Modeling, Model Evaluation And Correction For Hydrometallurgy Cementation Process

Given the rapid development of metallurgical industry, it is very important to use such resources effectively and economically for sustainable development of our country. As one of the two extractive metallurgy technologies, the remarkable advantages of hydrometallurgy are high comprehensive recovery rates of valuable metals in raw materials, environmentally friendly, and the easy fulfillment of continuous and automated production processes. Therefore hydrometallurgy is more suitable for recovering low-grade metal resources. The hydrometallurgical cementation process is one of the most important processes in hydrometallurgy. The control of cementation production process still remains as an off-line analysis, experienced adjustments and manual control, which lead to low efficiency, high resourses comsumption and unstable product quality. Therefore it becomes a bottleneck for hydrometallurgy industrial development in our country.This thesis aims at the cementation rate that can’t be measured online in hydrometallurgy cementation process. Based on the deep analysis of the characteristics of cementation process, a comprehensive and systematic research was done on hybrid modeling, model evaluation and correction. The main researches are summarized as follows:1. Based on deep analysis of the cementation process of hydrometallurgy, the dynamic mechanism model of cementation process was established according to the material balance of zinc and gold cyanide ions in reaction process, kinetics principle of chemical reaction, and the characteristics of plate and frame filter press in the cementation process;2. To tackle the problems and the shortage of the mechanistic model, a parallel hybrid model was established which was based on mechanistic model and PLS compensation model. The mechanistic model provides a prior knowledge for the data model, and data model makes up mechanistic model’s unmodeled characteristics. Which cause the hybrid model to have a better prediction performance.3. This paper uses conformal prediction algorithm to evaluate the hybrid model. To tackle the problem of hybrid model that can’t update online, a model correction strategy was proposed which combines short-term correction and long-term correction. The PLS model is corrected by the method of discount recursive PLS, and the mechanistic model parameters is corrected by the propagation of particle swarm optimization method. The former belongs to short-term correction and the latter belongs to long-term correction.4. The effectiveness of the proposed method is verified by simulation analysis.