Error analysis of watt-hour meter based on improved long-term and short-term memory hybrid neural network

In order to analyze the running error of watt-hour meter, an improved hybrid neural network based on long-term and short-term memory is designed. This paper analyzes the theoretical error of watt-hour meter from three aspects, that is, the error caused by watt-hour meter operation, the error caused by internal components and the error caused by the line. In this paper, K-means clustering is used to select the index which has a great influence on the operation error of watt-hour meter, and the operation model of watt-hour meter is established. In this paper, the model is applied to the improved LSTM neural network, and the dync sliding window and attention mechanism are used to realize the daily running error analysis and error prediction. The experimental results show that the proposed scheme can effectively identify the error source of watt-hour meter and predict the error according to the operation conditions.