Research on Electricity Price Anomaly Verification Technology Based on Big Data Analysis

This research is devoted to developing a technology of electricity price anomaly verification based on big data analysis, aiming to improve the verification efficiency and accuracy. We first built a comprehensive database containing historical electricity price, electricity consumption, weather conditions and other multi-source data, and carried out strict data cleaning and pre-processing to ensure the quality and consistency of the data. In terms of algorithm selection, we use long short-term memory network (LSTM) to capture the longterm dependence in the electricity price time series, and convolutional neural network (CNN) to extract features from the graphical time series and identify potential price patterns. In order to improve the accuracy of verification, we combine LSTM with CNN to form a hybrid model. The hybrid model can deal with both temporal and spatial characteristics and understand the changing law of electricity price more comprehensively. In addition, we also introduce isolated forests as an anomaly detection algorithm to identify electricity price behavior that deviates from the normal pattern. In the process of model training, batch gradient descent method was used to optimize the parameters. The experimental results show that the proposed hybrid model and anomaly detection algorithm perform well in electricity price anomaly verification.