A Fault Detection Method for Multi-Sensor Data of Spring Circuit Breakers Based on the RF-Adaboost Algorithm

In the context of increasing the complexity and intelligence of modern power systems, traditional maintenance approaches for circuit breakers have shown limitations in meeting both reliability and economic requirements. This paper proposes a multi-sensor data fusion fault detection method based on the RF-Adaboost algorithm for spring-operated circuit breakers. By integrating pressure, speed, coil current, and energy storage motor sensors into the mechanism, multi-source operational data are acquired and processed via denoising and feature extraction techniques. A fault detection model is then constructed using the RF-Adaboost classifier. The experimental results demonstrate that the proposed method achieves over 96% accuracy in identifying typical fault states such as coil voltage deviation, reset spring fatigue, and closing spring degradation, outperforming conventional approaches. These results validate the model’s effectiveness and robustness in diagnosing complex mechanical failures in circuit breakers.