Prediction Research of Water Demand Probability Interval in Liuzhou City Based on BP Neural Network

Abstract

Water demand prediction is of great importance for regional water resources planning.In this paper,BP neural network is primarily applied to predict the water demand of Liuzhou in 2020 and 2025,and then 100 groups of forecast results are selected for statistically frequency distribution analysis in the case of the relative error of prediction results being less than 2%.The results show that the prediction results of Liuzhou obey normal distribution in 2020 and mildly positively skewed distribution in 2025.Moreover,combining the prediction results of Liuzhou in 2025 which are converted to normal distribution by SPSS,and the interval estimation of normal distribution,the prediction interval of Liuzhou water demand,i.e.2.147 8~2.162 2 billion m3 in 2020 and 2.181 3~2.202 3 billion m3 in 2025 is obtained under the condition of 95% confidence level.The research has certain theoretical and instructional significance for Liuzhou water resources planning in the future.

Keywords

• BP neural network
• normal distribution
• interval estimation