论文部分内容阅读
鉴于日用水量的时变性,本文提出一种基于变结构支持向量回归的动态预测模型.利用日用水量的历史数据训练支持向量机,得到模型结构参数历史数据序列,然后利用扩展卡尔曼滤波器对模型结构参数组进行估计,最后用模型结构参数估计量来更新模型结构并预测下一天日用水量.在实例分析中分别利用变结构支持向量回归模型和支持向量机预测模型对实际用水量性进行预测分析.结果表明,前者具有更好的动态跟踪能力和更高的预测精度,可应用于城市日用水量的预测.
In view of the time-varying of daily water consumption, a dynamic prediction model based on variable structure support vector regression is proposed in this paper.Using the historical data of daily water consumption to train SVM, the historical data sequence of model structure parameters is obtained, then the extended Kalman filter The structure parameters of the model are estimated, and finally the structure of the model is updated to update the model structure and forecast the daily water consumption in the next day.In the case analysis, variable structure support vector regression model and support vector machine prediction model are respectively used to estimate the actual water consumption The results show that the former has better dynamic tracking ability and higher prediction accuracy and can be applied to the forecast of daily water consumption in cities.