Prediction of Daily Water Consumption of Residential Communities Based on Correlation Analysis and BP Neural Network

Abstract

In recent years, with the rapid development of cities, water resources are becoming increasingly scarce. Reasonable prediction of community water use is an important basis for regional water resources allocation, effective management of water resources and water conservation. The prediction model of BP neural network based on correlation analysis is established. Firstly, the influence factors of water consumption are analyzed by the correlation analysis theory, the influence size is sorted, and then the partial autocorrelation theory is used to analyze the daily water flow sequence, and the optimal delay time is determined, and then the input variables are determined. Then the BP neural network model is used to predict the water consumption of the community. Finally, the BP neural network model based on the correlation analysis is compared with the other models. The results show that the average error of forecasting model for residential community water consumption is 2.21% and the maximum relative error is 6.87%. Compared with the BP neural network model without correlation analysis and the least squares support vector machine model, the error is the smallest.