Applicative Research Of Ant Colony Neural Network For Condensing Equipment Diagnosis In Power Plant

Based condensing equipment for the main targets, using neural network technology, combined with mixed-knowledge representation and knowledge acquisition, knowledge-based expert system technology, the paper have developed and researched the fault diagnosis system for condensing equipment. This paper completed a condensing equipment failures samples collated, established a 0, 0.25, 0.5, 0.75, 1. as a sign of samples and the corresponding ant colony neural network, fault diagnosis and interpretation, data query and analysis module. This system is programmed with Visual Basic6.0 and the fault simulation tests of condensing equipment is practiced.During the course of samples training, production rules, semantic networks, frame knowledge expression etc are synthesized to express the condensing equipment fault diagnosis knowledge effectively, ensuring data integrity, reduce data redundancy.Normalized database, thus simplifying the data structure of the data and avoid conflict. The paper used a new type of optimization algorithm -- ant colony algorithm to train the neural network. It has positive feedback, distributed computation, and use of a constructive greedy heuristic. Meanwhile, the influence to the accuracy and speed of network training ,for example ,the hidden nodes, the number of ants, Information residual factor etc. are discussed. This network parameters selected to provide a reasonable basis. To highlight ant algorithm to train the neural network, Ant neural network, BP neural network, Genetic neural network were compared .Comparative results showed Ant neural network training algorithm in speed and accuracy is better than the other two algorithms. Meanwhile, fault diagnosis simulations show that the fault diagnosis system could make error reach to the aim value quickly. This proves that the ant colony neural network model is correct, and also can judge the fault rapidly and accurately.