Measurement Accuracy Evaluation of Power Quality Online Monitoring Device Based on LSTM

At present, most power quality online monitoring devices do not carry out accuracy test, and the measurement data may not be accurate due to the degraded performance of software and hardware and environmental impact after a certain number of years of operation. In this paper, a measurement accuracy evaluation method based on LSTM in proposed. Referring to the regulation that the precision inspection cycle of power secondary equipment is more than one year and the power quality online monitoring device must pass the accuracy test before putting into operation, it is considered that the measurement data of the device in the first year is accurate and reliable. Firstly, box plot is used to clean the data. Aiming at the coexistence section of reliable data and abnormal data interval in box plot, the slope law of abnormal data is used to complete the data cleaning. Then, the standard voltage and current value is given by the voltage and current prediction model based on LSTM. Finally, DTW cumulative distance is used to evaluate the accuracy, so as to avoid the problem that the timing of standard value and measured value does not correspond. By using the PQTS automatic test system, the feasibility of this method is verified.