capturing internal structural responses continuously in real time, providing direct data support for mechanism analysis and prediction (Han et al., 2023). Nevertheless, most existing embedded strain sensors adopt a single-point independent layout mode. Large-scale matrix deployment is required to achieve full-area monitoring, resulting in complex construction technology, high cost and insufficient data consistency, which restricts systematic analysis (Wijaya et al., 2021).

Fiber Bragg Grating (FBG) sensing technology presents the advantages of enabling quasi-distributed and distributed measurement (Braunfelds et al., 2022; Golmohammadi et al., 2025; Golmohammadi et al., 2026), providing a new technical pathway for long-term strain monitoring of asphalt pavements. Wijaya et al. (Wijaya et al., 2021), Gu et al. (Sun et al., 2025), Liu et al. (Gu et al., 2025), Meng et al. (Liu Y. et al., 2022), Lu et al. (Meng et al., 2017), Zhao et al. (Lu and Zhang, 2025) and Gong et al. (Zhao et al., 2024) developed distributed optical fiber sensing systems with various packaging materials (e.g., stainless steel tube packaging (Wijaya et al., 2021), flexible packaging (Gu et al., 2025), silicone rubber packaging (Zhao et al., 2024)) on the basis of quasi-distributed and distributed optical fiber sensing techniques. Nevertheless, current quasi-distributed and distributed FBG sensors still suffer from bottlenecks in engineering application: on one hand, the coordinated deformation capability between the sensor packaging structure and asphalt mixture is insufficient, resulting in reduced strain transfer efficiency and degraded durability; on the other hand, linear monitoring can hardly achieve multi-directional synchronous monitoring with complicated fabrication procedures (Gong et al., 2019).

Overall, existing research on embedded FBG strain sensors for asphalt pavements has yielded substantial advances in packaging processes, networking configurations and performance testing, yet no integrated solution well adapted to the service conditions of asphalt pavements has been established.

Most relevant studies pursue engineering improvements around a single performance metric: they either optimize packaging materials to enhance deformation compatibility, or adjust fiber routing to increase the number of measuring points. As a result, the three core bottlenecks—low deployment efficiency, difficulty in multi-directional synchronous monitoring, and poor deformation matching between sensor substrates and the pavement—remain unsolved simultaneously, leaving significant room for improvement in the overall engineering performance of such sensors.

To address these technical gaps, this paper proposes an embedded dual-substrate fiber Bragg grating (FBG) strain sensor. The design delivers systematic integrated innovation across three dimensions: structure, networking, and material. Its layered dual-substrate layout integrates both longitudinal and transverse sensing points, enabling synchronous acquisition of bi-directional strains in a single embedding operation and greatly streamlining construction for multi-directional monitoring. A U-shaped single-loop fiber network is employed to minimize optical loss and reduce fusion splices while maintaining adequate point coverage, which boosts the long-term service stability of the sensing network. Low-modulus glass fiber reinforced polymer (GFRP) is chosen as the substrate material, balancing load-bearing strength and deformation compatibility and effectively alleviating strain transfer distortion caused by modulus mismatch between metallic substrates and asphalt mixtures.

To clearly demonstrate the performance differences among various technical routes and highlight the innovative value of the proposed scheme, Table 1 presents a systematic comparison of four typical pavement FBG sensing solutions across six dimensions: sensing direction, sensing density, packaging scheme, installation complexity, bi-directional monitoring capability, and long-term reliability.

TABLE 1. Comparison of different types of FBG strain sensing schemes.

Building on the above analysis, this study first selects the optimal fiber networking scheme through laboratory optical loss tests. It then determines the substrate dimensions and material by combining finite element simulations with mechanical property tests, and fabricates the sensor using a hybrid bonding and mechanical fixation process. Finally, a systematic test program covering temperature-sensing performance over a wide temperature range, mechanical sensing behavior and waterproof durability is carried out to verify the sensor’s feasibility for engineering applications.

2 Research ideas and methods

At present, to monitor the strain of asphalt pavement along the traffic direction and perpendicular to the traffic direction, strain sensors are usually arranged in a matrix combining longitudinal and transverse layouts, most of which are point-type sensors (Liu et al., 2024; Pan et al., 2021). However, such traditional arrangement involves complicated procedures. Each sensor requires independent positioning, slotting and embedding, as shown in Figure 1, which leads to tedious operations and low construction efficiency. Meanwhile, this scheme is costly, and the data consistency obtained by each independently packaged sensor is poor, which is not conducive to systematic comparison and analysis.

To solve the above problems, this study designs an embedded strain sensor based on a dual-substrate structure. The sensor arranges longitudinally and transversely distributed Fiber Bragg Grating (FBG) measuring points on two substrates respectively, which together form a synchronous monitoring network. Its overall structure is shown in Figure 2. On this basis, three optical fiber networking layouts are proposed.

2.1 Double U-shaped networking layout mode

This layout is a single-board parallel optical fiber networking scheme, whose core is to achieve multi-point coverage inside the substrate via two symmetric “U”-shaped FBG loops, as shown in Figure 3. Two transmission optical fibers are synchronously introduced from the center of the substrate top, extend to both sides to form inner and outer double U-shaped loops (small inner U and large outer U), and finally exit from the center of the top. The measuring points are mainly distributed at the horizontal bottom sections of the double U-shape and the vertical side sections of the inner and outer U-shapes, and arranged independently in a T-shape. A single substrate is equipped with 6 measuring points, with 3 measuring points connected in series for each optical fiber, forming a symmetric distributed optical fiber sensing structure.

