Regional standby auto-transfer system based on 5G communication and dynamic time warping: scheme and application

1 Introduction

In recent years, with the strategic deployment of “carbon neutrality and peak carbon emissions” and the rapid development of new energy technologies such as wind power (Xie et al., 2021), photovoltaics (Cherp et al., 2021), and electric vehicles (Kinyemi et al., 2022), more and more distributed generation (DG) sources have been connected to the distribution network (Jung and Dirk, 2022). The distribution network is gradually transitioning from a traditional radial network to a multi-source multi terminal network, and the requirements for relay protection performance are becoming increasingly stringent (Guo et al., 2022; Akhtar et al., 2021). Traditional three-level current protection can no longer accurately and quickly isolate fault points, which may lead to large-scale power outages in regional power grids (Zhang et al., 2022; Xia and Tang, 2023). The automatic backup power supply switching system is currently one of the most effective methods for quickly restoring power supply to the distribution network. However, traditional backup self switching systems require the presence of hot standby open-loop switches within the station, which cannot be used for distribution networks operating in a ring network (Li et al., 2019; Huang et al., 2022). Therefore, there is an urgent need to develop a regional backup automatic transmission system that can meet the panoramic information exchange of regional distribution networks, in order to effectively improve the power supply reliability of regional distribution networks.

Domestic manufacturers of backup automatic switching devices usually use the pressure free method for backup automatic switching. If a large number of distributed power sources are connected to the distribution network system and fail, the existing backup automatic switching devices usually adopt the strategy of quickly cutting off new energy and then restoring power supply through backup automatic switching devices. The main idea is to disconnect new energy to meet the operating conditions of voltage free backup automatic switching (Muzzammel and Arshad, 2022; Chen et al., 2019; Hu and Zheng, 2023). At the same time, the backup automatic switching device will separate the three-phase voltage of the working power bus and use whether there is current in the working power supply incoming line as an auxiliary standard to eliminate the influence of voltage transformer disconnection and ensure the accuracy of signal detection (Liang et al., 2017; Hu, 2024). (Niu et al., 2016) proposed a method for searching for new energy power supply circuits by establishing a node power margin matrix in a node tree. The practicality of this method in new energy distribution networks was verified through simulation using an improved IEEE37 node system. (JI, 2018) proposes that when a certain part of the system malfunctions and causes the new energy to operate on an isolated island, the new energy will be temporarily disconnected, and high-frequency cutting and low-frequency load shedding methods will be used to suppress voltage and frequency fluctuations at the grid connection point. (Yang et al., 2018) ensures islanding operation before waiting for synchronization with the power grid in the event of a failure in the 110 kV voltage series system model. However, when there is an imbalance in power supply and demand within the island operation, it is still necessary to cut off new energy sources. (Liu et al., 2014) proposed a backup automatic switching device based on real-time information calculation to generate switching standards and achieve more accurate adaptive switching. However, this may result in asynchronous closure, causing the device to withstand significant surge currents. The above studies are all aimed at regional power grids with a single voltage level. After the introduction of the 220 kV substation backup automatic switching system and the regional level backup automatic switching system, multiple sets of backup automatic switching devices with different voltage levels and purposes will restore power supply after a fault according to the preset priority action sequence. If the priority order is incorrect, it may trigger multiple backup automatic switching devices to operate simultaneously, resulting in unnecessary mode switching.

To meet the communication needs of regional backup self deployment systems, high bandwidth and low latency 5G wireless communication technology has been introduced as an information transmission channel (Chen, 2022). In recent years, the application of 5G in the power system has gradually expanded from general scenarios such as demand response and intelligent power consumption (Sun et al., 2021; Xiao et al., 2022) to professional fields such as multi terminal protection in distribution networks (Zhou et al., 2022), robust post disaster recovery (Liu et al., 2024a), and unmanned aerial vehicle assisted emergency self-healing (Huang et al., 2025). On this basis, some scholars have further explored the application of 5G in conventional fault scenarios of distribution networks: (Hong et al., 2024; Lu J. et al., 2024) have achieved rapid self-healing after faults through distributed communication and centralized MILP optimization; (Guo and Tuxun, 2024) constructed a distributed control model based on 5G intelligent terminals; (Han et al., 2019; Liu et al., 2024b) proposed online self-healing schemes based on MILP and distributed feeder automation systems, but both schemes focus on isolation and load transfer of feeder level faults, without involving coordination and cooperation between multiple power stations and voltage levels. In terms of adaptive protection, (Chen et al., 2024; Wu et al., 2024) have implemented topology adaptive differential protection for distribution networks using 5G low latency characteristics and sampling point interpolation synchronization method. For the scenario of backup self deployment, (Yin et al., 2023; Baocheng et al., 2024) explored the optimization of self-healing of intra ring interconnection switches and data mapping of communication ring networks, while (Li et al., 2023) introduced 5G routers to achieve point-to-point communication in remote backup self deployment. However, the above-mentioned research on backup self deployment considers 5G as an alternative channel to fiber optics, and has not applied it to regional backup self deployment scenarios with multiple voltage levels and coordinated power stations. It has also not quantitatively analyzed the impact of 5G transmission delay and jitter on backup self deployment criteria, and has not yet achieved deep coordination between communication and control protection.

