Research On Multi-fault Diagnosis Of Planetary Gearbox Based On Vibration And Current Signals

With the development of modern science and technology,rotating machinery and equipment have been widely used in many engineering fields,such as precision manufacturing,transportation,aerospace,wind power and nuclear energy,mining and smelting.In recent years,the planetary gear box is becoming more and more common in application as the rotating machinery develops towards high-speed and heavy-load direction.As an important part in the rotating equipment,the operation of the planetary gear box determines that of the entire equipment.Therefore,it is of important safety and economic significance for the status monitoring and fault diagnosis of the planetary gear box.However,in the event of multiple faults that happen to the planetary gear box simultaneously,it is very difficult to accurately and timely extract the fault features due to the mutual modulation and submersion of the fault features.This paper made multi-fault diagnosis study on planetary gear box from three aspects,namely vibration signals,current signals and fusion signals according to the technical difficulties resulting from mutual modulation,mutual submersion and extraction of weak fault signals and others when multiple faults occur at the same time in a multi-stage planetary gear box.The main contents of this paper mainly include the following aspects:(1)It proposed the fault diagnosis method of variational mode decomposition and envelope analysis based on vibration signals.By using the difference in engaging frequency of gear pair within the multi-stage drive gear box,this paper used methods to decompose the vibration signals into high and low frequency bands at first to separate the vibration characteristics of planetary gear and spur gear,and then decomposed the component signals further by envelope demodulation to extract the fault characteristics of planetary gear and spur gear respectively.(2)It proposed the fault diagnosis method of resonance-based spare decomposition and envelope spectrum of instantaneous frequency based on the vibration signals.As the vibration characteristics caused by fault of planetary gear and spur gear vary,this paper separated the fault characteristics of planetary gear and spur gear with resonance-based spare decomposition,and then further demodulated the decomposed signals to extract the fault characteristics of planetary gear and spur gear respectively by the method of instantaneous frequency and envelop spectrum.(3)In order to realize the multi-fault diagnosis in variable-speed working conditions,it proposed the fault diagnosis method of uniform angle sampling and variational mode decomposition based on vibration signals.At first,this paper used uniform angle sampling to transfer the vibration signals into steady state signals,and then realized multi-fault diagnosis by means of variation empirical mode decomposition.(4)In order to realize the multi-fault diagnosis based on current signals,this paper proposed the intelligent diagnosis method based on current pre-processing and deep neural network.First,it used the method of delay addition to remove the power frequency component effectively,and then modified the current signal frequency spectrum after removal of power frequency,and finally took the modified frequency as sample to classify the types of multi faults effectively by using intelligent classifier of deep neural network.(5)In order to further improve the accuracy of deep neural network intelligent diagnosis method in the vibration of planetary gear box with multiple faults,this paper proposed the intelligent diagnosis method based on fusion dimensionality reduction,First,it obtained the original high dimensional multi-domain feature set representing the running state of planetary gear box;and then,used the maximal relevance and minimal redundancy method to select features and filter features;next,used linear discriminant analysis to reduce the dimension of fault characteristics;finally,adopted deep neural network method to classify the fault types effectively.