The Research And Application Of Multi-sensor Information Fusion

The precision-guided weapons enable the quickly development of multi-mode guidance system. And one of the main applications of multi-mode guidance system is multi-sensor information fusion technology which integrates measurements from different sensors and processes the information through data preprocessing, data correlation, data decision-making and data fusion in order to make the right decision and enhance the precision and accuracy of the multi-mode guidance system. The theoretical study and practical application of multi-sensor information fusion in multi-mode guidance system were researched in this paper.The least-squares time alignment algorithm was introduced in data preprocessing. Three outliers elimination algorithms which were Kalman filtering algorithm, effective outliers elimination algorithm and least-squares filtering algorithm were focused in data preprocessing and simulated and compared.In data association, the nearest neighbor data association algorithm, the probabilistic data association algorithm, the joint probabilistic data association algorithm and data association algorithm based on gray correlation were discussed in the paper. What's more, data association algorithm based on entropy weight and gray correlation analysis was put forward which could adaptively give weights of the various indicators by entropy.In data decision-making, the paper discussed DS evidence theory and its improved algorithms. Then a new algorithm was summarized in the paper named time and space domain target-identification based on entropy-weight gray correlation and DS evidence theory. The algorithm first formed sensor report from the target gray correlation degree which was got by the information of the target characteristics through entropy weight and gray correlation analysis, then integrated the report in multi-sensor time and space domain under DS evidence theory, at last come the result of the decision. The results showed that the algorithm was able to eliminate redundant information, reducing uncertainty and improving the accuracy rate of target identification greatly. Because DS evidence theory could not resolve the conflict situation and non-mutex evidence, the paper introduced convex optimization algorithm which can deal with the problems above well and the results from simulation showed that the algorithm was effective.Track fusion algorithm, joint Kalman filtering algorithm and least-squares fusion algorithm were researched, simulated and compared in data fusion.In practical application, the hardware of information fusion was debugged and the interface of the host computer was wrote and debugged. So far, the hardware circuit debugged could deal with multi-sensor information by pre-processing, data correlation, data decision-making and data fusion in real-time, and was able to improve the accuracy and enhance the precision of the multi-mode guidance system and was prepared to system debugging.