In summary, this article proposes a research and application of a regional standby auto-transfer system based on 5G communication technology and dynamic time planning (DTW) method. Firstly, based on the limitations of traditional standby auto-transfer systems, the design ideas and implementation functions of multi-level regional standby auto-transfer systems are elaborated in detail. Then, utilizing the high bandwidth, high reliability, and low latency characteristics of 5G communication technology, precise fault location, isolation, and power recovery in the distribution network can be achieved. Introducing DTW method to solve the transmission delay and delay jitter problems in 5G communication technology. Finally, the proposed research method was applied to the regional distribution network demonstration project in Hubei Province, and its effectiveness was verified through the priority coordination and action time of backup automatic switching devices at two voltage levels of 110kV and 35 kV.

The main innovation of this article lies in: (1) proposing a multi-level unified standby auto-transfer system architecture of “station domain region”. A unified backup and self switching system architecture covering four levels has been established - the first level is the backup and self switching layer for the 220 kV power substation, the second level is the backup and self switching layer for the 110 kV area, the third level is the backup and self switching layer for the high-voltage side of the 110 kV substation, and the fourth level is the backup and self switching layer for the 35 kV medium and low-voltage side of the substation. This architecture clarifies the priority coordination relationship between different levels and replaces traditional time step coordination with comprehensive logic control instructions, fundamentally solving the risk of multiple sets of backup automatic switching devices operating simultaneously or coordinating incorrectly. (2) The DTW algorithm is applied for the first time to compensate for delay jitter in regional standby auto-transfer. For the first time, the DTW algorithm was introduced into the electrical quantity comparison criteria for regional backup automatic deployment, effectively eliminating the influence of 5G transmission delay and jitter on the calculation of backup automatic deployment action criteria, and enabling the action speed of regional backup automatic deployment to reach the same millisecond level as that of station domain backup automatic deployment.

2 Traditional standby auto-transfer system

Traditional backup automatic switching systems are usually set according to voltage levels, as shown in Figure 1. Even within the same substation, there may be a lack of information exchange between backup automatic switching devices corresponding to different voltage levels. To ensure coordination and cooperation between different voltage levels, the operating time on the medium and low voltage sides is usually selectively extended. If the automatic activation of the backup power supply on the high voltage side does not meet the operational requirements, power can only be restored on the medium and low voltage sides, but due to its long delay time, rapid response cannot be achieved. If an information exchange mechanism is established between devices of different voltage levels and a direct logical cooperation method is adopted, its action can be triggered first when the high voltage side conditions are met; On the contrary, when the high voltage side cannot operate, the equipment on the medium and low voltage sides can be immediately activated to achieve more precise coordination without extending the operating time on the medium and low voltage sides. In addition, there is currently a lack of information sharing among substations within the same regional power grid. Even if there is a backup power source in the system, the hot standby switch cannot be turned off if it is located in another substation. Therefore, traditional backup automatic switching systems can no longer meet the increasingly rapid development speed and high reliability requirements of distribution network systems.

3 Multi level regional standby auto-transfer system

The core concept of the multi-level regional backup automatic switching system is to comprehensively realize the exchange and sharing of backup power supply automatic switching information between substations of different voltage levels within the regional power grid, and to construct an automatic switching mechanism for regional backup power supply based on this. The system integrates multiple sets of backup automatic switching devices with different purposes and voltage levels in the region into a highly coordinated and unified whole through comprehensive logic control instructions.

To achieve the above goals, it is necessary to first configure station backup automatic switching devices in each substation. In addition to completing the basic backup power automatic switching function required for the station, the equipment also needs to achieve level 2 information exchange: first, integrate multiple sets of backup automatic switching information of different voltage levels within the equipment; The second is to reserve communication interfaces for data exchange with other substations. Subsequently, a real-time communication platform based on fiber optic communication network was established, which achieved information sharing of the entire regional distribution network through the reserved interactive interface of each substation’s backup automatic switch equipment. On this basis, a regional standby auto-transfer strategy control device will be added, responsible for formulating and implementing a coordinated strategy for standby auto-transfer devices applicable to the entire region. By using comprehensive logic control commands, priority coordination is given to backup automatic switching devices of various categories and voltage levels within the area, forming a clear and orderly operating system for each set of equipment with clear coordination relationships.

3.1 Station domain standby auto-transfer device shares station end information

The advancement of intelligent substation technology has promoted the centralized development of relay protection equipment, enabling a single device to integrate multiple protections, as shown in Figure 2. Through this mode, the backup automatic switching device can also integrate multiple sets of devices with different voltage levels distributed within the station into the same hardware platform through a digital interface, forming a “station domain backup automatic switching device”. This device collects digital analog signals, circuit breaker positions, and other switch information from the station. It uses a multi CPU board architecture for logical processing and is specifically equipped with a communication interface for transmitting regional power grid data between substations.

When building a multi-level regional backup automatic switching system, it is necessary to deeply explore the information sharing advantages of substation equipment. The core lies in achieving seamless interaction of automatic switching information of backup power sources with different voltage levels within the equipment, and using internal logic instructions to directly coordinate the priority of actions between high, medium, and low voltage sides. This method can fundamentally overcome the inherent shortcomings of traditional time-varying constant value coordination. Under this architecture, the action time limit on the medium and low voltage sides can be independently set according to one’s own needs, without the need for human extension to avoid actions on the high voltage side. When the high voltage side meets the action conditions, the internal logic will directly lock the action of the medium and low voltage sides; If a fault occurs on the high voltage side, the locking conditions on the medium voltage side and low voltage side should be immediately released to allow for quick continuation of operation. If the power loss range is limited to the medium and low voltage sides, and the high voltage side does not have the conditions for action, the medium and low voltage sides can directly act after their own delay, which not only accurately solves the problem of priority coordination, but also avoids the time delay of action on the medium and low voltage sides.

