Soft Measurement Modeling of Turbidity in Flocculation Process of Drinking Water Treatment Using Gaussian Process Regression

The unpredictable changes of raw water quality and large lag characteristic of flocculation process bring great difficulties to the flocculation process. The soft measurement for the turbidity of sedimentation tank outlet can provide in-time predictive value for the feedback control of flocculant dosage and is therefore essential for the flocculation process of drinking water treatment. Gaussian process regression (GPR), as an efficient nonlinear modeling method, can effectively interpret the complicated features of industrial data by using covariance functions derived from base kernels. In this study, a hybrid model control scheme based on GPR consisting of a long-term part and a short-term part is proposed to predict the turbidity of sedimentation tank outlet. The proposed real-time control method can cope with seasonal and uncertain changes of the raw water quality. Experimental studies have been carried out and implemented for the alum dosing process control system, and the results demonstrate the effectiveness and practicality of this real-time control method.