论文部分内容阅读
为了提高城市供水调度的品质和效率,需要高精度的日需水量预报信息作为参考。分析影响城市需水量变化的主要因素,以近期需水量、降水及气温实测值为输入,辅以星期、节假日信息校正,采用RBF神经网络与支持向量机相结合的数据驱动建模技术,进行超前一天需水量预报研究。为了提高黑箱模型的训练效果,对数据进行一系列预处理,包括分离出历史需水量中的变化量;提取降水量的连续等级信息;非线性处理温度对需水量的影响。通过模型验证,结果表明预报误差在1%以内的占总预报天数的62.0%。
In order to improve the quality and efficiency of urban water supply dispatch, high-precision daily water demand forecast information is needed as a reference. The main factors influencing the change of urban water demand are analyzed. Inputs of recent water demand, precipitation and air temperature are used as input, supplemented by week and holiday information correction. Data-driven modeling combining RBF neural network and support vector machine One day water demand forecast research. In order to improve the training effect of the black box model, a series of preprocessing of the data was carried out, including the separation of historical changes in water demand, the extraction of continuous grade information of precipitation, and the effect of nonlinear treatment temperature on water demand. The results of model validation show that the forecast error is less than 1%, accounting for 62.0% of the total forecast days.