Hadoop-based parallel BP neural network energy consumption prediction method

Abstract

The invention discloses an energy consumption prediction method of a parallel BP neural network based on Hadoop, and the method combines a current mainstream ensemble learning related algorithm with a Hadoop distributed platform, improves the model building speed through a parallel calculation mode, and improves the generalization capability of the algorithm through a method of building a plurality of models or even combining a plurality of different models. According to the method, accurate prediction of future energy consumption of the metro can be realized, the processing capability, the calculation speed and the generalization capability of a prediction algorithm are improved, and reliable data support is provided for energy efficiency management and energy distribution of the metro by predicting the energy consumption of a future period and optimizing a control strategy and daily operation of the metro.

Description

technical field [0001] The invention belongs to the technical field of energy consumption forecasting of urban rail transit, in particular, an energy consumption forecasting method based on a Hadoop-based parallelized BP neural network. Background technique [0002] In the process of urban rail transit operation management, accurately predicting the energy consumption of subway trains in the future cycle is conducive to optimizing the control strategy and daily operation of the subway, providing reliable data support for the energy efficiency management and energy distribution of the subway, and can continuously and effectively assist the operation of the industry and service. The traction energy consumption model is a time series model. With the advancement of machine learning and deep learning, the prediction methods for rail transit traction energy consumption are also expanding. [0003] In the research on traction energy consumption prediction, at the level of statisti...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an energy consumption prediction method that can support parallelized BP neural network calculations. The parallel establishment of an energy consumption prediction model can improve the generalization ability and prediction accuracy of the model, and solve the problem of high time complexity for model establishment.

Images