Research On Radar/infrared Distributed Fusion Filtering Algorithm

Multi-sensor information fusion Kalman filtering is the combination of information fusion and filtering theory,the main purpose is to correlation,estimation and fusion local information and data to get a more accurate fusion estimation than single source estimate.Now the commonly used method of information fusion filtering is centralized fusion Kalman filtering and distributed fusion Kalman filtering.The process of distributed fusion structure is:each sensor has its own processor,the sensor measurement data is first estimated in the local processor before entering the fusion center,and the local estimation is then sent to the fusion center,then the fusion center aligns and correlates estimation data,filtering according to local estimates of each node,finally the fusion estimation is calculated.Compared with centralized fusion of Kalman filtering,distributed fusion of Kalman filtering has less calculation and more flexible and stable structure,but the system will lose some information after local processing.The process of centralized fusion structure is:sending all measurements into the fusion center,then the fusion center aligns and correlates the measurements,filtering to get the fusion estimation.Compared with distributed fusion of Kalman filtering,centralized fusion of Kalman filtering has least information loss but has heavier Computational burden,which is not conducive to fault detection and isolation.A variable dimension synchronous filtering fusion algorithm for multi infrared sensor is proposed in this paper.In the case of variable number of infrared sensor and centralized fusion structure,the algorithm uses augmented measurements matrix to carry out synchronous fusion filtering.In this algorithm,two or more infrared sensors are used for simultaneous fusion Kalman filtering,and the filter doesn't diverge.Compared with other infrared sensor filtering algorithm,this algorithm is more simple and practical.The presented algorithm has high fault tolerance,which can also run normally if an infrared sensor failure cannot provide measurements.In the case of radar and infrared sensor filtering,this paper designs two algorithms,which are divided into two kinds:synchronous work and asynchronous operation.In the case of asynchronous operation,the sequential fusion filter is adopted in the case that the infrared sensor and radar sampling frequency are inconsistent;radar and infrared sensor asynchronous sequential fusion algorithm in distributed fusion architecture is designed in this paper.According to the idea of sequential fusion,the fusion center estimates the given state in time when the measurements arriving instead of central estimating after getting all measurements.The algorithm ensures flexibility and real-time of the filtering process,avoids operation of high-dimensional extended matrix,reduces requirements of the filter on system performance.For synchronous work,covariance intersection is adopted in fusion.In the case that the covariance of mutually independent sensors is unknown,the algorithm can obtain higher fusion filtering precision and avoid the divergence problem of fusion Kalman filtering.In this paper,different number of radar and infrared sensor synchronous fusion filtering are compared,and the simulation results show that the filtering accuracy has been greatly improved.