3.2 The standby automatic switching device in the station area shares regional power grid information

Based on the reserved inter station information exchange interface of the backup automatic switching device in the substation domain, a real-time communication platform is constructed to convert the real-time operating status of the substation into GOOSE signals that comply with the IEC 61850 standard, achieving information exchange between various backup automatic switching devices in the regional distribution network. By utilizing the high bandwidth and low latency advantages of 5G communication technology, an effective regional standby auto-transfer information sharing system can be established. The use of a regional backup self switching strategy control device to collect panoramic information of the regional power grid and perform regional level backup self switching functions can not only improve the efficiency of power supply recovery after regional power grid failures, but also overcome the limitations of traditional backup self switching systems that can only restore power supply to a single substation. The tripping and closing instructions issued by the regional backup automatic switching device are also sent to the connected substation equipment in the form of GOOSE signals through the information sharing platform for execution.

With the low latency characteristics of 5G and the fast transmission mechanism of GOOSE signals, millisecond level data exchange can be achieved between regional backup automatic deployment strategic control equipment and various station backup automatic deployment devices. Therefore, the operating speed of the regional backup automatic switching device can reach the same level as that of the substation equipment. When formulating priority coordination strategies, all devices can be considered at the same level and given priority action rights to the regional standby auto-transfer system. This mechanism improves the shortcomings of previous regional standby auto-transfer based on scheduling systems, which can only be used as a backup measure due to slow actions. The regional backup automatic switching system quickly restores power supply based on the panoramic information of the distribution network, and prioritizes the execution of regional level backup automatic switching operations. If its action is successful, there is no need to restart the backup automatic switching action of the corresponding individual substation. Otherwise, if the substation level equipment is activated first, the regional system may restart, resulting in frequent switching of operating modes. Based on the above design ideas, the multi-level backup self feeding system structure is constructed as shown in Figure 3.

3.3 Multi-level unified standby power automatic input system

In the dual power structure of the 220 kV distribution network, the regional power grid adopts a unified backup automatic switching system with a multi-layer architecture, and its typical configuration is shown in Figure 4. The first layer is the substation domain backup self switching layer, deployed in 220 kV substations, usually using incoming line backup self switching or 110 kV bus tie switch backup self switching methods. When the power station loses voltage or a section of the 110 kV busbar loses voltage, the first layer should be used to achieve preliminary power supply recovery of the regional power grid. The second layer belongs to the regional level backup automatic switching system, which is suitable for situations such as 110 kV line faults, 110 kV power station voltage losses, and on-site backup automatic switching devices that do not meet operating conditions. The regional backup automatic switching system will be activated to restore power supply to the lost voltage substation. The third layer is the high-voltage side of the backup automatic switching device in the station area. When the current two layers do not meet the operating conditions, they will automatically start running and be responsible for restoring power supply to the substation from the high-voltage side. The fourth floor is the low-voltage side of the backup automatic switching device in the station area. Its application scenarios include voltage loss of low-voltage busbars in series power supply substations, or if the first three layers do not meet operating conditions, this layer will be used as a backup measure specifically for restoring low-voltage side busbar power supply.

When configuring a multi-level unified backup automatic switching system in a regional power grid, the above structure clarifies the priority of actions between different backup automatic switching devices. If traditional time limited differential coordination methods are used, the accumulated delay of low priority backup automatic switching devices will result in excessively long action times, affecting the speed of power recovery. To address this issue, a multi-level unified standby auto-transfer system has introduced information sharing and comprehensive logic judgment mechanisms to overcome the above problems. Each backup automatic switching device only needs to set its own required delay, without gradually coordinating with other devices in terms of time, thereby improving overall coordination and action speed.

To clearly describe the behavior and state transition relationship of the system under different operating states, a multi-level regional backup automatic deployment system state machine model is designed as shown in Figure 5, and the state transition rule table is shown in Table 1. This state machine covers four core working states of the system and their transition logic:

• Normal: The system is in normal operation, the voltage of each bus is normal, all backup automatic switching devices have completed charging, and are in standby mode.
• Regional ATS Acting: When a 110 kV line fault or a 220 kV power substation voltage loss is detected and the station’s regional ATS Acting does not meet the operating conditions, the system enters the regional ATS Acting state, and the regional strategy control device issues the regional ATS Acting instruction.
• HV Side Station ATS Acting: When the regional backup automatic switching does not meet the operating conditions or fails, the system enters the 110 kV HV Side Station ATS Acting state.
• MV/LV Side Station ATS Acting: When only the 35 kV side bus loses voltage or the first three layers do not meet the operating conditions, the system enters the MV/LV Side Station ATS Acting state.

TABLE 1. Typical operation mode of serial power supply network.

