Early Warning Algorithm for Thermal Fault Diagnosis of Electrical Equipment based on Dynamic Early Warning Threshold

Abstract

For electrical equipment in the thermal fault warning due to alarm thresholds in any case is static and lead to leakage, false alarm problem. An early warning algorithm for thermal fault diagnosis of electrical equipment based on dynamic warning thresholds is proposed. The temperature prediction model based on BP neural network is established by considering load current, ambient temperature and other influencing factors, and the normal operating temperature is accurately predicted by training and calibrating the measured data in normal operation at different ambient temperature and load current. Using the relative temperature difference method, the corresponding dynamic early warning threshold is determined by the thermal fault defect standard, and the early warning for different fault degrees is realized according to the dynamic threshold. This paper effectively solves the problem of false alarms and omissions caused by traditional static thresholds, improve the accuracy of temperature monitoring and fault warning of electrical equipment and reduce the workload of maintenance workers, Moreover, it lays the foundation for realizing the intelligence of thermal fault diagnosis and fault warning of electrical equipment, and provides a guarantee for the safety, reliability and continuous and stable operation of electrical equipment.