MACHINE LEARNING-BASED DIRECT CURRENT FAULT ARC DETECTION METHOD FOR PHOTOVOLTAIC SYSTEM

摘要

A machine learning-based direct current fault arc detection method for a photovoltaic system, comprising the following steps: building a random forest model, a support vector machine model and a decision tree model respectively, and performing training (S1); collecting a real-time current signal of a photovoltaic array (1) (S2); analyzing the real-time current signal to obtain a time-domain feature and a frequency-domain feature (S3); inputting the obtained time-domain feature and frequency-domain feature into the trained random forest model, support vector machine model and decision tree model respectively to obtain respective status label values (S4); setting the sum of all status label values as the total status label value, and determining whether the total status label value is greater than or equal to two (S5); when the determination result is yes, increasing the determination frequency value of a determination frequency counter by one, and further determining whether the increased determination frequency value is equal to a predetermined determination frequency value (S7); and if the increased determination frequency value is equal to the predetermined determination frequency value, then a circuit breaker operates so as to disconnect a circuit, and an alarm message is issued (S8).