The Fault Diagnosis Approaches Of Mine Hoist Braking System And Reducer Based On Information Fusion

With the development of sensor technology and computer technology, Simple monitoring of the mine hoist has been developed to comprehensive, real-time and multi-sensor on-line monitoring. The monitoring system saved a lot of information on the state of the hoist, using information fusion techniques to extract useful and identified information, to reduce the uncertainty of the information, and these identified information is applied in state assessment and fault diagnosis of the hoist.In order to improve the accuracy and reliability of the monitoring and diagnosis system for hoist and make full use of historical data, three methods for hoist braking system are proposed. The first method is principal component analysis, making full use of the data. This method can only qualitatively detect and locate faults, but can not distinguish the type of fault. The second method is a two-layer structure of information fusion fault diagnosis, including feature layer and decision-making layer. In feature layer RBF(Radial Basis Function) neural network is used for spatial integration and in decision-making layer DS evidence theory is used for time integration. This approach greatly improves the accuracy of diagnostic systems, but has the inadequacies of not making full use of the data. The third method is a three-layer structure of information fusion fault diagnosis, including data layer, feature layer and decision-making layer. This method is proposed for the inadequacies of the first two methods. In data layer PCA (Principal Components Analysis) is used for data integration, using Elman neural network for feature integration in feature layer and using DS evidence theory for decision integration. Finally, utilizing the method Q statistic contribution plot in order to locate the fault location. The approach greatly improves the accuracy and reliability of the diagnostic system, and also compensate for the inadequacies of the first two methods.A decision level fusion structure model proposed for hoist reducer is applied to gear fault diagnosis. First of all, neural network is used for property judgment of the information of each sensor. Then, fuse the attribute judgments of each neural network to get a decision fusion. This approach has strong flexibility. When the number of sensors increases, it only need to increase the number of neural network.Through Experimental study, proves that the application of the above information fusion fault diagnosis method in the diagnostic system of hoist reduce the miscarriage of justice and the missing and improve the reliability of the system, ensuring the safe operation of the hoist and improving the economic efficiency of coal mine.