基于IEA的需水预测投影寻踪模型研究

需水预测是一个由城市人口、工业水平、社会经济水平共同作用的多因素、多层次的复杂非线性系统,其结果将直接影响受区域水资源承载力约束的产业结构、布局形态等决策。PP方法通过投影将高维数据转换为低维数据,克服了维数祸根问题,能在一定程度上解决分类、函数逼近和时间序列预测等问题。将PP方法引入需水预测领域,建立了需水预测PP模型。同时,将免疫进化算法和PP耦合起来,简化了参数优化运算。在珠海市的应用实例表明:PP模型较好地解决了高维非线性和非正态问题,在需水预测中有较强适用性。

PP Model of Water Demand Prediction Based on IEA

Water demand prediction is a complicated multifactor,multi-level non-linear system influenced by the urban population,industrial and economic level.The predicted water demand is the important reference in decision of regional industrial structure and layout.Projection Pursuit(PP) method can avoid dimension disaster by transforming high dimension data to low dimension data.In this way,PP method can resolve such problems as classification,approximation,time series forecasting,and so on.Thus,PP method is applied to building a PP water demand prediction model in this paper,where PP is coupled with IEA to optimize parameters.The application of the model in Zhuhai shows that PP model has good applicability to water demand prediction by solving high dimension,non-linear,non-normal problems,etc.