Research On Data Fusion Technology In Heterogeneous Multi-sensor

Multi-sensor data fusion is a comprehensive discipline of processing multi-source information, is the product of the rapid development of high-tech science and technology, widely used in many fields of military and civilian. Heterogeneous sensor data fusion is a key and difficult topic in the study of multi-sensor data fusion, experts and scholars have been attention to the issue research, how to use different types of sensors measure the complementary and redundancy of multi-source data, obtain more accurate target state and location information than a single sensor. This paper studied data association 、 data fusion and state filtering problem in the heterogeneous multi-sensor fusion, established a heterogeneous multi-sensor data fusion tracking system, and then detected the target.First of all, this paper studied the characteristics and performance of a variety of different type of sensors, the structure model of multi-sensor data fusion, and analyzed the function and their advantages and disadvantages of the three level position data fusion model, and introduced the key technology and method of data fusion, advanced a kind of heterogeneous structure model of multi-sensor data fusion.Secondly, studied the data correlation、data fusion and filtering method in the heterogeneous multi-sensor data fusion. Analyzed amendment act、recent clinical domain method and sequential method and other related algorithms based on statistical theory, and verified correct association rate and error association rate of the different algorithms through the simulation. Several data fusion algorithms were introduced, and on the basis of the weighted fusion algorithm, proposed the weighted fusion based on the local state estimation(LSE) algorithm, and verified its effective performance through the simulation. The filtering algorithm of state estimation was studied, on the basis of kalman filter, studied extended kalman filter 、insensitive kalman filter and particle filter of the three nonlinear filtering algorithm, and analyzed their tracking performance through the simulation.Finally, after researched the data association method、data fusion method and filtering method, combined the characteristics of different types of sensors and structural model of fusion, and then established a heterogeneous multi-sensor data fusion tracking system from radar, infrared and laser consisting, detecting and tracking the target of air, and then verified the reliability and better tracking performance through the simulation.