Research on the Prediction Model of Urban Power Grid Short-Term Load Based on BP Neural Network Algorithm

Abstract

Objective: The paper aims to establish the prediction model of urban power grid short-term load based on BP neural network algorithm. Method: Five factors influencing the urban power grid short-term load are used to establish the neural network model: date type, weather, daily maximum temperature, daily minimum temperature and daily average temperature. Matlab toolbox is used to develop the testing platform through VC++ programming. Result: The variable learning rates are 0.35 and 0.64. With 23410 iterations, the model is converged, and the global error is 0.00032. Conclusion: Through the data comparison and analysis, the relative error is within 5%, thus indicating the model is accurate and effective, and it can be used to predict the change of urban power grid short-term load.