以上海市青草沙原水智能调度管理系统为背景,采用基于改进粒子群的最小二乘支持向量机为原水需水量预测的方法，得到了较为准确的预测效果.通过对需水量数据进行特征分析，发现在节假日需水量预测与实际供水量有较大误差.建立基于时差系数的小时级与天级原水需水量预测模型，用以改善和优化原天级预测模型.最后，结合水厂的实际运行情况，将优化改善后的预测模型应用于水厂，为其提供更为精确的需水量预测并取得较好结果.
The support of water supply system has been a concerned focus of urban construction. The accurate prediction of short term water quantity is important for the whole water system operation and maintenance. In this paper, the intelligent scheduling management system for raw water based on least square support vector machine with improved particle swarm optimization is proposed by means of the project Shanghai Qingcaosha Intelligent Raw Water Dispatch and Management System. After analyzing the characteristics of water quantity data, the results of water quantity prediction have a big deviation from the actual water supply during the holidays. So forecasting model of daily and hourly water demand is built based on time difference coefficient to optimize the original prediction model. Combined with the actual operation and process conditions of water plant, this optimized model is applied to water plant to provide more accurate water supply scheduling suggestion.
［1］上海市水务局, 上海市统计局. 上海市第一次水利普查暨第二次水资源普查公报［EB/OL］, (201388) ［2014112］.http:∥shzw.eastday.com/shzw/G/20130808/u1ai111884.html.
［2］HERRERA M, TORGO L, IZQUIERDO J, et al. Predictive models for forecasting hourly urban water demand［J］. Journal of Hydrology, 2010, 387(1): 141150.
［3］TIWARI M K, ADAMOWSKI J. Urban water demand forecasting and uncertainty assessment usingensemble waveletbootstrapneural network models［J］. Water Resources Research, 2013, 49(10): 64866507.
［4］王春超, 王丽萍, 曹云慧, 等. 改进多变量灰色模型在城市用水量预测中的应用［J］. 水电能源科学, 2013, 31(2): 2729.
WANG Chunchao, WANG Liping, CAO Yunhui, et al. Application of improved multivariable grey model in urban water consumption prediction［J］. Water Resources and Power, 2013, 31(2): 2729.
［5］JAIN A, VARSHNEY A K, JOSHI U C. Shortterm water demand forecast modelling at IIT Kanpur using artificial neural networks［J］. Water Resources Management, 2001, 15(5): 299321.
［6］岑健, 危阜胜, 张多宏, 等. 最小二乘支持向量机用于水量预测［J］. 计算机仿真, 2009, 26(7): 212215.
CEN Jian, WEI Fusheng, ZHANG Duohong, et al. Least squares support vector machines for water quantity prediction［J］. Computer Simulation, 2009, 26(7): 212215.
［7］岳琳, 张宏伟, 王亮. 粒子群优化算法在城市需水量预测中的应用［J］.天津大学学报, 2007, 40(6): 742746.
YUE Lin, ZHANG Hongwei, WANG Liang. Application of particle swarm optimization in prediction of urban water demand［J］. Journal of Tianjin University, 2007, 40(6): 742746.
［8］戢钢, 王景成, 葛阳, 等. 城市小时级需水量的改进型引力搜索算法最小二乘支持向量机模型预测［J］. 控制理论与应用, 2014, 31(10): 13771382.
JI Gang, WANG Jingcheng, GE Yang, et al. Gravitaional search algorithmleast squares support vector machine model forecasting on hourly urban water demand［J］. Control Theory & Applications, 2014, 31(10): 13771382.
［9］PALMER R N, HOLMES K J. Operational guidance during droughts: Expert system approach［J］. Journal of Water Resources Planning and Management, 1988, 114(6): 647666.
［10］ZHOU S L, MCMABON T A, WALTON A, et al. Forecasting daily urban water demand: A case study of Melbourne［J］. Journal of Hydrology, 2000, 236(3): 153164.
［11］DING B, HUANG B. Constrained robust model predictive control for timedelay systems with polytopic description［J］. International Journal of Control, 2007, 80(4): 509522.
［12］任崇光, 高俊芳, 张黛华. 时间序列分析在水量预测中的应用［J］. 预测, 1985(S1): 7784.
REN Chongguang, GAO Junfang, ZHANG Daihua. Application of time series analysis in water quantity forecast［J］. Forecasting, 1985(S1): 7784.
［13］郑毅. 时间序列数据分类、检索方法及应用研究［D］. 合肥：中国科学技术大学研究生院, 2015.
［14］吕谋, 赵洪宾. 城市日用水量预测的实用动态模型 ［J］. 哈尔滨建筑大学学报, 1998, 31(3): 3944.
L Mou, ZHAO Hongbin. A practical dynamic model for forecasting urban daily water consumption ［J］. Journal of Harbin University of Civil Engineering and Architecture, 1998, 31(3): 3944.
［15］SHVARTSER L, SHAMIR U, FELDMAN M. Forecasting hourly water demands by pattern recognition approach［J］. Journal of Water Resources Planning and Management, 1993, 119(6): 611627.
