Research On Processing Method Of Delayed Measurements Based On Nonlinear Gaussian Filter

The on-line detection of key quality parameters in complex industrial process is very important for product quality control and optimization,which is the emphasis of process monitoring and the implementation of advanced control strategy.Based on the nonlinear state space model,it is an effective method to realize the on-line detection of key quality parameters by state estimation using nonlinear Gaussian filter.Due to the delay of sensor transmission,long-time sampling,off-line analysis and other reasons,parts of the measurement information is inevitably reach the data processing unit with delay,which cause the phenomenon of measurement with delay.Delayed measurements cannot be directly used by nonlinear filters,thus the estimation error cannot be effectively corrected only by fast measurements.The delayed measurement obtained by offline analysis is accurate,utilizing the delayed measurements can further improve the results of the on-line detection accuracy of key quality parameters in complex industrial process.Therefore,it has important theoretical meaning and application value to research processing method of delayed measurements based on nonlinear Gaussian filter.Based on the analysis of the nonlinear Gaussian filter algorithm and the processing method of the delayed measurements,the establishment of discrete-time state space model for nonlinear system with delayed measurements and the cubature Kalman filter were researched,and a nonlinear state estimation method for system with delayed measurements was proposed.Based on the covariance fusion algorithm and the distributed state fusion method,a covariance fusion cubature Kalman filter algorithm was proposed.The proposed method does not need to transform the system model,and is capable to deal with the situation that the sampling time of delayed measurements is known after arrival,which has strong engineering applicability.A continuous fermentation model and a polymerization model were used to verify the proposed methods.The experiment results show that the proposed nonlinear state estimation method can be applied to the on-line state estimation of nonlinear system with delayed measurements,has good generalization and effectiveness.The covariance fusion cubature Kalman filtering algorithm can utilize the information of delayed measurements,and the accuracy of nonlinear state estimation is significantly improved,which provides an effective approach to realize the on-line detection of the key quality parameters in the complex industrial process.