Research On Modeling And Adaptive Real-time Optimization Methods For Gold Cyanidation Leaching Process

As one of the two extractive metallurgy technologies,hydrometallurgy is a process to use some solvent to separate the metal of raw mineral materials by chemical reactions,which has the advantages of high metal recovery,good flexibility,simple equipment,good environmental conditions,less investment,quick effect and good recovery of associated components.Moreover,hydrometallurgy processes are easier to fulfill continuous and automated production and meet the requirements of sustainable development of mineral resources.As the most important procedure of hydrometallurgy process,the quality of leaching solution affects the final economic benefits strongly.Although leaching technology in hydrometallurgy has reached the world advanced level,the level of automation in the real leaching process is still not high and still remains in the way of off-line test and analysis in laboratory,manual adjustment by experience and manual control,which leads to low production efficiency,high consumption of materials,high production cost and low profit.Therefore,under the condition that the quality of the leaching products are ensured,reasonable distribution of raw materials and then reduction of the total consumption,production cost,and improvement of economic efficiency have become serious problems that need to be solved urgently in hydrometallurgy industrial processes,which are solved based on the online prediction of key variables and process optimization.According to the difficulties of on-line measuring the leaching rate in gold cyanidation leaching process,in this dissertation the theory and application problems about the mechanistic modeling and hybrid modeling are investigated systematically after analyzing the reaction mechanism of leaching process deeply.The serial hybrid model is established by combining the mechanistic model with the data-driven models.Whereafter,according to the problem of randomness of manual operation in real leaching process,the optimization model of gold cyanidation leaching process is established based on the serial hybrid model.And then the solving method based on GA(Genetic Algorithm)and SQP(Sequential Quadratic Programming)and three adaptive real-time optimization strategies to solve plant-model mismatch are also proposed.The main researches are summarized as follows:(1)Based on the detail analysis of the reaction mechanism of leaching process,the single-level and multi-level dynamic mechanistic models of gold cyanidation leaching process are established,which consists of the mass conservation equations of gold in the ore,gold in the liquid and cyanide ion in the liquid,as well as the corresponding kinetic reaction rate models of gold and cyanide ion.Moreover,the simulation platform to simulate the real gold cyanidation leaching process is built and the effect of key variables on the leaching rate is analyzed by the simulation platform,which has laid an important foundation for the hybrid model.(2)The serial hybrid model is proposed to predict the leaching rate of gold cyanidation leaching process,which consists of the mechanistic model and data-driven models.The known part of leaching process is modeled by the mechanistic model(gold in the ore and cyanide ion in the liquid),however,the unknown kinetic reaction rates are estimated by the two KPLS models.Furthermore,due to the fact that the outputs of the KPLS models are unmeasurable,an effective estimation strategy based on the Tikhonov regularization method is proposed,which can reduce the effect of measurement noise on the estimates as much as possible.Finally,the efficiency of the serial hybrid model is verified by the industrial application,which has provided an important guiding significance for the operation optimization of leaching process.(3)After deeply analyzing the technology requirements of leaching process,an optimization model matched with the real plant is established and then the solving method based on GA and SQP is proposed.Considering the characteristic of the real leaching process,that is,due to the limitation of production technology,production cost and measurement noise,it is impossible to obtain an accurate process model that matches with the real plant absolutely and there is some uncertainty in model(parameters,structures and process disturbance).Therefore,the plant hierarchy structure used to solve plant-model mismatch is investigated deeply and three adaptive real-time optimization strategies are proposed to compensate for the effect of plant-model mismatch on the original model-based optimization problem:classical two-step approach,the direct input adaptation method based on the logarithmic-linear barrier-penalty function and the output feedback and the modifier adaptation approach.The efficiency of the proposed optimization methods are verified by simulation experiments,which has laid an important foundation for successfully implementing the plant-wide optimization and control for hydrometallurgy process.(4)On the basis of the above theory studies,the software of the prediction and optimization operating system for gold cyanidation leaching process has been designed and developed.And then the software is applied to the gold cyanidation leaching process in a hydrometallurgy plant and good application results are gained,which has laid an important foundation for successfully implementing the integrated intellectual optimization control technology for the plant-wide process of hydrometallurgy.