Design of Mine Ventilator Fault Diagnosis System

The common faults of mine ventilator are researched in this paper, and rotor misalignment, unbalance, oil whirl, surge and other faults and fault characterization of the generation mechanism are analyzed. The faults diagnosis system is designed based on rough neural network. First, the characteristics of the type of fault for fan failure data collection, including vibration and temperature signals. Then, the pretreated sample data using rough set attribute reduction method to delete redundant attributes. Finally, the sample data is divided into training and testing samples, were used to train and test the neural network classifier. Experiments show that the system is reliable, diagnostic yield, improved ventilator system security, expanding the scope of application of rough sets.