Constraint: (1) state mutual exclusion: The system can only be in one of the four states mentioned above at the same time; (2) hierarchical progression: State transition has directionality and does not allow direct jumping from S3 to S1; (3) timeout protection: any backup self throw action state has an action time window constraint, and if the timeout is not completed, it is judged as an action failure; (4) communication degradation: When 5G communication is interrupted, the system automatically degrades to the independent operation mode of station domain backup and self switching

4 Case study analysis

4.1 5G communication technology

As a new generation wireless communication technology that is about to be popularized, 5G has the advantages of high bandwidth, high reliability, and low latency. Its application in the backup automatic switching system of regional distribution networks provides new development prospects for accurate fault location, isolation, and power supply recovery of distribution networks. This article conducted 5G communication performance testing under laboratory conditions, and the test results showed that the time synchronization accuracy between terminals was less than 300ns. The wireless timing accuracy drifted by 10 microseconds within 26 min, and the end-to-end delay of measurement information between terminals was less than 15 ms. The business bandwidth was 2.724Mbit/s. In the communication availability test, the online rate of communication terminals is 100%, and the packet loss rate of power services is 0%. According to the test results, the automatic switching of regional backup power based on 5G communication can meet the communication requirements of the backup automatic switching system, ensuring that the automatic switching of backup power uses the same sampling frequency as traditional technology, that is, 24 points per cycle, and the sampling value is transmitted to the other party/network at a frequency of 1200 Hz. The information transmission path of the 5G based backup self delivery system architecture is shown in Figure 6.

The end-to-end communication delay τ _total of 5G consists of the following components, and the calculation formula is shown in Formula 1:

τ total = τ air + τ core + τ process (1)

In the formula, τ _air represents the air interface transmission delay, which is affected by factors such as wireless channel quality and scheduling algorithms; τ _core represents the transmission delay of the core network; τ _process represents the latency of data encapsulation/de-encapsulation and encoding/decoding processing on the terminal side.

In actual 5G networks, the air interface latency τ _air exhibits random fluctuations. According to actual measurement data, the end-to-end delay of 5G can be controlled within 15 ms, and the distribution of delay jitter Δτ can be modeled as a truncated normal distribution or a mixture Gaussian distribution and the calculation formula is shown in Formula 2:

f Δ τ x = ∑ k = 1 K ω k · 1 2 π σ k exp − x − μ k 2 2 σ k 2 x ∈ − J max , J max (2)

Where: ω _*k* represents the weight of the k-th Gaussian component, μ _*k* and σ _*k* represent the mean and standard deviation of the component; J _max represents the maximum jitter boundary.

4.2 DTW algorithm

The transmission delay and delay jitter of 5G communication result in different packet transmission delays, leading to packet distortion in the time domain. The waveform of the data sampled after voltage/current normalization and the distortion waveform of the delayed data during transmission are shown in Figure 7. The blue dots indicate that the sampled data waveform is uniformly obtained in a periodic manner. The red dot indicates that the transmission delay of the first sampled data is relatively small, while the transmission delay of the second sampled data is relatively large, resulting in distortion of the sine waveform corresponding to the sampled data obtained by the protection device.

To eliminate the impact of transmission delay and jitter in 5G communication on the calculation of regional standby automatic switching standards, the DTW algorithm (Li et al., 2023; Cai et al., 2025) is introduced in the optimization processing of time series. Its core idea is to use a nonlinear dynamic time axis to regularize the signal sequence on a non-uniform scaling time axis, in order to compare two distorted time series with different rhythms and lengths, as shown in Figure 8. The DTW algorithm does not require the additional buffer depth required by traditional methods. The algorithm starts the calculation immediately after receiving enough sampling points to complete a data window. After the occurrence of a fault, data collection can be completed and DTW calculation can be triggered after a data window length, thereby achieving faster action speed. By combining time regularization and distance measurement techniques, it can effectively handle local jitter in voltage/current sequences, with tolerance for distorted sequences. Through nonlinear alignment, it accurately identifies whether waveform distortion originates from time-domain misalignment or real electrical quantity changes, eliminating the impact of transmission delay and jitter on the comparison of power consumption in regional backup automatic switching actions. The processed sampled data is shown in Figure 9.

Let the local side current sampling sequence X and the opposite side current sampling sequence Y after 5G transmission have data lengths of m and n, respectively, as shown in Formulas 3, 4:

X = x 1 , x 2 , ⋯ , x m (3)

Y = y 1 , y 2 , ⋯ , y n (4)

When applied to the regional backup automatic switching system, the voltage/current data length on both sides of the line is the same, that is, m = n. Under the ideal condition of no delay jitter, the two sequences satisfy a point-to-point correspondence relationship x _*i*→y _*i* (Wang et al., 2024). The actual transmission delay τ _*k* of 5G communication is time-varying and can be decomposed into: τ _*k* = τ ₀+Δτ _*k*, where τ ₀ is the average transmission delay and Δτ _*k* is the delay jitter component. Actual testing has shown that the end-to-end latency jitter of 5G is usually within ± 3 ms. For a system with a sampling frequency of f _s = 1200 Hz (sampling interval T _s = 0.833 ms), the Δτ _*k* can reach 3–4 sampling intervals, resulting in non-uniform stretching or compression of the receiving sequence Y relative to the original transmitting sequence on the time axis. The actual corresponding relationship becomes: x _*i*→y _*i*+δ(t)。 Among them, δ(t) is a time-varying delay offset function. If point by point comparison is still used at this time (assuming x _*i*→y _*i*), There will be significant phase deviation (Hao et al., 2022).

