Sensors´ decision fusion algorithm based on the learning strategy

In the target detection of radar and sonar systems,it´s difficult to give prior probability of the target´s appearance and the cost of system´s wrong decision. In some practical applications,the probability of target´s appearance will continually change. It is difficult for the existing distributed system´s decision fusion algorithm to solve the decision fusion problem of unknown and variable targets.In this paper learning strategies is used to estimate target probability in real-time and to achieve adaptive decision fusion.Analysis shows that,in the detection of unknown and variable targets, this algorithm can adaptively modify related parameters according to the detected objects.The detection performance has good convergence with the increase of study time and the algorithm performance is better than NP and Bayes algorithm.