Information Fusion Technique Based On Neural Network And Fuzzy Inference

As information fusion technique was presented, more and more attention has been paying to this technique by scientists and governments in different countries. They spent a lot of manpower and capital in developing and researching this new technique. The applied area of this technique is also getting wider and wider. At the same time, academic papers, academic conferences and scientific research items on information fusion are getting more and more. The number of information fusion methods is particularly increasing fast. In all of the information fusion methods, neural network and fuzzy inference, which have some intelligent properties, especially attract the scientists and governments in different countries. When applied to information fusion, neural networks and fuzzy inference have their own advantages and disadvantages. Comparatively the advantages of fuzzy inference lie in its capabilities of knowledge representation, and that the underlying structure of the system and the reasoning results are clearly understandable. The disadvantages of fuzzy inference system lie in the fairly limited learning abilities and the difficulties in the knowledge acquisition process. On the contrary, the advantages of neural networks lie in its superb learning abilities. The disadvantages of neural networks lie in that it is very difficult to come up with a reasonable interpretation of the overall structure of the network because the knowledge acquired during training is stored in an implicit manner. Hence in order to benefit from the strengths of each technique it is necessary to employ them both in combination rather than exclusively. After deeply researching and studying both native and foreign references on information fusion, the author mainly did the work that followed: (1)Outline the current research situation on information fusion technique, then explain the necessity of employing them both in combination after stating the advantages and disadvantages of neural network and fuzzy inference in practical application; (2)Present the basic theory of information fusion, including the fundamental principles, advantages, levels, basic framework and primary methods of information fusion; (3)Explore how to use neural network to realize information fusion. Present the basic thoughts and main steps to utilize neural network for information fusion, then testify the validation of the method through using MATALB language emulator to solve a practical problem, namely the fault diagnosis decision of electronic circuits; (4)Study the primary thoughts and basic steps on how to use fuzzy inference for information fusion, then show that the feasibility of this algorithm through utilizing computer emulator to solve a practical problem, namely the control decision of the valve in a certain system; (5)On the basis of researching the combination mode between neural networks and fuzzy inference, present the algorithm using adaptive neural network-fuzzy inference system for information fusion. Then testify the validation and advantage of the algorithm before-mentioned through applying the computer emulator to solve a practical problem, namely the recognition of the fighter plane.