Fault Diagnosis Method for Control Cabinet Based on Quantum-Behaved Particle Swarm Optimization and Kernel Extreme Learning Machine

As the core equipment integrating primary electrical devices with secondary intelligent control units, the control cabinet plays a vital role in smart substations, and its operating reliability is crucial to the security and stability of the entire power grid. In order to solve the problems of low accuracy and insufficient generalization ability of traditional control cabinet fault diagnosis schemes, this paper proposes a fault diagnosis method based on QPSO-KELM. Firstly, the structure of the control cabinet and the current characteristics of the switching coil are introduced. Then, by combining the global optimization capability of the QPSO with the nonlinear feature extraction advantages of the KELM, a fault diagnosis method for the control cabinet is proposed. Finally, the coil current signals collected by the operating mechanism of the integrated primary and secondary switches in the control cabinet are extracted for experimental analysis to validate the proposed model. Compared with the existing techniques, the proposed approach not only raises the diagnostic accuracy but also shortens the convergence time considerably in the fault-diagnosis task of the primary–secondary integrated switch housed in the control cabinet. It achieves a diagnostic accuracy of 97.5% and saves 13.55 ms in a single training time compared with PSO-KELM, providing a new approach for the intelligent operation and upkeep of substation assets.