Short-Term Bus Load Forecasting Method Based on CNN-GRU Neural Network

摘要

In recent years, with the rapid development of the power market and smart grid, higher and higher requirements are put forward for load forecasting technology, and deep learning is also widely used in the filed. Aiming at the small load base of bus load, strong time series and large influence by relevant factors, this paper proposes a short-term bus load forecasting method based on CNN-GRU neural network. The method processes the input historical load, date type, weather data, renewable energy generation, time-of-use electricity price and other related factors through the CNN network, intelligently extracts the dominant factors and compresses the generated timing feature vectors, and then performs bus load forecasting through the multi-layer GRU network. Taking several bus load data from a certain city in the east from 2012 to 2018 as samples/tests, CNN-GRU and traditional BPNN forecasting methods were used for prediction. The experimental results show that CNN-GRU deep neural network has higher precision and better prediction effect when dealing with bus load forecasting.