Optimization Of Operating Parameters Based On Data Driven Alumina Roasting Process

In the Bayer process for the production of alumina,the roasting process is the last critical process and has a very important impact on the quality and yield of the alumina product.Due to the long process flow,the detection has hysteresis,the variables have strong coupling and nonlinearity,and the process mechanism is complex,resulting in high energy consumption in the production process.Guangxi has a very rich bauxite resource,the reserves of the deposit are large,and the ratio of aluminum to silicon in the ore is high.According to the characteristics of strong hysteresis and non-linearity between the operating parameters and the roasting state parameters,it is difficult to model and optimize by the traditional mechanism method.This paper takes an aluminum plant in Guangxi as the research background,based on the alumina roasting production process and production rules.Combined with artificial intelligence technology,data driving and data mining methods,a global optimization method for roasting process oriented to production targets is proposed.The main contents in this paper are summarized as follows(1)Firstly,according to the process mechanism of the roasting process,the relationship between the operating parameters and the state parameters of the roasting process and the production index are analyzed in depth,the optimization control target of the roasting process is determined,and an optimization strategy of the alumina roasting process is proposed from the perspective of global optimization(2)Aiming at the difficulty of roasting process modeling,the state parameters affecting the burning reduction of alumina chemical index and the operating parameters affecting the roasting temperature and oxygen content in flue gas were selected as input variables of soft sensing model by mechanism analysis,and the prediction models of roasting temperature oxygen content in flue gas and burning reduction were established.In order to improve the prediction accuracy of the model,the chaotic Grey Wolf algorithm is used to optimize the penalty factor and kernel parameters of LSSVM.The simulation expeiments on production data show that the prediction model can accurately predict roasting temperature,oxygen content in flue gas and burning reduction,and meet the requirements of industrial soft sensing and optimization.(3)On the basis of the prediction model,in order to make the alumina roasting process run in an optimal state,the optimization of the whole roasting process is divided into two parts:the reduction optimization and the operation parameter optimization of the roasting process;firstly,the adaptive differential evolution algorithm is used to obtain Optimized set point for calcination temperature and flue gas oxygen content.Then,the optimized value of the operating parameters is obtained by the adaptive differential evolution algorithm,so that the gas suspension roaster operates within the range of the calcination temperature and the optimum set value of the flue gas oxygen content,thereby improving the stability of the alumina production process and the quality of the product.(4)The operation parameter optimization control system of alumina roasting process was designed.The actual production operation data verification of an aluminum plant roasting workshop in Guangxi showed that the optimization method of roasting process oriented to the production target proposed in this paper has certain feasibility and obtained operating parameters(aluminum hydroxide).The optimal combination of cutting amount,ID fan power and Roots blower speed ensures stable calcination production.