2.2 “A-U”-shaped networking layout mode

This layout is a differentiated series-connected optical fiber networking scheme based on dual-substrates, which achieves multi-point coverage inside Substrate A and Substrate B through “A”-shaped and “U”-shaped FBG sensing loops, respectively, as shown in Figure 4. The transmission optical fibers are both introduced from the center of the top of Substrate A and Substrate B: Substrate A adopts an “A”-shaped path, where the optical fiber extends symmetrically to the left and right sides from the center of the top to form an “A”-shaped loop; Substrate B adopts a “U”-shaped path, where the optical fiber extends from the center of the top to form a complete “U”-shaped loop. The two paths are finally led out from the center of the top, forming independent series-connected sensing networks. In terms of measuring point layout, Substrate A and Substrate B are each equipped with 3 series-connected measuring points, which are located at the middle of the left and right sides and the middle of the bottom of the substrates, respectively, presenting an overall “T”-shaped distribution.

2.3 U-shaped networking layout mode

This layout is a differentiated single-loop U-shaped optical fiber networking scheme based on dual-substrates. The core is to form a U-shaped transmission path inside Substrate A and Substrate B through the transmission optical fiber, covering four directions of each substrate, as shown in Figure 5. The transmission optical fiber is introduced from the center of the top of Substrate A and Substrate B, extends leftward along the top horizontal section, downward to form the left vertical section, then rightward along the bottom horizontal section, and upward to form the right vertical section, finally returning to the center of the top, thus forming a closed U-shaped loop. In terms of measuring point layout, Substrate A and Substrate B are each equipped with 2 series-connected measuring points: the measuring points of Substrate A are located at the middle of the left and right vertical sides, while those of Substrate B are located at the middle of the top and bottom horizontal sections, presenting an overall “T”-shaped distribution.

Compared with existing point-type sensors, the embedded dual-substrate fiber Bragg grating (FBG) strain sensor proposed in this study achieves comprehensive advantages of more monitoring points, wider monitoring range, faster anomaly identification, stronger structural durability and higher integration consistency via an innovative networking design.

3 Structural design of embedded dual-substrate fiber bragg grating (FBG) strain sensor

3.1 Comparison of optical fiber networking layout modes

A comparative analysis of the above three networking layout modes was carried out through laboratory optical loss tests. As the developed sensor adopts an embedded structure, the substrate size should be controlled to minimize the influence of the strain sensor on the asphalt pavement. Therefore, the specific layout parameters for each networking mode are set as follows: ① In the double U-shaped networking layout mode, the bending radius of the large U-shape is 20 mm and that of the small U-shape is 10 mm; ② In the “A-U”-shaped networking layout mode, the minimum bending radius of the “M”-shape is 8 mm and that of the U-shape is 20 mm; ③ In the U-shaped networking layout mode, the bending radius is 20 mm. Three repeated tests were conducted for each networking mode. The optical fiber optical loss test is shown in Figure 6, and the specific optical loss results are listed in Table 2.

TABLE 2. Test results of optical loss for networking layouts.

As shown in Table 2, the three networking schemes differ noticeably in optical loss performance, driven by variations in fiber bending radius and layout structure. The tests adopted the relative loss method, with system calibration referenced to the fiber in its free state, which effectively compensated for systematic errors such as the instrument’s absolute reading deviation. Statistical checks reveal that for all groups, the standard deviation across three repeated tests is no greater than 0.012 dB, and the coefficient of variation is ≤0.10%. Such low data scatter demonstrates the stability of the test setup and the reliability of the results. The free state describes an ideal transmission condition with no external force or bending constraint applied to the fiber. Over three repeated measurements, its optical throughput held steady at 11.80 dB, with a standard deviation of 0 and a total fiber loss of 0.00 dB—providing a precise quantitative baseline for optical loss analysis of each networking layout.

In the double U-shaped networking layout mode, the optical loss characteristics of the large U-shaped and small U-shaped structures differ significantly, and the data from the three repeated tests shows good consistency. The large U-shape produces only a minor loss of 0.02 dB relative to the free state, indicating that its bending radius is reasonably designed with no obvious macro-bending loss, and the optical signal transmission is barely affected. In contrast, restricted by structural space, the small U-shape has a smaller bending radius, causing the total optical fiber loss to surge to 1.80 dB. This loss exceeds the typical optical loss tolerance of fiber grating sensor networks used in road engineering, resulting in significant optical signal attenuation and directly impairing the accurate demodulation and stable transmission of strain data at grating measuring points. In addition, this mode requires two optical fibers (large and small U-shapes) to be routed simultaneously on the same substrate. The limited substrate space greatly increases the technical complexity of fiber installation, which not only raises on-site construction difficulty and reduces efficiency but also increases fiber material consumption and the number of external fusion splices. This directly raises the material and construction costs of the network and introduces further risks. As weak points in optical fiber transmission, fusion splices are prone to loosening and breakage under the construction loads of asphalt pavement paving and rolling, as well as long-term vehicle dynamic loads. This greatly increases the failure risk of the sensor network and reduces the reliability of long-term monitoring.

In the “A-U” shaped networking layout mode, the optical loss performances of the “A”-shaped and “U”-shaped structures show a sharp contrast. The three test data of each structure have no obvious fluctuation, and the loss characteristics are definite. The total optical fiber loss of the “U”-shaped structure is only 0.03 dB, indicating a good optical signal transmission state. In contrast, the “A”-shaped structure involves small-radius bends due to its structural design, resulting in a total optical fiber loss as high as 3.77 dB, far exceeding the acceptable level for engineering applications. Severe macro-bending loss directly leads to signal interruption at the grating measuring points on this structure, making effective strain monitoring impossible. The structural design defects of this mode lead to extremely poor stability of the overall optical loss of the network, which can hardly meet the core requirement of stable operation of all sensor measuring points in road engineering. Meanwhile, its grating measuring points adopt a multi-node series connection design. Although the measuring point coverage is improved to a certain extent, an excessive number of series connections not only greatly increases the procurement cost of fiber grating devices but also reduces the fault tolerance of the network. Fiber breakage or grating failure at a single measuring point may cause monitoring interruption of the entire series of measuring points.