DTW adjusts the correspondence based on dynamic programming to find an optimal path that minimizes the distance between two sequences along that path, resulting in the strongest correlation between the two sequences (Li et al., 2023; Cai et al., 2025). The effective path formed at this point can be represented as shown in Formula 5:

W = w 1 , w 2 , ⋯ , w s , ⋯ , x m (5)

Where: w _*s* represents the s-th point of the path, with coordinates of (i _*s*, j _*s*), corresponding to the i-th point of time series X and the j-th point of time series Y.

After the DTW optimal matching, the residual time axis misalignment between the local i _*s* sampling point and the contralateral sequence j _*s* sampling point is defined as the residual time delay difference e _res(i), and the calculation formula is shown in Formula 6:

e res s = j s n · m − i s · T s s = 1 , 2 , … , m (6)

Where: (i _*s*, j _*s*) is the coordinate of the matching point corresponding to the local s-th sampling point in the optimal path W. The smaller the residual delay difference, the more thorough the compensation of DTW for delay jitter.

The formula for calculating the distance between two points is shown in Formula 7:

d w s = d x i s , y j s = x i s − y j s (7)

The calculation formula for the distance matrix D between all corresponding points in time series X and time series Y can be obtained as shown in Formula 8:

D = d x 1 , y 1 d x 1 , y 2 ⋯ d x 1 , y n d x 2 , y 1 d x 2 , y 2 ⋯ d x 2 , y n ⋮ ⋮ ⋮ d x m , y 1 d x m , y 1 ⋯ d x m , y n (8)

The DTW algorithm stipulates that the effective path must meet the following four constraints:

• The starting coordinates of the path must be (1,1), and the ending coordinates must be (m, n);
• The total number of steps K in the path must meet the following requirements:
• Continuity: When the path moves from p _*s* to p _*s*+1, it needs to satisfy the condition that when advancing along the path to D (i, j), the previous step must be one of D (i-1, j-1), D (i-1, j), D (i, j-1), as shown in Formula 7:
• Monotonicity: When the path moves from p _*s* to p _*s*+1, the phenomenon of “time reversal” is not allowed, as shown in Formula 8:

i s ≤ i s + 1 j s ≤ j s + 1 (11)

There are multiple paths W that satisfy the constraint conditions, and all paths W form the path space M. The shortest path length is the DTW distance between time series X and Y, as shown in Formula 9:

D T W X , Y = min M ∑ s = 1 K d w s (12)

At this point, DTW (X, Y) will serve as the core electrical quantity for the action criterion of the backup automatic switching device. Its physical meaning is: the residual waveform difference between two sequences after optimal delay compensation. Solving the DTW distance refers to solving an optimization problem with Formula 12 as the objective function and Formulas 9–11 as constraints.

• Normal operation/out of zone fault: The current waveforms on both sides are highly similar, with only minor time axis misalignment caused by time delay jitter. DTW effectively eliminates this misalignment through nonlinear regularization, keeping DTW at a low level (DTW (X, Y)<DTW _set) and protecting it from action.
• Fault within the area: The waveform of the current on both sides undergoes essential changes (amplitude, phase, harmonic content, etc.), and even after optimal delay compensation, the difference in waveform between the two sequences is still significant. DTW quickly rises and exceeds the action threshold DTW _set, ensuring reliable protection.

Introducing Sakoe Chiba constraints to improve the significant increase in computational complexity and time caused by long time series in the DTW algorithm (Sakoe and Chiba, 1978), limiting the path search range to the bar shaped area along the diagonal direction. The calculation formula is shown in Formula 13:

n * i s m − f ≤ j s ≤ n * i s m + f (13)

Where: f represents the Sakoe-Chiba constraint parameter.

Introduce a data standardization strategy to eliminate the problems of reduced computational accuracy and increased computation time caused by amplitude differences in the original sampling sequence, eliminate amplitude differences in data, and limit the range of sequence values. If the mean x ¯ of sequence X is and the standard deviation is σ, then the standardized new sequence X 'can be expressed as shown in Formulas 14, 15:

X ′ = x 1 ′ , x 2 ′ , ⋯ , x m ′ (14)

x m ′ = x m − x ¯ σ (15)

4.3 Dynamic model for synchronization error and fault recovery

Considering abnormal situations such as packet loss and out of order in 5G communication, synchronization errors e _syn need to introduce an error accumulation factor and the calculation formula is shown in Formula 16:

e s y n − k = α · e s y n − k − 1 + β · Δ t drift + η k (16)

Where: α is the error attenuation coefficient; β is the clock drift coefficient; Δt _drift is the time interval between adjacent synchronization moments; η _*k* is the residual error random term at the kth synchronization.

The delay jitter compensation Δτ _*k* of the DTW algorithm and the processing of system synchronization error e _syn form a complementary relationship mathematically. The collaborative mechanism between the two can be described as follows: assuming the local sampling time is t _*i* and the contralateral sampling time is t _*j*, the total time deviation between the two is shown in Formula 17:

Δ t i j = t j − t i = e syn + Δ τ k (17)

The dynamic behavior of a multi-level regional backup self delivery system can be described by a discrete state space model. Define the system state vector as shown in Formula 18:

S t = s reg , s HV , s MV / LV , s bus T (18)

The state transition equation of the system can be expressed as shown in Formula 16:

S t + Δ t = F S t , E t , C t (19)

Where: E(t) is the fault event vector; C(t) is the control instruction vector.

