Research On Gear Fault Signal Analysis And Diagnosis Method

Gear is one of the important parts in mechanical equipment.Because of its high transmission efficiency,transmission accuracy,reliability,and wide power transmission range,it is widely used in various mechanical equipment.As a key component,it is also a component that is more susceptible to damage.If the damaged gear can be found and repaired in time,a lot of economic costs can be saved and serious accidents can be avoided.The use of vibration signals generated during the operation of gears for gear status detection and fault diagnosis classification is of great significance in the actual work of gear equipment.If the fault features can be extracted accurately and effectively in the process of gear fault diagnosis,the accuracy of fault diagnosis will be improved,and the analysis of fault types and fault levels will be more specific.Therefore,this paper builds a gear failure simulation experiment platform to collect the vibration signals of the faulty gear running at different speeds.Two fault diagnosis methods,which are based on the one-dimensional feature and two-dimensional feature obtained by signal processing,are used to classify faults for gear fault diagnosis.The main research contents of the article are:1.The typical gear damage types are introduced,the gear damage mechanism is analyzed using the gear dynamics model,and the common characteristic indexes of gear failures are compared and analyzed.A gear failure test bench was designed and built.The test bench can perform failure simulation experiments on multiple types of gears at multiple speeds and multiple load conditions.A typical damage simulation of the gear required for the experiment.Debug the parameters of the data acquisition system and perform fault simulation experiments,collect and store vibration data,and obtain data samples required for subsequent fault diagnosis experiments.2.Introduce the basic principles and calculation process of EEMD and SVM.Use EEMD to extract IMF from the collected gear vibration signal data samples,in order to make the results of the fault diagnosis experiment comparable,the energy features of the first five-order IMF are extracted to form the feature vector,and several time-domain features of the vibration signal are extracted to form the feature vector.Use the particle swarm optimization(PSO)algorithm to optimize the penalty coefficient C and the kernel function parameter σ in SVM.After that,the two feature vectors are used as the input parameters of SVM to realize the classification and identification of gear faults based on EEMD-SVM.3.Introduce the basic principle of WT and the working principle of each layer structure of CNN.The vibration signals of different speeds and different types of faults are transformed by WT to obtain a two-dimensional time-frequency characteristic map of the signal.The characteristic map is used as the input parameter of CNN,and the CNN is used to classify and identify the fault in the time-frequency characteristic obtained by the WT transformation.The output features of the convolutional layer are processed by PCA dimensionality reduction,and the better experimental results verify the applicability of the method in fault diagnosis.