Research On Information Fusion Theory Based On Multi-Sensor For Network Environment And Their Applied Models

As a burgeoning theory and technology, the Information Fusion has provided the advanced, credible theoretics and method to solve information disposing and decision making in information times. The research on the information fusion is just unfolding in all the world at present, and it begins just now in our country. There is not any successful scheme for network environment in national bound. This subject is able to create the theory frame about the information fusion based on all-source and multi 梥ensor for network environment. The model and technique of the information fusion based on multi 梥ensor for network environment has been created, and it can determinately guide the technical innovation and application in this domain. As the problem is complex and universai, I research only on the fuzzy theory, neural network and fusing both of them, the failure diagnosis based on the knowledge, the measurement based on the the industrial field bus and connecting with Internetflntranet, the transmission on the network based on the various plate and so on. The major contents and research results are as follows: 1. Research those questions that there are a lot of incomplete or uncertain information, and that it is very difficult to state by mathematics formula. Establish the theory frame about the information fusion based on all-source and multi 梥ensor for network environment. Found the basic theory and key technology. 2. Study on the fuzzy optimization theory of the multi-target. Describe the sensors data by relative membership function of the fuzzy sets. Set up the fuzzy optimization evaluation model of power system based on the relative membership function and the concept of regulative factor. Build the evaluation model of the electric motor function, compare and analyse the factor about estimation results. 3. Study on the method fusing fuzzy theory and ANN. Every node in this new network has the certain physics means. Prove the equivalence of the fuzzy pattern recognition forecast model and the fuzzy pattern recognition ANN forecast model. Improve on the Simplex algorithm of the fuzzy pattern recognition ANN forecast model. It can settle the foundation question in the motor protection that the paper build the ascending temperature forecast model III of the motor. Research on the qualitative analysis in complex forecast system and present the concept of the qualitative factor network so as to improve the precision of it. 4. The frame and method of the failure diagnosis system on industrial devices is established based on the knowledge. The idea which the technology of devices diagnosis based on the data management now grows up the technology of devices diagnosis based on the knowledge management is presented. Based on the technology of knowledge management, it realizes that the symbolic process fuses with the numeric process, the inference course and the algorithm, the knowledge base and database on the layer of the knowledge management. The thinking of the integrated and multi-step diagnosis is presented. It provides the numeric, symbolic and linguistic fusion operators to process various data information extracted from all sensor source reliably based on the factor space theory. Establish the general formula for the fuzzy set membership function based on entropy. The fuzzy reason mechanism is changed, and the function of self-learning and self-adaptation are designed in the example. 5.