| Grinding process directly affects the power consumption, steel consumption, product quality and recovery of the entire beneficiation process. But many significant parameters (such as different material content, water content, grinding balls content) is difficult to measure and even can’t be measured due to technical or economic problem, we can only use measurable parameters (such as various flow rate of ore at the outlet) to estimate these parameters.Firstly, state-space equations using the power-consumption model, ideal model, mechanism analysis and conservation of mass are built, which describe the relationship among the size distribution of product, flow rate of water, ore, ball and size distribution of feed.Secondly, analy sizing the characteristics of nonlinear with constraint in the grinding process, we found that it is difficult to gain precise state estimation value. To determine suitable state estimation method for grinding process, a typical2-order CSTR model is used in simulation study. On the one hand, the performance of four nonlinear state estimation methods (EKF, UKF, EnkF and PF) is compared in Gaussian and non-Gaussian noise. The results show that:in Gaussian, UKF and PF have better performance than EnkF; in non-Gaussian, PF has the best performance. On the other hand, an in-depth study of nonlinear state estimation methods with interval-constraint is given, nonlinear state estimation methods based on projection are discussed and the projected position is analyzed. The ICPT based on interval method is proposed based on the ICUT, which improve ICUT to non-Gaussian noise and the ICPT is compared with ICUT. Simulation result illustrated that the state estimated accuracy gained by ICPT is improved significantly.Finally, proposed ICPT is used to estimate the state variables of the grinding process and obtain accurate state estimation value for grinding production process optimization, which is significant for realizing the optimal control of production process. |
PDF Full Text Request