Research On Fault Diagnosis Method Of One-Dimensional Convolutional Network Equipment Based On Wireless Sensor Networks

Wireless sensor networks have attracted the attention of researchers in the field of fault diagnosis because of their low cost,flexible and convenient installation and certain data processing ability.Most of the existing researches only use the data transmission function of wireless sensor networks to transmit the original data to the computer to realize fault diagnosis,but wireless sensor networks often use low-speed wireless data transmission protocol,and its limited bandwidth can not transmit a large amount of data generated by the equipment fault diagnosis system in time.In addition,the mechanical equipment fault diagnosis method based on vibration signal often needs to rely on manual experience and expert knowledge for feature extraction,and the fault diagnosis method under the traditional machine learning theory has insufficient learning depth and poor diagnosis effect.In order to solve the above problems,exploring an efficient algorithm that can make full use of the data processing ability of the nodes,reduce the amount of data transmission and deeply learn the fault characteristics from the original vibration signal has become the key of the current research.This paper presents a fault diagnosis method based on wireless sensor network and one-dimensional convolutional neural network.This method can directly process the original data on the wireless sensor network nodes to realize fault diagnosis,and only transmit the diagnosis results.This method gets rid of the dependence of traditional fault diagnosis algorithms on expert experience,reduces the amount and time of data transmission,and saves the energy consumption of nodes.The specific work contents are as follows:Firstly,the mainstream fault diagnosis methods were sorted out,and two convolution network models suitable for direct processing of one-dimensional vibration signals were explored and designed.Then,the simulation experiment was completed based on MATLAB platform.Finally,a more suitable fault diagnosis model for wireless sensor network nodes is selected thro ugh comparative analysis.Secondly,the experimental hardware platform based on wireless sensor network nodes was built,including End Device nodes,Coordinator node,upper computer,oscilloscope for judging node energy consumption and 10Ω resistance.The software design of fault diagnosis system was completed by using C language programming.Including the realization of fault diagnosis and wireless transmission based on one-dimensional separable convolution network model on the wireless sensor network End Device node,the realization of wireless reception and wired transmission functions on the Coordinator node,and the design of serial port display software using Lab VIEW language programming on the computer.Finally,the experimental verification was completed by using the data of Case Western Reserve University storage bearing data center.The experimental results show that the diagnosis accuracy of the proposed fault diagnosis system can reach 100%,and compared with the diagnosis method of directly trans mitting original data,the system can reduce the payload transmission data of wireless sensor networks from2048 bytes to 2 bytes,saving 15.3% of node energy consumption.