After a fault occurs, the action process of the backup automatic switching system can be divided into the following five timing stages: fault detection and data acquisition, 5G communication transmission, DTW calculation and criterion judgment, logical judgment and instruction issuance, and circuit breaker action. The time dynamic model for fault recovery can be expressed as:

T total = T detect + T 5 G + T DTW + T logic + T beraker (20)

The priority coordination of multi-level standby auto-transfer can be formalized as a hierarchical decision model, defining a priority function P(l), where l∈{1,2,3,4} corresponds to four levels (1 is the highest priority): P (1)>P (2)>P (3)>P (4). The action condition of the l-th level standby auto-transfer can be expressed as shown in Formula 16:

A l = C o n d l ∩ ∩ p < l S u c p ∩ ∩ p < l B l o c k p → l (21)

Where: Cond _*l* is the local action condition of the l-th layer; Suc _*p* is the success flag for the p-th layer action; Block _*p*→*l* is the locking instruction of the p-th layer to the l-th layer.

The revocation delay T _release of the blocking instruction Block _*p*→*l* is much smaller than the differential time in traditional time differential coordination, fully explaining that logical hard coordination can achieve precise priority coordination without prolonging the low priority backup self switching action time.

When 5G communication is interrupted or the quality deteriorates, the system automatically degrades to the independent operation mode of station domain backup and self switching. The downgraded action time model is shown in Formula 22:

T total _ ⁡ deg = T detect + T local + T logic + T beraker (22)

Where: T _local represents the information transmission delay based on the on-site fiber optic network, without 5G communication delay and DTW calculation delay. Although the downgrade mode sacrifices regional level coordination capabilities, it can ensure basic power supply recovery functions.

5 Engineering application of multi-level regional standby auto-transfer system based on 5G communication and DTW algorithm

Taking the distribution network in a certain area of Hubei Province, China as an example, the 110 kV #1 substation installed a backup automatic switching device to obtain GOOSE input, SV input, and conventional output signals through the original process layer network, covering two backup automatic switching functions on the 110kV and 35 kV sides. The 110 kV side adopts a three circuit incoming line extension internal bridge connection structure, while the 35 kV side is responsible for the mutual switching of two transformer switches CB103 and CB104, as well as the automatic switching of the section switch CB105, as shown in Figure 10. It is worth noting that there is an optimal synergistic relationship between the automatic activation process of the 110 kV internal bridge switch and the 35 kV section switch. The backup automatic switching devices of 35 kV #2 substation and 35 kV #3 substation can operate in coordination to achieve regional level power supply recovery function (Zang and Li, 2020; Gao et al., 2026; Jia, 2025). This pilot project relies on 5G communication technology and DTW algorithm to achieve two key functions: firstly, optimizing the timing coordination of upper and lower level actions between the backup automatic switching device of 110 kV #1 substation and the backup automatic switching devices of 35 kV #2 and # 3 substations; The second is to establish a regional backup self switching cooperation mechanism between 35 kV #2 substation and 35 kV #3 substation. To achieve optimal timing coordination between upper and lower level backup automatic switching, two trip time settings are set for the backup automatic switching devices of 35 kV #2 substation and 35 kV #3 substation, respectively. Setting 1 is specifically used to coordinate the automatic switching action of the superior backup, while setting 2 is used to coordinate the protection tripping and reclosing time of the main power supply line itself at that level.

To verify the reliability of the regional standby auto-transfer system based on 5G communication technology and DTW method, 9 typical operating modes of the series power supply network are shown in Table 2. By simulating permanent ground faults at different positions under the 9 operating modes, the operation of the regional standby auto-transfer system is verified, as well as the reliability of 5G communication used in the regional standby auto-transfer system and the correctness of the logic of the standby auto-transfer device.

TABLE 2. Typical operation mode of serial power supply network.

5.1 Simulation model

The hardware composition of the loop simulation system built on the RTDS platform is shown in Figure 11. This architecture is specifically designed for regional standby auto-transfer systems. The system builds three simulation models of the distribution network running in series, which can simulate fault situations under nine series mode operating conditions, including single-phase grounding, two-phase grounding, phase to phase short circuit, and three-phase short circuit grounding faults, and reproduce the switch action process under protection logic control. By configuring the interaction interface between the analog, input, and output signals of the RTDS system and the standby auto-transfer device, a closed-loop real-time simulation testing platform combining software and hardware of the standby auto-transfer device is constructed, which can be used for single function testing and verification of the entire transmission logic of the standby auto-transfer device integrated with 5G communication and DTW algorithm. The RTDS real-time digital simulator is equipped with RSCAD software, and the simulation step size is set to 50μs. RTDS uses a 3 PC interface board (including 24 DAC channels and ±10 V analog output) to drive a backup automatic switching device through a power amplifier, achieving hardware in the loop closed-loop testing. The 5G communication test uses Huawei MH5000 industrial 5G module, operating in the 3.5 GHz frequency band and 100 MHz bandwidth. It is connected to the RTDS simulation system through a 5G industrial intelligent gateway via a fiber optic data port, and the network type is 5G private network slicing. The measured end-to-end latency is less than 15 ms, the business bandwidth is 2.724Mbit/s, the time synchronization accuracy between terminals is less than 300ns, the packet loss rate is 0%, and the communication terminal online rate is 100%, which meets the performance requirements of regional backup automatic switching for communication channels.