In the U-shaped networking layout mode, the optical loss performances of the structures on Substrate A and Substrate B are both close to those in the free state, and the data from the three repeated tests fluctuate slightly around the baseline value, showing excellent and stable optical transmission characteristics. The total optical fiber loss of Substrate A is 0.00 dB, which is completely consistent with the optical transmission effect in the free state; the total optical fiber loss of Substrate B is only 0.01 dB, and the loss is negligible. This quantitative result indicates that the fiber bending radius of the U-shaped networking mode is reasonably designed without macro-bending loss, which fundamentally guarantees the stable transmission of optical signals and provides a key guarantee for the accurate collection of strain data at grating measuring points. Meanwhile, the structural design without small-radius bends greatly reduces the technical difficulty of fiber layout, with strong on-site operability, which can adapt to the complex operating conditions of asphalt pavement construction and effectively improve construction efficiency. In addition, this mode does not require multi-fiber parallel layout and multi-measuring-point series connection, which reduces the consumption of optical fiber materials and fusion splicing devices and significantly lowers the comprehensive cost of networking. The reduction in the number of fusion splices also further reduces the failure risk in engineering applications, greatly improving the long-term stability and operation and maintenance convenience of the sensor network.

From the perspective of long-term engineering service performance, the U-shaped layout offers notable advantages across multiple dimensions. Structurally, it features a symmetric folded design across two substrates, with the optical fiber fully embedded in the substrate grooves along its entire length. This configuration shields the fiber from impact loads during asphalt paving and rolling as well as shear forces from the surrounding matrix, delivering enhanced mechanical robustness. Meanwhile, its low initial optical loss provides ample margin for additional loss caused by long-term structural deformation.

In terms of strain transfer, the layout is dominated by straight segments supplemented with large-radius bends, which simplifies the stress state of the optical fiber from a structural standpoint. Coupled with the packaging process—spot fixation with UV adhesive followed by groove filling and full bonding with epoxy adhesive—sufficient interfacial adhesion is guaranteed. This creates a continuous, unbroken strain transfer path and more stable transfer efficiency.

Based on the quantitative optical loss test data, structural design rationality of each networking mode, as well as the core requirements of road engineering applications for construction, cost and long-term monitoring, the U-shaped networking layout mode is significantly superior to the double U-shaped and “A-U” shaped modes in optical signal transmission stability. It also features strong construction operability, low networking cost and high long-term monitoring reliability. This mode can comprehensively balance optical loss characteristics with the practical application requirements of road engineering, and is therefore determined as the optimal networking layout mode for this fiber grating sensor.

3.2 Substrate size determination and material selection

3.2.1 Substrate size determination

To determine the substrate size of the sensor, in accordance with the specification that the sensor resolution shall be ≤0.1% FS in Optical Fiber Grating Strain Sensor for Civil Engineering (JG/T 422–2013) (the measuring range of the sensor in this study is −1,500 με to 5,000 με, with a corresponding resolution of 6.5 με), a numerical model of asphalt pavement structure with dimensions of 2.0 m × 2.0 m × 0.3 m (length × width × height) was established using the Abaqus finite element software. The entire model adopts eight-node linear brick elements (C3D8R). This element type can effectively avoid shear locking and is suitable for fine-scale strain calculation of asphalt pavement structures. A zoned mesh refinement strategy is adopted for meshing: the core area under wheel load and the key area for sensor installation are locally refined, while gradually coarsened meshes are applied to the pavement edge areas far from the load. This approach improves computational efficiency while guaranteeing calculation accuracy. The standard highway BZZ-100 axle load is adopted for loading, with a standard tire pressure of 0.7 MPa. The load is applied as static uniformly distributed load on the pavement surface to simulate the stress state of pavement under vehicle loads. Under the applied wheel load, the strain distribution characteristics inside the pavement structure were simulated and analyzed, as shown in Figure 7. The detailed model parameters are listed in Table 3. The variation law of strain with the distance (L) from the load center along the x-axis is presented in Figure 8. The boundary conditions of the model are set as follows: the bottom surface is subjected to fully fixed constraints (simulating rigid support with zero vertical displacement), and the interfaces between different pavement structural layers are assumed to be fully continuous.

TABLE 3. Parameters of asphalt concrete pavement structural layers.

As can be seen from Figure 8, when the distance from the load center is 29.4 cm, the strain value at the bottom of the asphalt pavement surface layer is 6.54 με (slightly higher than the sensor resolution of 6.5 με), indicating that this position is at the critical state identifiable by the sensor. In accordance with the requirement that asphalt pavement strain sensors shall be arranged with a horizontal spacing of 60 cm × 60 cm specified in Scientific Observation Network for Long-Term Performance of Highway Infrastructure—Part 1: Construction Specification (JT/T 1504.1–2024), the sensor is placed at the center of the grid in this study (Figure 8).

Considering the most unfavorable working condition (wheel load acting on the edge of the layout area), the theoretical minimum gauge length of the sensor is derived as (60/2–29.4) × 2 = 1.2 cm. This value only reflects the minimum sensing space requirement under extreme load. However, Optical Fiber Grating Strain Sensor for Civil Engineering (JG/T 422–2013) clearly specifies that the gauge length of pavement strain sensors shall be controlled within 5.0–25.0 cm. This specification comprehensively takes into account the non-uniformity of pavement strain, measurement accuracy and long-term stability. Therefore, the theoretical lower limit of 1.2 cm cannot be used for the substrate size.

On the premise of meeting the specification requirements, this study optimizes the substrate size under three constraints: (1) Networking space constraint: Ensure the bending radius required for fiber routing and packaging, so as to avoid excessive loss caused by an over-small bending. (2) Engineering practice constraint: Refer to the typical dimensions of asphalt pavement strain sensors used in similar projects. (3) Process feasibility constraint: Guarantee the operability of substrate fabrication, fiber bonding and packaging procedures, while satisfying the measurement accuracy requirements.

Based on the above conditions, the size of a single substrate is finally determined as 12 cm × 12 cm × 0.3 cm (length × width × height), which not only meets the standard gauge length requirement but also balances engineering practicality and process feasibility.

