Research And Development Of Mine Ventilator Fault Detection System

Mine fan is a key equipment for providing fresh air underground coal mine.Once a failure occurs,the lighter will affect the use and cause economic losses.In severe cases,it will cause damage to the machine and even casualties.The main failures of mine ventilators are imbalance,shaft misalignment,looseness,rubbing and bearing failures,among which bearing failures include inner ring,outer ring,rolling element,cage and other failure types.In the fault diagnosis of mine ventilator,the vibrationbased fault diagnosis method is the one with the easiest signal acquisition and the best application effect.In order to make accurate judgments on the failure of mine ventilator and give accurate maintenance suggestions,this paper proposes a new diagnostic process based on the vibration signal analysis method.It first determines whether a failure occurs through time-domain characteristic parameters,and then The fault type is determined by the convolutional neural network,and finally the degree of fan damage is determined by the PeakVue method.In order to verify the feasibility of the program,the experiment table was used to simulate the type of failures that may occur during the operation of the fan.The normal state of the experiment table,the inner ring failure,the outer ring failure,the rolling element failure,the cage failure,the loose base,and the friction Vibration data under 9 operating conditions of fault and shaft misalignment fault and unbalance.First,seven major time-domain feature parameters were selected as the criteria for failure,and the average value of 10 sets of feature values under normal conditions was obtained as the basis for judgment.Then use convolutional neural network to judge the type of failure.Due to the problems of low accuracy and poor stability using the traditional convolutional neural network model,parameter optimization and model optimization methods are used in the model.A one-dimensional convolutional neural network model based on batch normalization.Since the convolutional neural network is usually used in the field of two-dimensional images or three-dimensional video,the direct convolution of the collected one-dimensional vibration data is realized by improving the convolution kernel to a one-dimensional convolution kernel,and batch normalization is adopted Layer to prevent overfitting,the method is validated using data collected from the experimental bench.Experimental results show that the average diagnostic accuracy of this method is as high as 98.43%,and it is more stable than other models.This method realizes the adaptive extraction of fault features and the accurate identification of fault types for different fault types of rotating machinery under a large number of samples.In order to judge the degree of damage to the ventilator and give reasonable maintenance suggestions,the PeakVue peak method is used to detect the stress wave generated during the failure of the ventilator,and the impact value after analysis and treatment is used as the judgment standard.And according to the range of impact value,maintenance suggestions are given.Based on the diagnosis process mentioned above,a set of mine ventilator fault detection software was developed to realize the detection function of the ventilator from fault diagnosis to fault type identification to damage degree judgment.This is for the mine ventilator.The operating status diagnosis has practical significance.