采用LS-SVM的自来水厂沉淀池出水浊度建模

Forecast Model for Settling Tank Water Turbidity in Waterworks Using LS-SVM

加药混凝过程是自来水厂生产工艺的一个重要环节,如何对沉淀池出水浊度进行预测一直是个热点问题。自来水厂加药混凝过程是大滞后、非线性和时变的复杂动态系统,针对这一过程进行机理模型分析非常困难的特点,采用最小二乘支持向量机(LS-SVM)对这一过程进行建模研究,给出了沉淀池出水浊度预测模型。通过实际应用表明建立的预测模型拟合误差小、推广性能好,具有较好的预测效果,可以应用到对加药混凝过程进行优化和控制中。

Chemical dosage and flocculation are important procedures in production of waterworks ; and prediction of outlet water from precipitating tank is always the key point. It is very difficult to do the analysis on mechanism model for the dosing and flocculation process because the process is a complicated dynamic system which features large time lag, nonlinear, and time varying. By using least square support vector machine （ LS-SVM ）, the modeling research is conducted for the process ; and the predictive model of turbidity of outlet water is established. The practical application shows that the model offers small fitting error,good to be propagated ,and better predicted effects. It can be used on optimization and control of the dosing and floceulation process.