Research On PLS Modeling And Correcting Of Sludge Bed Height Of Thickener

Thickener is the essential process equipment of mineral processing and water treatment. Underflow density is the most important indicator of the production. However, real-time detection of sludge bed height of thickener which is control basis of the underflow density is difficult to achieve. It reduces the efficiency of thickener greatly. Therefore, researching on modeling and correcting of sludge bed height of thickener has important theoretical and practical significance.This paper first introduces thickener production process and technology. And then the mechanism model of sludge bed height of thickener is analyzed and modeled which is based on operation mechanism of thickener. Thus, we have a deep understanding of Settling and thickening mechanism and slow time-varying characteristics which provides mechanism theoretical basis and simulation data support for follow-up research. Based on the mechanism model, partial least squares (PLS) method is introduced into the slow time-varying system modeling which is based on data-driven. And then the prediction model of sludge bed height of thickener is modeled by PLS, recursive PLS and discounted-measurement block recursive PLS algorithm. The result of simulation shows that the prediction model which is modeled by discounted-measurement block recursive PLS can solve the problem of the prediction of sludge bed height of thickener effectively. Although the problem of the prediction of sludge bed height of thickener with its slow time-varying characteristics can be solved, there exist some problems in the general continuous online modeling strategy of recursive PLS, such as wasting resource and erroneous learning. Therefore, the model is monitored by PCA monitoring method and then corrected by online PLS which is based on variable contributions in this paper. And then a new online modeling method:PCA-PLS algorithm is conducted which models of slowly time-varying systems. Simulations on prediction of the sludge bed height of thickener show the improved PCA-PLS algorithm is effective.