The relevant parameter settings of the DTW algorithm are shown in Table 3. The Sakoe Chiba window parameter f = 8 limits the search for the optimal warping path to a diagonal band with a width of ± 8 sampling points, limiting the computational complexity to about 22% of the full matrix. At the same time, it provides a jitter compensation range of ± 6.67 ms, which is sufficient to cover the measured ± 3 ms 5G delay jitter, the probability distribution of end-to-end delay jitter in 5G is shown in Figure 12. A data window length of 30 ms and a sampling frequency of 1200 Hz generate 36 sampling points per window, ensuring timely fault detection and statistically significant waveform comparison. Independently apply the minimum maximum normalization method within each data window to compress the current/voltage sequence within the range of [0,1], eliminating the adverse effects of amplitude differences on DTW distance calculation. The single window calculation time of 3–5 ms measured on the embedded protection platform, combined with a 30 ms data window and a 40 ms circuit breaker operation time, results in a total action time of approximately 78 ms under typical conditions.

TABLE 3. Key parameter configuration of DTW algorithm.

5.2 Result analysis

In order to verify the reliability and logical integrity of the communication function of the standby auto-transfer device based on multiple communication modes, the operation of the substation and the operation time of the regional standby auto-transfer system under 9 modes were tested, and the results are shown in Table 4 and Figure 13. According to Table 4, among the 9 models, the action time of the regional standby auto-transfer model based on 5G communication and DTW method proposed in this paper is the smallest under permanent faults in the serial power supply system, with the longest action time being 187 ms, fully meeting the practical application requirements on site. Compared with 5G communication, the action time has been reduced by 33 ms, which verifies that the method proposed in this paper can effectively improve the speed and sensitivity of the standby auto-transfer device; This model effectively operates under permanent faults in serial power supply systems, verifying the feasibility of using 5G communication and DTW algorithm for standby auto-transfer systems.

TABLE 4. Test results.

To quantitatively evaluate the adaptability of the proposed method under different DG penetration rates, simulation tests were conducted in Model 2 with three penetration rate scenarios of 20%, 50%, and 80% added to the 110 kV power grid framework. The calculation results are shown in Table 5. The RTDS simulation test results are shown in Figure 14. The simulation results show that as the penetration rate of new energy increases, the action time of automatic backup also gradually increases. At 80% penetration rate, the longest action time of the proposed method in Model 2 is 207 ms, which is 22 ms more than the 185 ms without DG scenario and only 13 ms more than the 20% penetration scenario. The method used in this paper can operate correctly at different levels of penetration rate, because the DTW algorithm is based on waveform similarity rather than amplitude for judgment and can still effectively identify faults.

TABLE 5. Calculation results of different renewable energy penetration rates.

Summarize the research method of this article with the action time data of 5G based distribution network protection and self-healing systems, as well as traditional/regional backup self switching systems, both domestically and internationally, as shown in Table 6. From the table, it can be seen that the action time of 187 ms in this article is at a leading level in the specific functional domain of backup automatic deployment, significantly better than traditional on-site backup automatic deployment (in seconds) and existing regional backup automatic deployment (5–30 s). The difference between 5G differential protection (50 ms) and differential protection is due to the essential difference in functional complexity: differential protection only needs to distinguish a single line and trip, while the proposed solution needs to complete multi station information fusion, regional level strategy decision-making, and multi-level circuit breaker sequential operation.

• Communication is normal: In the event of a power loss due to a malfunction in the higher-level power supply, the standby auto-transfer device of the higher-level incoming line will first perform the recovery operation. If the operation successfully restores power supply, the standby auto-transfer in the lower-level area does not need to be started. If the superior operation fails, it will send an acceleration action command to the subordinate area’s standby auto-transfer device. After receiving this command, the standby auto-transfer device in the lower-level area will quickly perform a disconnection operation to disconnect the main power switch and then close the standby power switch.

TABLE 6. Comparative analysis of research methods.

When a fault occurs within the power supply range of this level, the standby auto-transfer of the upper-level incoming line does not operate. The standby auto-transfer device of this level will operate according to its preset tripping time setting, first trip the main power switch on this side, and then switch on the standby power supply. When manually opening the upper-level main switch to lock the upper-level incoming line standby auto-transfer, the locking signal can be synchronously transmitted to the lower level, thereby achieving linkage locking of its standby auto-transfer function.

• 2. Communication abnormality: If the superior power supply fails and loses power, the standby auto-transfer of the superior power supply incoming line will operate as usual. At the same time, the standby auto-transfer switch in the lower-level area will initiate a delay process based on its set tripping time of 1 and execute the logic of tripping the main supply switch and then closing the standby switch after the delay ends. If the upper-level incoming line standby auto-transfer is successfully restored to the main power supply during this process, the operation being carried out by the lower level will be suspended, and the entire process will automatically reset.

When the fault source is within the power supply range of this level, the higher-level standby auto-transfer also does not operate. At this time, the standby auto-transfer of the incoming line at this level will activate the tripping time setting of 1. After the corresponding delay, the sequence operation from tripping the main power switch to closing the standby power switch will be completed.

