State fusion of fuzzy automata with application on target tracking

For better target tracking, this paper presents a new viewpoint that is the states fusion of various fuzzy automata, and mainly discusses the algorithm of the fusion on the states of the fuzzy automata based on Bayesian theory and fuzzy knowledge in detail, respectively. The recurrent neural network architecture corresponding to fuzzy automata is provided for the states fusion of fuzzy automata and the simulation results are given. The simulation results show target tracking based on the states fusion of the fuzzy automata is better than that of single state information of the fuzzy automata relatively. Moreover, the application of fusion of fuzzy automata to track recognition is given. Thus, it will be a theoretic base for the application of any automata. Finally, some problems and development trends on fuzzy automata and neural networks are presented for future research.