论文部分内容阅读
影响城市日用水量的因素众多,供水部门由于缺乏有效的日用水量预测模型,造成了调度过程中严重的水电等资源浪费现象。针对日用水量变化的趋势性和周期性特点,提出了基于GA-BP神经网络与LSSVM支持向量机的组合预测模型,即选择不同影响因素分别输入到两个子模型,可达到最优效果。在对两个子模型的训练过程中,同时获得预测结果的置信概率,利用置信概率结合两子模型的预测结果,建立组合预测模型,并与传统组合模型进行了对比分析。在上海市某区域自来水公司的应用表明,与单项预测模型、传统线性和非线性组合模型相比,该组合模型具有更高的精度和泛化能力。
There are many factors affecting daily water consumption in urban areas. Due to the lack of an effective forecast model for daily water consumption, the water supply department has caused serious waste of resources such as hydropower during the dispatching process. In view of the trend and periodicity of daily water consumption, a combined forecasting model based on GA-BP neural network and LSSVM is proposed. That is, different influencing factors are input to the two sub-models respectively to achieve the optimal effect. In the process of training the two sub-models, the confidence probability of the prediction result is obtained at the same time. Combined with the confidence probability and the prediction result of the two sub-models, the combination prediction model is established and compared with the traditional combination model. The application of a water company in a certain area in Shanghai shows that the combination model has higher accuracy and generalization ability than the single prediction model and the traditional linear and nonlinear combination models.