5.3 Reliability analysis and fault recovery

To ensure the correctness of the backup self switching system, this paper establishes a multi-level reliability index system using 5G redundant dual channel technology and completes quantitative calculations (Li et al., 2022; Lu C. et al., 2024). Two different operators' 5G wireless networks are used to construct a redundant dual channel experimental verification environment. Real time monitoring of link delay is carried out using 1 kHz heartbeat packets, and time series analysis technology is used to predict delay fluctuations to achieve optimal link switching. The feasibility of applying 5G redundant dual channel communication technology in distribution network protection is verified. The single channel availability rate of 5G communication is 99.99%, and after dual channel parallel redundancy, R5G = 1−(1–0.9999)² = 99.999999%, combined with R _DTW≥99.999%, R _logic≥99.99%, R _breaker≥99.9%, The calculation result of the comprehensive reliability model is R _sys = R _5G × R _DTW × R _logic × R _breaker = 99.888%, which meets the high reliability requirements of the backup automatic switching device in the power system.

Simultaneously increasing the two core indicators of action time reduction ΔT and correct action rate R _cor for statistical analysis. The calculation formula is shown in Formulas 23, 24:

Δ T = T 5 G − T DTW (23)

R cor = N cor N total × 100 % (24)

Where: N _cor represents the number of correct actions taken by the backup system; N _total represents the number of backup self throwing actions.

For 9 operating modes and 4 types of faults (single-phase grounding, two-phase grounding, phase to phase short circuit, three-phase short circuit), a total of 36 operating conditions were tested. Each operating condition was independently simulated 30 times (1,080 simulations in total), and the action time of each simulation was recorded. The DTW scheme in this article has a highly statistically significant improvement in action time compared to the pure 5G scheme. The 95% confidence interval of ΔT is 31.3–33.9 ms, which is roughly the same as the difference in action time under different methods in Table 2. Among the 1,080 simulations tested, the backup automatic deployment system correctly operated 1,078 times, with an accuracy rate of 99.81% and a 95% confidence interval of [99.07%, 99.98%].

To further verify the robustness of the method under different types of faults, the mean and standard deviation of ΔT were calculated for four types of faults, and the results are shown in Table 7. The 95% confidence intervals of ΔT for all four types of faults are completely within the positive range, and the lower limit of action time reduction is greater than 29 m, indicating that the DTW scheme can stably improve the action time of the pure 5G scheme under different fault types, proving that the effectiveness of the proposed method does not depend on specific fault types.

TABLE 7. Mean and standard deviation of ΔT under four types of faults.

Compared with a single 5G link solution, the dual channel redundant configuration increases the availability of communication links by four orders of magnitude, and the predictive active switching mechanism through time series analysis technology can complete channel switching before the delay deteriorates, rather than passively waiting for link interruption to recover, fundamentally avoiding protection misoperation or refusal caused by sudden link degradation. In terms of fault tolerance and recovery, this article establishes a three-level fault tolerance mechanism: the first level compensates for single packet loss and instantaneous jitter in real time through the DTW algorithm (recovery time<5 ms); Level 2 achieves automatic switching in case of continuous packet loss or delay exceeding the standard through dual operator redundant channels (recovery time<20 ms); When both links are interrupted, Level 3 will automatically downgrade to the independent operation mode of the station domain backup self switching (recovery time<100 ms) to ensure that the basic power supply recovery function is not affected. After communication is restored, it will synchronously switch back to the regional collaborative mode. The fault recovery strategy and related required time are shown in Table 8.

TABLE 8. Fault recovery strategy and related required time.

6 Conclusion

In order to improve the power supply reliability of regional power grids and compensate for the functional limitations of traditional standby automatic switching devices, this paper proposes a research and application of a regional standby automatic switching system based on 5G communication technology and DTW method, which can effectively shorten the action time of standby automatic switching devices on the medium and low voltage sides and quickly restore power supply to the distribution network. The following conclusions were drawn:

• Using a multi-level regional standby automatic switching system to replace traditional standby automatic switching devices, integrating standby automatic switching devices of different voltage levels within the station domain standby automatic switching device, and then using 5G communication technology and DTW methods to achieve data exchange between different substations, ensuring the accuracy of information sharing in the entire regional distribution network.
• The proposed research method was applied to 9 typical operating modes of the regional distribution network in Hubei Province. The experimental results showed that the action time of the standby automatic switching device of the proposed method was the smallest under permanent faults in the serial power supply system, with a maximum action time of 187 ms. Compared with 5G communication technology, the action time was reduced by 33 ms, which can effectively improve the speed and sensitivity of the standby automatic switching device. At the same time, all 9 typical operating models operated correctly under permanent faults in the serial power supply system, verifying the feasibility of using 5G communication and DTW algorithm for standby automatic switching systems.
• Future research will also expand the scale of cross station testing, including more voltage levels and complex fault scenarios, to explore the reliability of the proposed method in large-scale distribution networks. At the same time, we will also study redundant backup solutions and network security threats for 5G communication under extreme conditions, strengthen network security protection design, and enhance system robustness and risk resistance.

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: Fateh Krim, University Ferhat Abbas of Setif, Algeria

Reviewed by: Kenneth E. Okedu, Melbourne Institute of Technology, Australia

Zhanghao Huang, University of Macau, China