3.2.2 Substrate material selection and performance analysis

To evaluate the applicability of different substrate materials, three typical materials—titanium alloy (Ti), spring steel, and glass-fiber-reinforced polymer (GFRP)—were selected in this study. Tensile mechanical property tests were conducted using a microcomputer-controlled universal electronic testing machine with high–low temperature capability. A loading rate of 10 mm/min was used for tensile strength testing, and 2 mm/min for elastic modulus testing, to ensure accurate strain data acquisition. No fewer than 5 valid specimens were prepared for each material. The mechanical property test results of the three materials are presented in Figure 9.

As shown in Figure 9, in terms of tensile strength, the average value of spring steel reaches 1,221.03 MPa, which is the highest among the three materials, followed by titanium alloy (1,026.54 MPa), and GFRP is the lowest (600.51 MPa). From the numerical comparison, the tensile strength of spring steel is approximately 18.9% higher than that of titanium alloy and 103.3% higher than that of GFRP; the tensile strength of titanium alloy is about 71.0% higher than that of GFRP. Even GFRP, the material with the lowest strength, has a tensile strength of 600.51 MPa, which is far higher than that of conventional pavement matrix materials such as asphalt and aggregate. It has sufficient strength safety margin and fully meets the stress requirements of pavement structures. Meanwhile, owing to its non-metallic composite characteristics, GFRP has excellent corrosion resistance that titanium alloy and spring steel do not have, ensuring a longer service life in harsh pavement environments such as salt corrosion and humidity.

In terms of elastic modulus, spring steel has a modulus of 207.86 GPa, which is 1.92 times that of titanium alloy (108.24 GPa) and 8.17 times that of GFRP (25.44 GPa); the modulus of titanium alloy is 4.25 times that of GFRP. The high modulus of spring steel provides high stiffness and outstanding load diffusion capacity, but it matches poorly with the low-modulus, large-deformation characteristics of asphalt pavement, and is prone to interfacial debonding and stress concentration caused by asynchronous deformation. Titanium alloy has a moderately high modulus with ordinary deformation compatibility with the pavement, and as a metallic material, it carries the risk of long-term corrosion failure. In contrast, GFRP has an elastic modulus of only 25.44 GPa, which matches the deformation characteristics of asphalt pavement optimally. While ensuring tensile load-bearing capacity, it can better deform coordinately with the pavement, which theoretically reduces the probability of distresses such as interlayer debonding and fatigue cracking.

To further determine the substrate material, an asphalt concrete beam model with an embedded substrate was established using Abaqus software. The model dimensions are 500 mm × 150 mm × 150 mm (length × width × height). The material parameters of the structural layer are: density 2.440 g/cm³; dynamic modulus 8,500 MPa; Poisson’s ratio μ = 0.35. The sensor substrate is arranged at the center of the beam, with an embedding depth of 10 cm. Surface-to-surface contact is adopted between the sensor and the asphalt mixture. The normal contact is defined as “hard contact”, allowing separation after contact; the tangential contact uses the penalty function friction model with a friction coefficient of 0.7. The stress distribution contours of the beam specimen and the substrate are shown in Figures 10, 11, and the influence of different substrate moduli on the stress of the asphalt concrete beam specimen is presented in Figure 12.

As shown in Figure 11, the substrate modulus is a key parameter regulating the interfacial mechanical properties between asphalt mixture and sensor substrate. As the substrate modulus increases from 0 to 140 GPa, the average interfacial stress rises from 1.6 MPa to 16.6 MPa, the stress concentration factor (β, β = σ n σ 0) increases gradually from 3.6 to 10.1, and the cooperative deformation coefficient (γ, γ = β n + 1 − β n β n) decreases from 0.52 to below 0.01. This coefficient directly quantifies the sensor’s macroscopic strain transfer efficiency. Strain in the asphalt matrix is first transferred to the substrate via interfacial shear, and then further transmitted to the embedded FBG gratings through the substrate. A higher γ indicates better deformation synchrony between the substrate and the matrix, smaller interfacial shear slip, more complete strain transfer from the matrix to the sensing element, and thus higher monitoring accuracy. Conversely, a lower γ corresponds to degraded strain transfer efficiency, causing monitoring results to deviate significantly from the actual strain state of the pavement. When the modulus is ≥60 GPa, the γ value is always lower than 0.05, indicating that the deformation compatibility between the high-modulus substrate and asphalt mixture is reduced. The substrate acts as a rigid constraint body, aggravating stress concentration instead of bearing load together. In terms of material properties, GFRP has an elastic modulus of 25.44 GPa, which is lower than the critical threshold of 60 GPa and falls within the advantage zone of cooperative deformation. Its tensile strength is 600.51 MPa, meeting the strength requirements of pavement structures. Meanwhile, the low modulus endows it with excellent deformation compatibility, which can effectively alleviate interfacial stress concentration. In addition, as a non-metallic material, GFRP can avoid corrosion and is suitable for corrosive environments such as humidity and salt erosion. Titanium alloy has an elastic modulus of 108.24 GPa, whichfalls into the rigid constraint zone. Although its tensile strength reaches 1,026.54 MPa, the cooperative deformation coefficient is only 0.02–0.03, making the interface prone to debonding and fatigue damage. The corrosion risk of metallic materials also affects its long-term service reliability. spring steel has an extremely high elastic modulus of 207.85 GPa, far exceeding 140 GPa, and is located in the stress concentration saturation zone. The β value is close to 10.1 and the γ value approaches 0, meaning it basically loses cooperative deformation ability, significantly increases the local stress gradient of asphalt mixture, and raises the risk of interfacial cracking.

To sum up, the tensile strength of all three materials can meet the pavement bearing requirements, but their matching degrees of elastic modulus with asphalt mixture differ significantly. GFRP achieves the optimal balance among tensile strength, modulus compatibility and corrosion resistance, making it suitable as the substrate material for pavement monitoring sensors in corrosive environments. Titanium alloy and spring steel deteriorate the interfacial mechanical state and increase the probability of pavement diseases due to their excessively high modulus. Therefore, GFRP is the preferred choice.

