组合优化的城市供水量预测模型

城市供水系统是一个复杂的大系统,供水量受多种因素的共同影响.为了节约水资源,供水公司需要根据日供水量记录估计未来一段时间的用水量,以便安排未来的生产调度计划.以过去七年的历史数据为基础.建立了供水量的最优组合预测模型.首先采用霍尔特指数平滑法预测模型进行预测,然后采用BP神经网络预测模型进行了预测,最后采用最优组合预测模型对前两种模型的预测结果进行了优化,模型具有较高的预测精度,组合预测的预测效果优于任意一种单一预测的预测效果,供水量预测结果对未来供水的短期或长期规划能起到重要的宏观指导作用.组合优化的预测结果更为稳健、精度更高,在其它的能源消费预测中具有一定的推广应用价值.

Compound Optimized Model of Forecast of Energy Consumption

This paper introduces a model of optimal compound forecasting of water supply discharge from history data of the past seven years.Firstly,a prediction using Holt Exponential Smoothing is discussed.The second prediction is using Genetic-BP Neural Network.Finally,this paper optimizes the prediction results from these two models with optimal compound forecast model,which produces better accuracy than any one of the single models.Such forecast provides important general guidance to short-term and long-term plan...