Research On Cubature Kalman Filter With Incomplete Measurements And Its Application In Fermentation Process

It will result in the phenomena of incomplete measurement when different sampling frequencies of the measurements, heavy computational load and certain key measurement delayed in the state estimation of nonlinear systems. How to utilize the incomplete measurements effectively is an important way to solve nonlinear state estimation with incomplete measurements. As so far, cubature Kalman filter(CKF) that has a good filter performance is a widely used nonlinear filter algorithm, especially for high-dimensional nonlinear system state estimation. There is a key method for the fermentation process, which is a typical complex high-dimension nonlinear system, to improve the accuracy of state estimation that uses incomplete informations such as delay measurements in fermentation process effectively in nonlinear filtering algorithm. Therefore, it has important theoretical meaning and application value to research on cubature Kalman filter with incomplete measurements and its application in fermentation process.An adaptive robust square-root CKF(ARSCKF) algorithm is proposed based on the research about adaptive robust algorithms in nonlinear filter. It effectively overcomes the impact of the model uncertainty and the incorrect statistical properties of noise on filter performance; A cubature Kalman filter with incomplete measurements combined with sample-state augmentation method by research on filter theories and methods with incomplete measurements is proposed. It can effectively use delayed measurements whose sample time and delay time are uncertain to improve the accuracy of state estimation; The adaptive robust SCKF filter algorithms with delayed measurements is applied to the state estimation of fermentation process according to research on the filter methods about using on-line and off-line measurements in fermentation process.Simulation results show that the adaptive robust SCKF algorithm can effectively overcome model uncertainty, get highly state estimation accuracy and stability, and this filter is more suitable for high-dimensional nonlinear systems; CKF algorithm under incomplete measurements that is proposed can effectively use variable sample time and delay time lag measured data to improve the accuracy of estimation of key variables, and it also has a smaller calculations; An adaptive robust SCKF algorithm with measurement delay applied to biomass fermentation process parameters to obtain a good result, and it also provides a new approach that solves biological fermentation process parameters online estimation.