Fault Prediction Method For Leaching Process Based On Dynamic Causality Diagram

With the development of economy,the demand for rare metals such as gold is increasing year by year.As the main metallurgy technology,hydrometallurgy is widely used because of its unique advantages in dealing with low grade complex ores.But along with the scale of production process of hydrometallurgy continues to expand,the production process complexity and the risk are increasing.Once a failure occurs,it may cause great losses.Therefore,it is of great significance to predict faults in hydrometallurgy process.Data-driven fault prediction methods or model based methods are often unsatisfactory due to the coexistence of qualitative information and quantitative information in the process of hydrometallurgy.The fault prediction technology based on Dynamic Causality Diagram(DCD),to a certain extent,can make up for the shortcomings of fault diagnosis methods that rely entirely on process data or system model,and better avoid the loss of information,thus attracted wide attention of experts and scholars.In addition,in the large scale production process,the causality between many variables is time-dependent,that is,there is a delay in the occurrence of the result.But the traditional Dynamic Causality Diagram does not consider this time information.This paper proposes the concept of Temporal Dynamic Causality Diagram(TDCD),which adds time information to causality.The temporal dynamic causality diagram model not only contains the causality between events,but also contains the delay time information between events.Based on the hydrometallurgical leaching process as the research background,this paper predicts two kinds of fault.In this paper,the fault prediction model is established and applied.Firstly,analyze the process mechanism and common faults of hydrometallurgical leaching process.The structure was established by mechanism knowledge and improved structure learning algorithm.The model parameters were obtained according to the parameter learning algorithm.Among them,the learning algorithm of time parameter is proposed.The delay time between related events is obtained by the mobile search maximum MIC algorithm.In the process of reasoning,this paper research a time interval reasoning method.The advancement and effectiveness of the fault prediction method are verified by simulation.Finally,the method is applied to the hydrometallurgical leaching process.The simulation results of the conventional method and the improved method are compared,and the effectiveness and advancement of the improved method are proved.In this paper,a fault prediction method is studied for small faults,and the fuzzy Dynamic Causality Diagram method is used to predict the fault.The characteristics of the small faults are not obvious in the early stage,but the variables under their influence will have slight abnormalities.In this paper,the small abnormalities detected are fitted to obtain the predicted values of the monitored variables.By enriching the evidence of the abnormal membership degree of the predicted value,it is used for causal graph reasoning to obtain the predicted result.Finally,the method was applied to the hydrometallurgical leaching process,which proved the effectiveness of the method.