Charging pile fault diagnosis method, device, equipment and medium

• Main Authors: ZHANG YUN, USHIGAKI QUI, KANG QI, YUAN JIAQI, GUO TAILONG, YANG JINGJING, ZHANG LEI, WANG YAQUN, DING DANLEI, ZHAO YONGLIANG, WANG CHAO, QI TIAN
• Format: Patent
• Language: Chinese; English
• Published: 11.08.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention belongs to the technical field of fault diagnosis, and particularly discloses a charging pile fault diagnosis method and device, equipment and a medium. Comprising the following steps: acquiring historical operation data of a charging pile, and dividing the operation data of the charging pile into a test set and a training set; inputting the data in the training set into a CNN-LSTM-Attention model for model training, and obtaining the weight of each piece of data; substituting the weight to obtain a to-be-tested model, testing the to-be-tested model through the test set, and outputting a prediction model through the test; acquiring real-time operation data of the charging pile, and inputting the real-time operation data of the charging pile into the prediction model to obtain a prediction result; and judging whether the charging pile fails or not according to the prediction result. According to the method, the CNN-LSTM model is adopted to organize the charging pile time sequence data, and the at