Research on Vibration-AE Fusion Detection Technology for Metal Particle Defect in GIS under Electromechanical Coupling Condition

Gas-Insulated Switchgear (GIS) is a critical transformation and switching equipment in power systems, and its operating condition directly affects grid safety and stability. Studies have shown that under high-voltage and high-current conditions, GIS is simultaneously subjected to electric field and mechanical stress, and there is a correlated influence between its insulation and vibration conditions, making insulation defect and mechanical defect the two main causes of GIS failures. For example, metal particle contamination is a typical insulation defect related to mechanical vibration, while AE and vibration signal measurements are effective for detecting insulation and mechanical conditions, respectively. However, traditional single-signal detection fail to capture the correlation and complementarity between Partial Discharge (PD) and mechanical vibration. The lack of data fusion analysis constrains diagnosis efficiency, and the evolution process of defect cannot be researched. Therefore, the optimal design and development of a vibration-AE fusion sensor, the exploration of intelligent sensing technology for GIS condition detection, the investigation of metal particles motion and PD characteristics under combined power-frequency voltage and mechanical vibration, and the proposal of a vibration-AE fusion detection method, together provide both theoretical and technical support for on-site GIS defect detection, offering significant academic and engineering value. To overcome the limitations of current sensing technologies for detecting mechanical and insulation defects in power equipment, specifically the separation measurement and resulting restricted diagnosis, a vibration-AE fusion sensor is developed which enables synchronous and co-located measurement of mechanical vibration and PDAE signals. An electromechanical-acoustic equivalent model for piezoelectric vibration-AE measurement is established, and a fusion sensing structure is designed. Simulation analysis clarifies the relationships between structural parameters and measurement performance for both vibration and AE signals. Furthermore, an intelligent structural optimization method based on data augmentation-algorithmic optimization is proposed. A multifunctional embedded circuit is designed, and a prototype fusion sensing device is developed and calibrated. The developed fusion sensor exhibits a vibration sensitivity of 950 mV/g and an average AE sensitivity of 101 dB within the target frequency band, with a resonant frequency of 54 kHz, enabling synchronous detection of vibration and AE signals. To address the current issues of low intelligence and poor efficiency in power equipment sensing, incorporating the features of low-power wireless communication, edge computing, and self-calibration, an intelligent architecture for fusion sensors is proposed, which enables autonomous enhancement of sensing, storage, computing, and transmission. An integrated intelligent wireless module is designed, developed, and assembled within the fusion sensor. Test results show that in wireless low-power mode, the sensor achieves a data rate above 1 Mbps within 8 meters in substation environment, with a theoretical battery life of up to four years. An edge computing strategy tailored for vibration-AE fusion signal processing is proposed, along with an embedded algorithm that enables signal feature extraction at the sensor level. A novel calibration method based on pseudo-random M-sequence excitation is firstly proposed for efficient impedance and sensitivity calibration of piezoelectric sensors. Simulation and experimental results verify that the method offers high accuracy, low time and power consumption, and stable, reliable operation. This work improves fusion sensing efficiency and lays the foundation for distributed on-site measurement applications. Given that current studies on metal particle defect detection do not adequately account for the combined effects of mechanical vibration and operating voltage in GIS, this work experimentally investigates the motion and PD characteristics of free metal particles in GIS under simultaneous power-frequency voltage and mechanical vibration. A GIS experimental platform is constructed to apply synchronized vibration-voltage excitation. The effects of electric field strength and mechanical vibration on typical metal particle are analyzed, revealing the motion and PD characteristics. For spherical and linear particles, both the jumping initiation (Ei) and cessation (Ec) field strength increase with particle mass, while mechanical vibration reduces the thresholds. For flaky particles, the jumping behavior is influenced by both mass and geometry, small, elongated flakes jump more easily, whereas large, square-shaped ones are more stable. Once airborne, flakes and lines exhibit active motion, with heavier particles generating stronger discharges. Under high field strength, firefly phenomenon may occur. Multiple particles in motion may cause chain breakdown, leading to short-circuit fault. Under the same field strength, lighter particles move more actively with lower discharge, while heavier ones move less but generate stronger PD. These findings provide experimental reference for metal particle detection in practical GIS. Based on the research background and engineering need, the fusion detection technology is proposed for GIS and on-site tests are conducted in substation. Methods are proposed to evaluate the shape, mass, and quantity of metal particles. Shape is infered by flight patterns and discharge rates; mass is estimated from the flight slope; and quantity is assessed from the discharge-mass-quantity correlation. VMD is employed to denoise and separate fusion signal, and signal feature extraction methods are summarized. A vibration-AE fusion measurement and communication framework is proposed, along with a cloud-edge collaborative computing model, offering an innovative strategy for intelligent distributed sensing, processing, and diagnosis in substation. The developed vibration-AE fusion detection system demonstrates efficient signal sensing, processing, storage, and transmission capabilities, and is successfully applied to the condition monitoring of GIS in substation.