3.3 Sensor fabrication process

Based on the substrate size design, material selection, and FBG networking layout scheme for the embedded strain sensor, an embedded strain sensor with a dual-substrate structure was successfully fabricated. Its overall structure and fabrication process are shown in Figure 12.

3.3.1 Substrate pretreatment and FBG selection

The substrate adopts prefabricated glass-fiber-reinforced polymer (GFRP) sheets with dimensions of 120.0 mm × 120.0 mm × 3.0 mm (length × width × height). A strip groove with a width of 1.5 mm and a depth of 1.0 mm is formed by milling on the plate surface. Fixed round holes of 3.2 mm in diameter are drilled at the four corners and the optical fiber outlet. The surface is flat without defects such as cracking or delamination, and the dimensional error is controlled within ±0.1 mm. Bare FBGs with a grating length of 10 mm are selected, ensuring no damage or fracture and a clean surface. Two groups of grating measuring points correspond to Substrate A and Substrate B respectively: Substrate A is equipped with gratings of 1,530 nm (Measuring Point 30) and 1,535 nm (Measuring Point 35); Substrate B is equipped with gratings of 1,540 nm (Measuring Point 40) and 1,545 nm (Measuring Point 45). The centers of the grating measuring points are calibrated in the milled groove using a steel ruler. The marking lines are fine and distinct to avoid damaging the groove wall, ensuring that the positioning accuracy meets the design requirements.

3.3.2 Precise positioning and curing process of optical fiber

Optical fiber fixation is a critical step to ensure the strain transfer efficiency of the sensor. The optical fiber and grating must be precisely fixed in the milled groove of the substrate to avoid damage and relative sliding. First, wipe the surface of the optical fiber with alcohol to remove dust and impurities, and recheck the integrity of the grating region. Then, embed and position the optical fiber into the milled groove using UV curable adhesive, keeping the fiber straight without bending or twisting to prevent wavelength shift caused by stress concentration in the grating region. After positioning, irradiate the UV curable adhesive in the groove with an ultraviolet lamp for 10–15 s to achieve complete curing and preliminary fixation of the optical fiber and grating. After fixation, check the fitting degree between the optical fiber and the groove wall to confirm no looseness or offset, and ensure the positions of the grating measuring points meet the design requirements.

3.3.3 Uniform sealing of adhesive layer and pretreatment of bonding surfaces

The gluing process includes two steps: milling groove filling and coating on the bonding surfaces of substrates, aiming to seal the optical fiber, enhance the fixing effect, and provide a reliable interface for substrate assembly. After the preliminary fixation of the optical fiber, low-temperature curing epoxy adhesive is slowly injected into the milling groove to ensure that the voids are fully filled without bubbles or cavities, and the filling height is flush with the groove depth, realizing complete encapsulation and sealing of the optical fiber. After groove filling, the same type of epoxy adhesive is evenly applied to the bonding surfaces of Substrate A and Substrate B with a brush, with the thickness controlled at 0.1–0.2 mm, covering the entire area to avoid missing coating or adhesive accumulation. Allow a short standing time after gluing to prevent premature curing of the adhesive layer and prepare for the subsequent substrate bonding.

3.3.4 Assembly and molding

Assembly and molding is the final forming stage of the sensor. The glued Substrates A and B are assembled and fixed together with the optical fiber, and stress-bearing supports are installed to ensure a robust sensor structure that can deform coordinately with the pavement. A dual combination mode of bonding + mechanical fixation is adopted for substrate assembly. Substrate A is precisely placed over Substrate B to ensure perfect alignment without dislocation, so that the bonding surfaces fit tightly and the adhesive layer is uniform without voids. After alignment, M3 stainless steel bolts are inserted into the fixing holes at the four corners and tightened with a torque wrench to further strengthen the connection. The assembly is left standing at room temperature for 24 h to allow complete curing of the epoxy adhesive and achieve firm fixation of the substrates. The finished sensor has Substrate A as the front side and Substrate B as the back side, with a flat overall structure. To improve the integrity between the sensor and the pavement and avoid stress concentration at the top corners or excessive displacement under load, stainless steel stress-bearing supports (10 mm × 10 mm × 5 mm) are installed at the four corners of the sensor. The bottom of the supports is flush with the back surface of the substrates and fixed by bolts. The connection reliability of the supports is checked after installation to ensure no looseness or detachment.

After final assembly, each sensor receives a full visual inspection and dimensional check to rule out defects including cracking, delamination, loose optical fibers and adhesive peeling, and to confirm all dimensions meet design requirements. To ensure consistent performance and process repeatability for batch production, clear tolerance specifications were defined for all core fabrication steps. The overall dimensions, milled grooves and fixing holes of GFRP substrates are all machined by CNC milling, with tolerances kept within ±0.1 mm for overall geometry, ≤ ±0.05 mm for groove dimensions, and ≤ ±0.1 mm for hole positioning. Grating measurement points are located via scale calibration, with positional deviation limited to ±0.5 mm. Standardized parameters are enforced for UV adhesive curing time, epoxy adhesive layer thickness (0.1–0.2 mm) and bolt tightening torque, minimizing performance scatter introduced by manual operation across both substrate machining and assembly processes. Samples passing quality control inspection then move on to performance testing to verify their practical strain monitoring capability.

4 Performance study on embedded dual-substrate FBG strain sensor

4.1 Temperature-sensing performance

To analyze the temperature-sensing performance of this strain sensor, it was placed in a programmable constant temperature and humidity test chamber for heating tests. The test temperature ranged from −10 °C to 80 °C (−10 °C is the minimum temperature of the equipment, and 80 °C exceeds the extreme service temperature of asphalt pavement), with a heating interval of 10 °C. The temperature was kept constant at each target point for 30 min, and the test was repeated 3 times. During the test, the sensor was connected to a fiber Bragg grating interrogator (Nantong Zhixing Technology Co., Ltd., model ZX-FP-C04-1) via transmission optical fiber. The interrogator was linked to a host computer through Ethernet, enabling real-time acquisition and storage of wavelength data. The test process and results are shown in Figures 13, 14.

