城市供水预测的智能模拟

摘要：为了节约能源和用水，供水公司需要根据每天的供水量来预测未来的用水量，最优地规划未来生产计划。用水量具有很大的不确定性及随时间序列的非线性。专家学者们更多关心的是用水量预测估计，方法有很多种，多重回归分析法和灰度预测法是当前最主要的研究方法，考虑到非线性和时间变化等特性，这些方法很难给出一个满意的结果。鉴于以上方法的缺陷，本文提出一个新的智能模型，该模型结合了基于选择性通用机器学习算法和支持向量机算法。通过网络设计和支持向量机学习算法的构造来模拟复杂非线性的用水量。优化的支持向量机参数由网络搜索和现有资料交叉验证最终选出。相比灰度模型，该优化模型能有效模拟非线性过程，为用水量预测提供了简单而又可靠的智能方法。

 

Intelligent modeling of urban water supply prediction

Authors: Yangu Zhang, Yanlin Zhang

Keywords: Optimized support vector machine learning algorithm; Grid search and cross validation; Water consumption; Grey model

Abstract: To reduce energy and water, water supply company need estimate future water consumption according to the record of daily water supply, and best arrange future production planning and control, water consumption is uncertainty and is strong non-linear time series, water consumption prediction estimation is more concerned by academics, it is predicted through various methods, multiple regression analysis and gray forecast are the most common method at present, these methods are difficult to give a satisfactory result according to the characteristic of nonlinear and time varying. In accordance with the disadvantage above methods, a new intelligent model is presented to predict accurately water consumption of a city based on optimal common machine learning algorithm- support vector machine in this paper. Complex and strong nonlinear water consumption was simulated by network design and conformation of support vector machine learning algorithm and the optimized support vector machine parameters were selected by the method of network searching and cross validation according to existing data. Compared the errors with output value of the optimized model and output value from grey model, support vector machine whose parameter was optimized with cross validation had excellent ability of nonlinear modeling and generalization. It provides a simple and feasible intelligent approach for water consumption prediction.