Intelligent Signal Fusion and Fault Diagnosis of Complex Systems

Nonlinear and non-stationary signals are common across major engineering domains such as intelligent manufacturing, aerospace, energy and power, and high-end equipment. These signals act as a primary source for describing operational state evolution and fault characteristics in complex systems. Examples include vibration signals from power transmission systems, acoustic signals from fluid machinery, and electrical physical monitoring signals from electromechanical equipment.In practice, these signals often show multi-source coupling, time-varying nonlinear behavior, and hidden fault features masked by strong noise. These traits create persistent challenges for (i) adaptive signal decoupling, (ii) efficient deep feature mining, and (iii) robust operation under extreme working conditions ultimately limiting the performance of modern signal processing and intelligent diagnosis methods.Efficient signal analysis is a key enabler for overcoming bottlenecks across the “signal perception → feature extraction → state diagnosis” pipeline. It directly impacts the accuracy, real-time capability, and reliability of health management for complex engineering systems. Advancing this area has clear strategic value for improving the safe service of critical equipment, increasing industrial productivity, and reducing full life-cycle operation and maintenance costs.Although nonlinear signal processing and intelligent diagnosis have progressed rapidly in recent years, major scientific and technical gaps remain. These include multi-source heterogeneous signal fusion, suppression of feature drift under dynamic operating conditions, few-shot/zero-shot fault recognition, and lightweight deployment for engineering applications. This Research Topic welcomes interdisciplinary advances that deliver both theoretical innovation and practical, validated breakthroughs.Scope of Contributions (including, but not limited to)1. Theoretical innovations in nonlinear/non-stationary signal decomposition and denoisingExamples: adaptive modal decomposition optimization, deep learning-enhanced time frequency analysis, blind decoupling for multi-source heterogeneous signals.2. High-dimensional feature extraction and optimization for complex system state characterizationExamples: entropy-based feature fusion, cross-domain feature transfer, adaptive feature selection, redundancy suppression.3. Accurate state estimation and prognosis methods for complex engineering systems based on signal analysis.4. Signal recognition, localization, and traceability for early-stage faults and composite faults.5. Innovative applications of vibration/acoustic/electrical–physical signal processing for fault diagnosis of critical equipment.6. Integrated diagnosis frameworks combining signal analysis and data-driven modelsExamples: lightweight CNNs, attention-based architectures, graph neural networks.7. Robust signal analysis under extreme operating conditionsExamples: high/low temperature, strong electromagnetic interference, complex loads, long-term service degradation.8. Cross-domain method migration and adaptation across different engineering systemsExamples: sparse representation, dictionary learning, modal component reconstruction.9. Fusion of signal analysis with digital twins and edge computingExamples: virtual signal simulation, real-time deployment of diagnosis algorithms at the edge.10. Engineering implementation and validationExamples: hardware verification, monitoring system development, on-machine testing, large-scale industrial case studies. Topic editors: Gang Yu, Nan Wang Source: https://www.frontiersin.org/research-topics/77052/intelligent-signal-fusion-and-fault-diagnosis-of-complex-systems