As shown in Figure 14, within the temperature range of −10 °C–80 °C, the wavelength variation (Δλ) at each measuring point of the strain sensor increases monotonically with rising temperature, showing a high linear relationship (all linear correlation coefficients R ^*2* > 0.999). These results indicate that the fiber Bragg grating (FBG) measuring points inside the sensor present good synergy with the GFRP substrates during thermal deformation, verifying the sensor’s structural integrity and excellent interfacial bonding performance. This is mainly because the epoxy adhesive used has favorable thermal stability and bonding strength, without softening, brittle cracking or debonding under high and low temperature environments, thus effectively ensuring no relative slippage at the FBG–substrate interface. In addition, the thermal expansion coefficients of the GFRP substrates and the optical fiber are well matched, leading to consistent thermal deformation during temperature changes. This effectively avoids excessive interfacial stress caused by mismatched deformation and guarantees the overall cooperative deformation capacity of the sensor. The temperature sensitivity coefficient (K _*T*) of each measuring point is generally low: 0.030 nm/°C for point 30, 0.031 nm/°C for point 35, and 0.027 nm/°C for both points 40 and 45. All channels deliver good measurement repeatability, and K _*T* values show little variation across points, with a maximum difference of only 0.004 nm/°C. The average temperature sensitivity of the four points is 0.029 nm/°C, with a coefficient of variation of approximately 7.0%.These results demonstrate excellent uniformity in temperature-sensing characteristics across different measuring channels, along with sound thermal insulation and temperature interference resistance, meaning the sensor is minimally affected by ambient temperature fluctuations in practical engineering service. The main reasons are: (1) The low thermal conductivity of GFRP substrates effectively slows down internal heat conduction; (2) The sealant protection at the optical fiber lead-out end further blocks external heat exchange and strengthens the overall thermal insulation effect. From an engineering application perspective, the low K _*T* has important practical value. On one hand, in asphalt pavement monitoring, the sensor does not need complex temperature-compensated optical paths or multi-sensor combined compensation algorithms; accurate correction of temperature interference can be achieved only by simple linear fitting, which greatly simplifies on-site data processing and improves monitoring efficiency. On the other hand, the highly consistent K _*T* values at each measuring point effectively avoid data distortion caused by sensitivity differences, ensuring the comparability and reliability of synchronous monitoring results from multiple measuring points.

In summary, the strain sensor exhibits low temperature sensitivity, high linearity and consistent response at each measuring point over a wide temperature range, which validates the rationality of its structural design, material selection and fabrication process.

4.2 Mechanical sensing performance

To analyze the mechanical sensing performance of the strain sensor, vertical loading tests were carried out using a universal testing machine to investigate the relationship between the wavelength variation (λ) at each measuring point and the loading displacement. During the test, the sensor was placed with its front side facing upward (Substrate A on top). Displacement control was adopted for loading, with a loading rate of 0.2 mm min^-1, a displacement loading range of 0.4 mm–1.0 mm, and a loading step of 0.2 mm, the test temperature was maintained at 25 °C ± 1 °C. Each test was repeated 3 times. The detailed test process and results are shown in Figures 15, 16.

As shown in Figure 16, under front-side loading, the wavelength of each measuring point shows a strong linear correlation with the applied strain (R ^*2* ≥ 0.995), confirming the stable and reliable mechanical sensing performance of the sensor and its ability to accurately capture minute strain variations during loading. The fitting results indicate that the equations for points 30 and 35 are y _*30* = 0.0198x+1,529.957 and y _*35* = 0.0199x+1,535.156, respectively, with positive slopes representing a positive wavelength shift as strain increases. For points 40 and 45, the fitting equations are y _*40* = −0.0196x+1,540.131 and y _*45* = −0.0196x+1,544.998, with negative slopes corresponding to a negative wavelength shift with rising strain.

This differential response clearly reflects the stress distribution across the sensor: Substrate A (hosting points 30 and 35) is subjected to tension, while Substrate B (hosting points 40 and 45) undergoes compression. This pattern aligns well with the loading setup and the design expectation of opposing stress states on the two substrates, verifying that the sensor can faithfully and accurately capture the stress-deformation behavior of both the sensor itself and the surrounding matrix.

Derived from the slope of the fitting curves—defined as the mechanical sensing sensitivity K _*m* = Δλ/Δε—the sensitivities of the four points are K _*m30* = 0.0198pm/με, K _*m35* = 0.0199pm/με, K _*m40* = −0.0196pm/με, and K _*m45* = −0.0196pm/με, respectively. The negative sign denotes the compressive response and does not alter the absolute magnitude of sensitivity. Statistical analysis shows that the mean absolute sensitivity of the four points reaches 0.0197 pm/με, with a coefficient of variation of merely 0.8% among points, indicating excellent uniformity in the mechanical response of all measuring channels.

A comparison reveals that fiber sensitivity is slightly higher under tension than under compression. This occurs because axial tension uniformly elongates the grating period, while the photoelastic effect induces a regular reduction in the effective refractive index. The two effects work in concert to produce a pronounced, stable positive shift in the Bragg wavelength, hence delivering high linearity and sensitivity. Under compression, by contrast, axial pressure tends to trigger lateral fiber expansion, local micro-bending or wrinkling, which breaks the uniformity of the grating period. Meanwhile, the photoelastic effect and stress concentration cause fluctuations in the refractive index profile, diminishing the wavelength shift and slightly degrading linearity. As a result, the absolute sensitivity under compression is generally lower than under tension.

Furthermore, the maximum relative difference in Km values across all points is around 1.5%, a low level that reinforces the high consistency of mechanical response with no marked individual variation. Such uniformity helps prevent data distortion caused by excessive sensitivity discrepancies and ensures the comparability and coherence of synchronous multi-point monitoring data. When combined with the low fluctuation of the temperature sensitivity coefficient KT presented earlier, these findings further demonstrate that the sensor delivers consistent performance in both mechanical sensing and temperature interference resistance. Its structural design, material selection and fabrication process achieve well-matched performance across all measuring points, providing a reliable data basis for synchronous multi-region, multi-point health monitoring of asphalt pavements, and laying a solid foundation for subsequent intelligent integration of monitoring data and multi-field coupling error correction.

