Prediction Of Key Variables Of Thickening Process Based On Hybrid Model

Resource problems have already become the basic issue of the sustainable development strategy in China. Due to the continuous consumption of mineral resources, high grade ores of great economic value are rapidly decreasing, and the metal resource exhaustion situation is becoming severe; on the other hand, our country contains a large quantity of low grade non-ferrous metal resources. Therefore, how to utilize the low grade non-ferrous metal resources is very important of the sustainable development of economy and society. Since hydrometallurgy can deal with miscellaneous and low grade mineral, and benefit for environmental protection, it has been widely used.Thickening process is an important unit operation in hydrometallurgy. The underflow concentration of thickener is the key quality index. Currently the control of thickening processes is still in the state of manual operation, lead to the enormous fluctuate in concentration and flow, and has great effects on the downstream processing. Due to the harsh working conditions, many influence factors, nonlinear, large time-delay and slow time-varying of thickening process; the automatic control has been a difficult problem. To achieve the optimization control of thickening process, the prediction model of key variables that related to optimization control is of great importance.Aiming at the above mention problems and based on deeply analyzing the characteristics of thickening process, this thesis carries out a comprehensive and systematical research on soft sensor methods and applications for thickening process in hydrometallurgy using hybrid modeling method which combines the first principle and data-driven modeling methods. The main researches are summarized as follows:(1)Through analyzing the basic principle of sludge sedimentation and based on solid flux theory and mass conservation theory, the dynamic sludge concentration distribution model is established. Through the simulation of the proposed model, the characteristics of the thickening process are revealed to find the main factors of the thickening process, to determine the secondary variables of the soft sensor model, and to establish the foundation of soft sensor model.(2) Aiming at the difficulties of applying the dynamic mechanism model to the industrial field directly, soft sensor model of thickening process is established adopting parallel hybrid model. The model is composed of the simplified mechanism model and data compensated model. The simplified mechanism model is used to describe the overall trend of thickening process and improve the computational efficiency of the model while the data-driven model is used to compensate for the prediction error of the dynamic mechanism model. Considering the nonlinear characteristics of thickening process, least squares support vector machine (LS-SVM) approach is used to realize the nonlinear fitting of prediction error.In order to make the model better practical application, dynamic prediction model is established. Simulation results verify that hybrid method is more effective in comparison with the dynamic mechanism model.(3) Aiming at the difficulty of slowly time-varying characteristics, a model correction strategy is proposed based on model evaluation. Gaussian Mixture Model (GMM) is established to describe model prediction error distribution characteristics. Then the statistic is constructed from the GMM to evaluate the hybrid model performance. Based on the assessment results, the hybrid model is updated using model output offset updating or adaptive LS-SVM strategy. Through combining model output offset compensation and online model parameter updating, the model prediction accuracy is further enhanced. The effectiveness of the correction strategy is verified through the simulation results.