Research On The Intelligent Diagnosis Method Of Typical Mechanical Equipment Fault Based On Acoustic Signal

Currently gear fault diagnosis is mainly based on vibration signals.However,vibration signal acquisition is limited by its contact measuring in some special environments.Mechanical noise is the result of mechanical vibration propagating to the outside through elastic media,which contains abundant equipment status information.Fault diagnosis method,represented by acoustic signal measurement,has the advantages of non-contact measurement can partly replace vibration signal in some special environments as a means of fault diagnosis.In traditional acoustic diagnosis technology,only a single sensor just can be used to acquire the local acoustic characteristics of mechanical equipment,and the anti-interference ability is poor.The selection of measuring point location needs rich experience.While,by using the multisensor and multi-channel diagnosis method,the acoustic signals of the whole radiated sound field under different working conditions can be fully obtained.So the changes of the sound field at the key parts of the machine can be highlighted,and the changes of the sound field at the interference parts can be neglected,which will improve the diagnostic effect.However,the procedure of determining the weight value of multi-channel acoustic diagnosis is complex,which make it develop slowly.In recent years,the intelligent diagnosis technology based on deep learning theory has developed rapidly.Especially,the combination of deep neural network and time-frequency analysis has injected new vitality into machinery.Deep neural network has the function of parameter self-adapting adjustment.It does not depend on expert knowledge base and engineering practice experience.It has ability to enhance recognition accuracy through reasonable optimize the weight of multiply channel by training.According to above thinking,we propose an intelligent diagnosis method for typical mechanical faults based on acoustic signals.Taking the noise signal in fault state as the research object.We use the deep learning theory as the signal processing and pattern recognition method to study the gear fault diagnosis method based on acoustic signal in different working conditions.The main research work are as follows:First of all,By setting up an experimental platform to simulate the fault state of mechanical equipment in actual work,and aiming at the situation of multi-equipment and multi-sound sources in the process of sound signal acquisition,a reasonable measurement scheme is worked out.Through reasonable spatial layout of multi-channel and multi-volume microphones,the reflection of sound signal in air transmission can be reduced.Diffraction phenomenon,improve the anti-interference ability of sound signal,and then use the microphone to collect the noise signal of mechanical equipment failure parts under different working conditions to obtain fault data.Secondly,To solve the problems of the traditional acoustic method,a multi-channel acoustic diagnosis base on convolutional neural network for gears fault diagnosis is proposed.The fault information of different sensitivity is obtained by placing microphone at different measuring points.Then,the convolutional neural network is used as the fusion technology to fuse the four channel acoustic signals of gears at feature level to realize the fault diagnosis of multi-stage transmission gears.The experiment result show that the method that we propose for gears fault diagnosis based on multi-microphone information has achieve higher accuracy in fault type recognition of gears which is higher than traditional method that base on a single microphone with multiple manual features.Finally,a novel deep learning-based gear fault diagnosis method is proposed based on sound signal analysis.By establishing an end-to-end convolutional neural network(CNN),the time and frequency domain signals can be fed into the model as raw signals without feature engineering.Moreover,multi-channel information from different microphones can also be fused by CNN channels without using an extra fusion algorithm.Our experiment results show that our method achieved much better performance on gear fault diagnosis compared with other traditional gear fault diagnosis methods involving feature engineering.