The Research On The In-supremum Of Noise Limitation For Linear System State Estimation

The state estimation of the dynamic system is widely existed in the field of automatic control, fault diagnosis, target locating and tracking and economic prediction. It is concerned by related experts both at home and abroad, and people have made great progress on its estimation in recent years. Effective state estimation depends on the design of the filter, while the operation of which is closely related to the use of sensor. A sensor’s quality can directly influence the accuracy of filtering, and the improvement of it mainly involves two aspects. One is about how to optimize the filter’s structure and its working process by combining with the characteristics of the linear and nonlinear system. Besides, using the high-precision sensor to reduce the measurement error is a good way to improve the accuracy. However, it is indispensible to its cost. It will take a higher price to get a better measurement system. But in real applications, design filter will usually consider it in the upper and lower bounds of the satisfies the given index constraints can have major principles on the parameters of the filter This paper, based on the linear system, researches how to set limitations of the systematic noise and metrical noise in the precondition of loosing the sensor’s accuracy. Mainly include the following contents:1. Briefly introduces the modeling of the linear discrete system, linear continuous system and the multi model systems by studying related knowledge about linear system and concluding the literature review, and focuses on the introductions of the four typical optimized estimation rules and the realization of the kalman filter.2. This paper, based on the information filtering, then under known conditions sensor accuracy by solving LMI principle, achieve system noise and measurement noise is indeed bound, and analyzed as a function of the estimation error covariance matrix of the community and the system noise and measurement noise, in order to guide actual projects selected sensor. By setting the simulation environment and simulation results show the feasibility and effectiveness of the algorithm.3. Due to the interactive multiple model algorithm(IMM) achieved a soft handover between different models, to avoid a single model of the way the hard decision algorithm in motion and the motion model matching process, the time delay problem. In this paper, an interactive multi-model as a framework to build a multi-model system for linear measurement noise is indeed bound algorithm, and further study as a function of the estimation error covariance matrix and measurement noise, theoretical guidance and validation algorithm simulation analysis feasibility and effectiveness.