Method Of Operation Status Evaluation For Process Of Hydrometallurgy Of Gold Based On Improved Random Forest

The shortage of mineral resources and the decrease of high-grade ore resources are seriously restricting the development of China’s national economy.Therefore,how to use the low-grade ore economically and efficiently,to ensure the maximum comprehensive economic benefits under the premise of normal production,is the focus of the current metallurgical industry development.The evaluation method of the operation status of the complex industrial process provides a certain guarantee for improving the operation efficiency of the industrial production process and improving the production efficiency.Therefore,the research on the operation status evaluation method for various characteristics of complex industrial processes has become a hot research topic.In the production process of hydrometallurgy,there are not only a lot of on-line quantitative measurement information,but also a lot of qualitative information with high value.The coexistence of quantitative information and qualitative information is an obvious feature of hydrometallurgy production process,which brings some difficulties to the implementation of common operation status evaluation methods.In this paper,an improved random forest algorithm is proposed for the complex industrial production process with the coexistence of quantitative information and qualitative information.Based on this method,an operation status evaluation model based on the mixture of quantitative information and qualitative information is established to realize the online operation status evaluation of the process.The improved random forest algorithm is based on mutual information.Aiming at the problem of information redundancy between decision trees in traditional random forest,the improved random forest model is simplified by retaining the decision tree with high evaluation accuracy.And in the voting process,the evaluation precision is used as the voting weight of each decision tree.It solves the problem that traditional random forest ignores the performance differences between different decision trees.In this paper,cyanidation leaching and zinc powder replacement,the two most important processes of hydrometallurgy,are taken as the main research objects of the whole process operation state evaluation.We evaluated the operation status of the whole process of hydrometallurgy.In the offline modeling stage,according to the characteristics of long process and many processes of hydrometallurgy,the whole process is divided into sub block layer and whole process layer to model its own model respectively.In the online evaluation stage,input the online data into the corresponding models respectively for evaluation the operation state.When the result of operation state evaluation is not optimal,this paper uses a non-optimal cause identification method which is based on relative deterioration degree,which can quickly locate the nonoptimal causes and provide suggestions for the operation of the production process.With the simulation and analysis of the hydrometallurgical process operation status evaluation,the effectiveness and the advancement of the method proposed in this paper are proved.