4.3 Waterproof performance

Although asphalt pavement itself has a certain waterproof capability, it is necessary to investigate the waterproof reliability of the strain sensor to comprehensively analyze its mechanical sensing performance under harsh water environments. Following 1 h and 2 h of water immersion respectively, the strain sensors underwent mechanical loading tests at a controlled temperature of 25 °C ± 1 °C to assess their sensing performance. Each test under different water immersion conditions was repeated 3 times. The test process and results are shown in Figures 17, 18.

As shown in Figure 18, the strain sensors maintain favorable performance stability under the coupled effects of dynamic loading and water immersion after 1 h and 2 h of immersion. The wavelength variation trend at each measuring point stays consistent with that under dry conditions: points 30 and 35 remain in tension with positive wavelength shifts, while points 40 and 45 stay in compression with negative shifts, and all datasets yield strong linear correlations with R ^*2* ≥ 0.992.

This result confirms that the sensor’s stress transfer mechanism remains intact even under complex hydro-mechanical coupling conditions. The structural stress pattern—tension on Substrate A and compression on Substrate B—is not altered or compromised by water immersion. The mechanical response therefore remains reliable in humid environments, with no systematic distortion of sensing signals caused by water ingress.

The mechanical sensitivity K _*mw* (defined as K _*mw* = Δλ/Δε) of each measuring point under dynamic loading ranges from 0.0192 to 0.0200 pm/με. After 1 h of water immersion, the sensitivities are 0.0194 pm/με (point 30), 0.0196 pm/με (point 35), −0.0192 pm/με (point 40), and −0.0192 pm/με (point 45); the negative sign denotes the compressive response and does not affect the absolute sensitivity value. After 2 h of immersion, the corresponding values are 0.0196 pm/με, 0.0200 pm/με, −0.0195 pm/με, and −0.0197 pm/με for points 30 to 45, respectively.

Statistical analysis reveals a maximum relative difference in K _*mw* across all measuring points of only 2.1% after 1 h of immersion and 2.6% after 2 h. The maximum relative difference in sensitivity between the two immersion durations is merely 4%. These figures indicate that water immersion does not impair the consistency of the sensing response across channels, and the comparability and uniformity of synchronous multi-point monitoring are well preserved in humid conditions. Compared with the dry-state sensitivity, the maximum relative difference is only 2%, suggesting that water ingress has a negligible effect on the sensor’s mechanical sensing performance. Neither the waterproof sealing of the encapsulation nor the bonding stability of the fiber-matrix interface degrades after immersion, and stress transfer efficiency remains high with no sensitivity attenuation from water penetration.

During long-term service, asphalt pavements are repeatedly exposed to the combined action of rainwater immersion and vehicle dynamic loads. Water ingress typically causes structural degradation and reduces the mechanical transfer efficiency of the sensing interface. This study demonstrates that the immersed sensor delivers sensing performance highly consistent with its dry-state performance. This not only validates its sound waterproof capability but also indicates that the sensor can maintain stable monitoring performance under prolonged rainwater erosion, effectively suppressing moisture interference with sensing signals. This characteristic provides essential environmental adaptability for full-life health monitoring of asphalt pavements, and establishes an experimental basis for reducing follow-up maintenance costs and improving the reliability of monitoring data.

5 Discussion

To meet the engineering demand for multi-directional synchronous strain monitoring in asphalt pavements, this study presents the structural design, optimized fabrication process, and core laboratory performance evaluation of an embedded dual-substrate FBG strain sensor. The sensor demonstrates reliable performance across optical transmission stability, temperature sensing properties, mechanical response, and waterproof capability, providing a novel integrated technical scheme for asphalt pavement strain monitoring. However, constrained by the current research progress and experimental conditions, the work still has limitations that warrant further improvement in follow-up studies.

5.1 Research limitations

• Performance validation relies primarily on controlled laboratory tests, which differ from actual pavement service conditions. All tests were performed under ideal laboratory conditions with well-controlled single variables and stable environmental settings. In practical engineering, however, sensors are exposed to complex service conditions, including impact loads during asphalt paving and compaction, long-term cyclic traffic loading, alternating temperature and humidity, and hydro-mechanical coupling. In addition, factors such as interlayer constraints in the pavement structure, variations in aggregate gradation, and construction deviations can all affect strain transfer efficiency and long-term reliability. Since laboratory tests cannot fully replicate these field conditions, the in-service performance of the sensor requires further field verification.
• Systematic research on long-term service performance remains insufficient. The present study focuses on calibrating basic sensing performance, and no long-term aging or fatigue tests have been conducted. The long-term lifespan of the sensor is determined by the coupled effects of multiple factors, including substrate material aging, interfacial bonding degradation, and optical fiber bending fatigue damage, all of which require systematic evaluation through dedicated long-term experiments. Although measures such as corrosion-resistant GFRP substrates, low-loss U-shaped fiber routing, and fully sealed epoxy encapsulation have been adopted in the design phase to enhance long-term reliability, the long-term effectiveness of these design strategies still requires experimental validation.
• Mechanical test conditions are relatively simplified and do not fully reflect the actual stress state of pavement structures. Current mechanical tests are centered on basic performance calibration using vertical monotonic frontal loading, which enables accurate measurement of core parameters such as sensitivity and linearity. However, these tests do not cover more realistic loading scenarios such as flexural-tensile loading, shear loading, and cyclic loading, and therefore cannot fully characterize the sensor’s response under multi-axial stress and cyclic dynamic loads. The engineering relevance of the current test framework needs to be further strengthened.
• The evaluation of water resistance is limited in scope. Given that the sensor is designed for embedding at the bottom of the intermediate and lower asphalt layers, and that an intact asphalt pavement provides sufficient waterproofing to prevent long-term water accumulation under normal rainfall, short-term immersion tests of 1 h and 2 h were adopted to simulate occasional water seepage and verify short-term waterproof reliability and mechanical stability. Nevertheless, this test scheme only reflects performance under occasional water exposure and cannot fully represent long-term waterproof durability under complex aqueous environments. The evaluation system therefore requires further refinement.
• No dedicated structural design or systematic analysis of temperature compensation has been carried out. In this study, the interference of temperature fluctuations on laboratory measurements was eliminated by conducting tests under constant temperature conditions. However, no integrated temperature compensation structure has been developed, nor has the impact of temperature cross-sensitivity on strain monitoring accuracy been systematically analyzed over a wide temperature range. For field applications, a differential compensation scheme using an external FBG temperature sensor is proposed, but its actual compensation performance and adaptation accuracy remain to be verified through field tests.
• A systematic life-cycle economic analysis is lacking. Current research focuses on sensor structural development and basic performance verification, with only a preliminary breakdown of core cost components. No quantitative evaluation has been performed from a full life-cycle perspective covering equipment procurement, construction and deployment, operation and maintenance, and service life. There is also a lack of systematic cost comparison with similar monitoring schemes, meaning the economic feasibility of the proposed scheme is not yet fully demonstrated.

