Research On Nonlinear State Estimation With Delayed Measurements

The state estimation quality of key variables affects the advanced control of chemical process directly, and nonlinear filtering algorithm is an important realization approach for nonlinear state estimation. The actual chemical process is a complex nonlinear system with high dimensions and strong nonlinearity. The filtering of the existing nonlinear filtering algorithms may appear loss of accuracy, large calculation or poor stability. Thus a better performance nonlinear filtering algorithm becomes an urgent need. The measurement for key variables often need off-line test of the laboratory test, and the data will arrive after an inevitable measurement delay. With efficient using of nonlinear filtering algorithm to fuses the delayed data, the estimation errors caused by calculation error or sensor error will be corrected, and the state estimation precision will be improved. Thus, the research on nonlinear state estimation with delayed measurements has theoretic importance and application value.A high-degree cubature Kalman filter based on the diagonalization of matrix(DMHCKF) under the unified framework of Gaussian filter is proposed based on the research about existing nonlinear filtering algorithms. The algorithm uses the diagonalization of matrix to take place the covariance matrix factorization of HCKF. It does not need the positive definiteness of the covariance matrix, and the eigenspace remains unchanged after the decomposition, and that will benefit to enhancing the accuracy and stability of filtering with HCKF algorithm. The methods of nonlinear state estimation with delayed measurements for chemical process is studied, a DMHCKF with delayed measurements based on sample-state augmentation are proposed, and it has been applied to the nonlinear state estimation for chemical process.The simulations show that DMHCK has higher precision and stability of filter. The DMHCKF with delayed measurements can use the delayed data in a rational and efficient manner, and the precision of nonlinear state estimation is improved. The DMHCKF with delayed measurements is applied to the state estimation of chemical process to acquire better estimates, and it provides a new way to resolve chemical process parameters on-line estimation.