Research On Multi-sensor Data Fusion Algorithm In Target Tracking

Multi-sensor fusion technology has been a research hotspot in the information field in recent years.Because of its advantages such as high accuracy and low cost,it has been widely used in military and civilian fields.This article mainly studies the related technologies of multi-sensor fusion.The main research work of this paper is as follows:First,an Improved Probability Data Association algorithm is proposed to solve the problem of poor tracking performance under uncertain measurement conditions that can increase the association probability by referencing an improved weighting formula,the Improved Probability Data Association algorithm can distinguish the measured values??from the target and clutter and is combined with the Unscented Kalman Filter algorithm under the interactive multiple model to track the maneuvering targets under dense clutter.The Unscented Kalman is used for prediction,and the improved data association method is used for echo correlation to obtain a new combination information,finally,the combined information is used to complete the status update.Then,for the case where the cross-covariance between the local estimates of each node in the distributed system track fusion is unknown,the Convex Combination fusion algorithm is premised on the absolute independence of the estimation error between sensors,and the covariance Intersection fusion algorithm is premised on the complete correlation of the estimation error.Combining the advantages of the two retains its correlation and independence;at the same time,the fusion coefficient is optimized by the Fruit Fly Optimization Algorithm which further improves the complexity of fusion and is better applied to the dual radar fusion algorithm.When there are more than two radars,the fusion algorithm will be more complicated,and there will also be the deficiency that the fusion result will change with the fusion sequence.An Improved Covariance Intersection algorithm is proposed by using the inverse of the trace of the estimated variance matrix inverse at each node as the intermediate variable to calculate the fusion coefficient.Experimental results show that the improved algorithm in this paper significantly reduces the calculation time while obtaining better fusion results.Finally,the heterogeneous sensor data fusion algorithm based on improved particle filtering is proposed by combining Fruit Fly Optimization Algorithm with particle filtering.The algorithm firstly uses the Drosophila algorithm to optimize particle filtering and introduces a distance comparison formula during the resampling process.The problems of low particle accuracy and insufficient particle diversity in the particle filtering process are effectively improved.Finally,the improved filtering method is used in the weighted fusion of radar and infrared heterogeneous sensors.Experimental results show that the proposed method has better tracking effect and higher fusion accuracy than the traditional particle filtering method.