5.2 Future research directions

• Embedding tests on asphalt mixture beams and rutting slabs. Sensors will be embedded in asphalt concrete beam and rutting slab specimens to analyze strain transfer behavior and sensing accuracy in the embedded state, verify the coordinated deformation performance between the sensor and the asphalt matrix, and refine the strain transfer correction model.
• Tests on long-term service performance, waterproof durability, and temperature adaptability. Million-cycle fatigue loading, temperature-humidity cycling, and freeze-thaw cycle tests will be conducted to evaluate performance degradation patterns under the coupled effects of sustained loading and environmental factors. Additional tests, including prolonged water immersion, dry-wet cycling, and water-load-temperature multi-field accelerated aging, will be supplemented to reveal the mechanism of performance decay under long-term water exposure. Temperature-strain coupling tests over a wide temperature range will also be performed to clarify the influence of temperature cross-sensitivity, optimize the temperature compensation algorithm, and establish a performance degradation prediction method under multi-factor coupling. Furthermore, performance comparisons across multiple sample batches will be added to quantify the coefficient of variation of core indicators, verify the batch-to-batch stability of the fabrication process, and improve the comprehensive performance evaluation system.
• Field embedding verification and long-term tracking based on a full-scale test pavement. On-site deployment, construction protection, and long-term data acquisition of the sensors will be completed. Differences between laboratory calibration results and field measurements will be compared to develop an engineering application correction method. The long-term performance of the sensors under the combined action of natural environment and traffic loads will be continuously monitored to verify full-life service reliability. In parallel, a life-cycle economic analysis will be conducted to quantify the overall cost covering equipment procurement, construction and deployment, operation and maintenance, and service life. A systematic comparison with conventional monitoring schemes will be performed to strengthen the economic feasibility assessment and provide data support for large-scale application.

6 Conclusion

Aiming at the engineering pain points of traditional point-type strain sensors for asphalt pavements, such as cumbersome layout procedures, insufficient data comparability, low construction efficiency and high cost, this study innovatively proposes an embedded dual-substrate fiber Bragg grating (FBG) strain sensor. The main conclusions and prospects are as follows:

• Three fiber networking configurations—double-U, A-U and U shapes—were developed based on the dual-substrate structure. Optical loss testing confirmed the U-shaped layout as the optimal scheme, yielding total fiber losses of 0.00 dB for Substrate A and 0.01 dB for Substrate B. This design achieves the lowest optical transmission loss, and also benefits from simple fabrication, fewer fusion splices and strong long-term transmission stability, providing a viable structural solution to reduce transmission loss in fiber networks and improve the long-term reliability of optical links.
• Based on an analysis of asphalt pavement strain distribution and finite element numerical simulations, the optimal dimensions of a single sensor substrate were determined as 12 cm × 12 cm × 0.3 cm. Combined with material mechanical property tests and interfacial mechanical simulations, glass fiber reinforced polymer (GFRP) — with an elastic modulus of 25.44 GPa and a tensile strength of 600.51 MPa—was selected as the optimal substrate material. It exhibits outstanding deformation compatibility with asphalt mixtures, effectively mitigating the modulus mismatch and long-term corrosion risks associated with metallic substrates at the material level.
• The “wavelength-temperature” relationship of the strain sensor in the temperature range of −10 °C–80 °C was studied through temperature-sensing performance tests, showing a good linear correlation (R ² > 0.999), indicating that each measuring point of the sensor has stable temperature-sensing characteristics; moreover, the temperature sensitivity K _*T* (0.027 nm/°C ∼ 0.031 nm/°C) of the four measuring points has small differences, indicating that the sensor has high reliability in a wide temperature range.
• Mechanical sensing performance tests were carried out to examine the wavelength-displacement relationship of the strain sensor under dynamic loading. The results show a strong linear correlation (R ^*2* ≥ 0.995), confirming reliable mechanical response at all measuring points. The mechanical sensitivity K _*m* of the four channels falls within a narrow range of 0.245–0.249 nm/mm. Water immersion waterproof tests further reveal that the rate of change in sensitivity K _*mw* is no more than 4% after 2 h of immersion. This preliminarily verifies that the sensor retains stable mechanical sensing performance under short-term water exposure, and its encapsulation structure provides sound short-term watertightness.

This study completes the structural design, fabrication process optimization and laboratory fundamental performance validation of the embedded dual-substrate FBG strain sensor. It offers a novel technical approach for multi-directional synchronous strain monitoring of asphalt pavements, and provides an experimental basis and theoretical reference for subsequent engineering applications.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Mehdi Koohmishi, University of Bojnord, Iran

Reviewed by: Murat Kunelbayev, Institute of Information and Computers Technology, Kazakhstan

Yasir Mahmood, North Dakota State University, United States