基于混合算法软测量的污水处理厂污泥回流系统设计

Prediction of Wastewater Sludge Recycle Volume Based On PCA-GA-SVR

针对污水生化处理中的污泥回流量模型具有高度非线性的问题,提出了基于PCA-GA-SVR(Principal Component Analysis-Genetic Algorithm-Support Vector Regression)的污泥回流量软测量模型。该方法由两部分组成:主成分分析PCA和GA-SVR支持向量回归的混合算法,先使用PCA进行降维,然后再以遗传算法对SVR的影响预测精度的几个有关参数进行优化,建立污泥回流量模型。实验表明提出的预测模型是有效的,具有较快的收敛速度,相比SVR预测模型更高的预测精度及更快的运行速度。

PCA-GA-SVR Principle Component Analysis Cenetic Algorithm-Support Vector Regression model is constructed for that sludge recycle flow rate model of biological wastewater treatment,it is highly nonlinear.This method consists of two parts：the principal component analysis PCA and hybrid algorithm of GA-SVR supporting vector regression.First uses PCA to reduce the dimensionality.Then optimizes the SVR parameters which affecting prediction accuracy by genetic algorithm,and establishes sludge recycle processes model.Experiments show that the prediction model is effective with higher convergence speed.The prediction model has higher prediction accuracy and faster speed compared to SVR prediction model.