Research On Fault Diagnosis Method Of Motor Based On Low Quality Vibration Signal Processing Of Internet Of Things

In today's society,electrical equipment has a higher position in industrial manufacturing.Its emergence and development has brought huge direct or indirect benefits to industrial development.In addition,it also plays an important role in our daily life.Every industry sector and every household needs all kinds of electric motors.As the most common and most abundant power supply equipment and power machinery in the world today,it is involved in almost all fields.As the motor is widely used in domestic and foreign industries,the problem of equipment failure and various huge losses caused by failure deserve our attention.In order to solve the problems of equipment failure and improve the efficiency of industrial production,it is necessary to master the running state of the motor at all times,carry out real-time monitoring,and achieve accurate motor fault diagnosis.In this paper,the motor is taken as the research object,and the wireless sensor networks(WSN)are used to collect the vibration signals of the motor on-line,and time-frequency analysis technology,image enhancement technology and convolution neural network(CNN)are used to classify the motor faults.Traditionally,sensors first convert vibration,sound,motor current,and temperature signals into electrical signals.Then,the data acquisition system(DAS)acquires and quantifies these analog signals to obtain digital signals.The server collects data from distributed Das through cables.Finally,these data are analyzed by using specific algorithms,and then maintenance decisions can be made.With the development of Internet of things technology,it provides convenience for sensor installation,replacement and networking.Experts and scholars have carried out in-depth exploration and Research on artificial intelligence technology.WSN is a network composed of a variety of sensors,which is an important equipment commonly used for monitoring,this technology is also known as the Internet of things.The WSN is usually battery-powered and is often used to monitor the state of mechanical equipment.To extend the battery service life,the length of the acquired and transmitted signal should be short and the sampling resolution should be reduced.In this case,the accuracy of the diagnosis will be affected.To address this issue,this study proposes an enhanced feature extraction method for motor fault diagnosis using low-quality vibration signals acquired from a battery-powered WSN.First,the vibration signal is converted to an image using wavelet synchrosqueezed transform(WSST)technique.Second,the constructed image is enhanced using histogram equalization(HE).Finally,the enhanced image is inputted into a convolutional neural network(CNN)model,and the motor fault type can be recognized from the CNN output.The validity of the method is verified in the brushless direct motor test rig with different motor fault types,and the relationship between transmission time and energy consumption of image resolution data length is studied and discussed.The relationship between the fault diagnosis accuracy and the WSN performances is investigated and summarized.Compared with several traditional methods,it is proved that the method has better comprehensive performance when the length of vibration signal is limited.In conclusion,compared with the traditional methods,the CNN method is suitable for remote motor fault diagnosis,especially when the motor signal is collected and transmitted by WSN of limited power supply.The proposed method shows the application prospect of remote motor fault diagnosis using low-quality vibration signals obtained from WSNs with limited battery capacity.This method has potential application value in the condition monitoring and fault diagnosis of motor in remote areas such as wind farms and offshore platforms.