Intelligent Prediction And Optimization Control Methods For States In Lead-zinc Sintering Process

Lead and Zinc are widely used in many fields, such as military industry, electronic industry, etc. Stability of the lead-zinc sintering process (LZSP) and quality of sinter are essential to the LZSP. Since the state of sinter reflects the status of the LZSP, a stable and optimal state of sinter is of great help to increasing the quality and quantity (Q&Q) of agglomerate. Based on the features of strong nonlinearity and uncertainty, of the LZSP, this dissertation studies an intelligent integrated modeling and optimization control strategy for the LZSP, and produces achievements mainly in the following five aspects.(1) The analysis of the relationships between global production target and state parameters, and the structure of optimization and control for sintering processThe state of the LZSP reflects the status of the LZSP and infects the Q&Q of sintering agglomerate. Note that the number of state parameters is large and they have different effects on the global production target, this dissertation makes an in-depth analysis on the relationships between the operation parameters, state parameters, and global production target, and determines the target for state optimization control. Then, the structure of the state intelligent integrated optimization and control is devised. Finally, the principle of state integrated optimization and controller is presented. This method provides a new idea for optimization and control of the LZSP.(2) Prediction models of the state parametersSince permeability and burn through point (BTP) directly affect the Q&Q of sintering agglomerate, they are the most important stste parameters in the LZSP. To carry out the optimization and control of the LZSP, we require not only the current state parameters but also their future changing trends. Based on the features of time-vary and uncertainty of the permeability, we establish a radial basis function (RBF) neural network model to accurately predict the permeability. The BTP is mainly affected by a surface temperature, an experiment method which combines fixed measurement points with non-fixed measurement points is used to investigate the distribution of gas temperature in the sintering machine. Based on the analysis results, we use a back propagation (BP) neural network to establish the model of the gas temperature distribution (GTD), and further build a grey model for the BTP. To consider the influence caused by the status fluctuations, we use the support vector machine to establish a technology parameter model for the BTP. Then, we integrate these two models into an integrated state prediction model of the BTP using dynamic weights. MATLAB 7.0 is used to verify the validity of the presented optimization method. The experimental results show that the prediction precision of the integrated model is higher than that of a single prediction model.(3) Genetic-ant-algorithm-based state optimization and settingIn order to achieve the production target of high Q&Q, we need to optimize and control the permeability and BTP effectively, and stabilize the LZSP at an optimal state. Based on the analysis of machnism and control requirements, we express the relation between the state parameters and the production target as a synthetic profit function with inequality constrains. In order to solve this problem, the penalty function method is used to transform the muti-target-constrained optimization problem to an unconstrained optimization problem. Then, we use a genetic algorithm to perform coarse optimization, and an ant algorithm to carry out fine optimization. This gives us a suboptimal solution. The solution is then used as an optimal setting of the state. Simulation results show the validity of the method of the genetic-ant-algorithm-based state optimization and setting.(4) Self-adapt-immune-tabu-search-based state optimization and controlBased on the prediction, optimization and setting of the state in the LZSP, and according to the target of state optimization and control, we formulate the problem of state optimization and control of the LZSP as a nonlinear and multi-objective optimization problem. To deal with the problems of unmeasurable parameters, nonlinearity and a time delay in the LZSP, we utilize a self-adapt immune tabu search optimization algorithm to optimize the target function, and to obtain a set of optimal operation parameters. This allows us to implement the state stabilization and optimal control.(5) Application investigation of integrated optimization and controlBased on the controller of state intelligent integrated optimization, this dissertation presents a hierarchical configuration of state intelligent integrated optimization and control system. Simulation results show that the fluctuations of permeability and BTP are suppressed to a low level by the state optimization and control strategy. The system lays the foundation of implementing optimization and control of the LZSP.