| PS matte converting is a widely used metallurgy technique to extract blister copper from matte containing iron, sulfur and other impurities. It is known as a representative batch multiphase molten bath reaction process, and the whole process has intense dynamic characteristic.Hence, it is difficult to set up the steady state working point so that the operation optimization based on static model can not be used to optimize and control the matte converting process. At the same time, the fact that the key parameters in the converting process can not be measured timely pose difficulty to apply the prediction control or other control technologies.At present, the matte converting production is manually operated based on human operators' own experience, which leads to the fluctuation of process indexes and can not make the matte converting running in optimal mode. Research on optimization control of matte converting is significant to reduce energy consumption, improve technique and economy profits of production process, increase the efficiency of the resource utilization, and realize enterprise sustainable development.On the basis of investigating the technological mechanism of the matte converting process, the nonlinearity dynamic model is developed according to metallurgy kinetic theory. In addition, the dynamic optimization control scheme based on feedback adjustment of production quality indexes is proposed. It is further used in optimization supervising decision system of PS matte converting process. Results prove the effectiveness of the proposed scheme. The major innovation research achievements include:(1) After investigating the PS matte converting process, the dynamic optimization control scheme based on feedback adjustment for production quality indexes is proposed. The optimal control laws are calculated by dynamic optimization model. In order to eliminate disturbances and other uncertainties in converting process, feedback adjustment method based on production quality indexes is utilized. The feedback information is obtained by using soft sensor model according to the materials entering and leaving PS converter. The optimal control laws are compensated and adjusted by intelligent control unit using the errors between feedback quality indexes and expected quality objects. The initial value of state variable, boundary of control variable and terminal time of dynamic optimization model are computed by parameter initialization computation models.(2)By means of research on the metallurgy reaction kinetic of PS matte converting process, nonlinear reaction kinetic models are established, which include the dynamic models of slag-making and copper-making. The models lay a foundation for optimizing and controling the matte converting process. Simulation experiments of dynamic models are carried out with production data from a copper plant, and the results of simulation experiment are compared with the observed data. Results show that two dynamic models can effectively describe the time varying transformation of matte components and matte temperature. The computing results are proved to be accurate and credible.(3) In parameter initialization computation models, the prediction method of converting end point based on reduced fuzzy least squares support vector machine (LS-SVM) is proposed to set up the terminal time of dynamic model. The anti-noise property of LS-SVM is improved by means of data fuzzy processing, kernel matrix reducing, and kernel partial least squares identifying the regression parameters of LS-SVM. The relative root mean square error of converting end point prediction is controlled below 4%, and the accuracy of results meet the requirements of production operation. In addition, to compute the initial values of state variable and the control range of cold charge amount, the optimization admeasuring calculation model of cold charge amount is established by using the linear programming method so that the cold charge amount in slag-making period can be computed according to preparing cold charge in production field. The computing results of optimization admeasuring calculation model are compared with the production data. The analysis results show that the computing results of model are accurate and effective.(4)With the rolling computation method, the soft sensor model of the ratio of silicon to iron and matte temperature are established on the basis of multiphase and multi component equilibrium calculation method. Simulation experiments show that the computing results of soft sensor model are precise. The relative root mean square error of computing results is below 1%, and it achieves the demand of optimization control system.(5)The intelligent compensating control method based on the converter running status judging model is proposed. According to the converter running information provided by soft sensor model, the converter running status judging model computes the errors between production quality indexes and expected production quality objects. If the errors are larger than the thresholds, the fuzzy rules are used to adjust the control law of flux by the intelligent control unit, and the expert regulations are used to adjust the control law of cold charge. The intelligent compensating control method improves the robustness of optimization control system of PS converting process.(6)The optimization supervising decision system is designed and developed on the basis of dynamic optimization control scheme for PS converting process. Energy saving and the stabilization of production quality of PS converting process are achieved by the optimization supervising decision system. The percentage content of silicon in slag is controlled in 21%. The average cold charging amount per furnace is increased by 7%. The average oxygen enrichment consumption per furnace is decreased by 9%.The application results in PS converting process demonstrate that dynamic optimization control scheme has good adaptability and can be popularized in the similar metallurgy process.
|
PDF Full Text Request