Research On Fault Diagnosis Model Of Genetic Neural Network Of Local Ventilation System In Coal Mine

The fault of local ventilation system is one of the main factors to induce gas explosion accident at heading face in coal mine. The components of local ventilation system are comparatively complex which involve numerous fault types and reasons. For lacking of reliable gist for fault diagnosis of local ventilation system, the technologist can not judge the fault reasons quickly and accurately, which prolongs the ventilation break time without plan and that it brings a hidden danger of gas explosion. Therefore, the means of combining the neural network with the genetic algorithm is brought forward to establish fault diagnosis model of local ventilation in coal mine to improve efficiency and accuracy of fault diagnosis model in this dissertation. The main performed researches are as follows:(1) The abundant faults information of the mine local ventilation system was collected and the fault symptoms and reasons of local ventilation system were determined. The fault diagnosis tree model and the fault diagnosis samples are collected and established.(2) The technique of the artificial neural network is used to establish fault diagnosis model of local ventilation in coal mine. The problem of the number of hidden layer and the selecting of neurons nodes of the fault diagnosis model base on BP artificial neural network are analyzed, and the function of the complex network structure is used to optimize network structure of the fault diagnosis model. The fault diagnosis model base on BP artificial neural network of the local ventilation system in coal mine is formed.(3)As the disadvantage of easily getting into local minimum of the BP neural network, according to the global search property of genetic algorithm, the means of combining the neural network with the genetic algorithm is used to establish fault diagnosis model of local ventilation system in coal mine. The model inherits the advantages of traditional genetic algorithms and has powerful function of the approximation function in neural network, and it gets rid of the BP neural network defects.(4) On the base of above studied, Two kinds of neural network model procedure are programmed by applying Matlab7.0 software, and two kinds of neural network model is optimized through the training fault samples of local ventilation in coal mine. The result of experiment show that the training results of genetic neural network are better than the training results of BP neural network, and has a good diagnostic effect.The technology of combining the neural network technology with the genetic algorithm is applied to the fault diagnosis mode establishment of local ventilation system in coal mine and the feasibility and diagnostic capacity of the model is verified in this paper, which provides a new way for fault diagnosis of the mine local ventilation system .