| In modern mechanical equipment,rotary machinery has a wide range of application and a large number of proportion.Rolling bearing is one of the component that is prone to failure.Therefore,condition monitoring and fault diagnosis of rolling bearing become the key to ensure the normal operation of mechanical equipment.In the field of mechanical fault diagnosis,there are many kinds of fault diagnosis methods,among which vibration signal analysis is the most commonly used method.When machinery fails,its vibration signal often has obvious non-linearity and non-stationarity.Variational mode decomposition(VMD)can decompose complex signals adaptively.Its theoretical basis is solid and reliable,noise robustness is strong and frequency resolution is high.It is an ideal method to deal with non-linear and non-stationary signal,so it is widely used in fault diagnosis research.In this paper,the setting problem of decomposition layer number and penalty parameter in VMD is mainly studied.The specific contents are as follows:Firstly,aiming at the problem that the number of mode in VMD decomposition needs to be set in advance and the setting value will have an important impact on the decomposition result,a method to find the optimal decomposition level is studied.Based on the amplitude characteristic of instantaneous frequency,the optimal decomposition parameter is determined by analyzing the relationship between the maximum amplitude(MA)of each component in VMD process.The signal is processed by MA-VMD method and its root mean square entropy is calculated as the fault characteristic parameter.The support vector machine which is optimized by particle swarm optimization algorithm is used to distinguish different fault types.The validity and feasibility of this method are verified by comparative experiments.Then,aiming at the setting of another important parameter which is named penalty parameter in VMD decomposition method,an optimization method based on water cycle algorithms(WCA)is studied.Penalty parameter and decomposition level of VMD are considered as combination to be optimized in this method.The local minimum envelope entropy of each component of VMD is taken as the fitness value and the entropyminimization becomes the ultimate goal to obtain the final optimization values of the two parameters.The feasibility and superiority of the method are verified by simulation and experimental signals.Finally,A fault feature extraction method based on WCA-VMD and frequency weighted energy operator demodulation is studied for the non-stationary and periodic impact characteristic of rolling bearing fault signal.The FWEO demodulation,Hilbert demodulation and symmetrical differential energy operator demodulation are compared by simulation signal.The optimal component is obtained by decomposing the measured signal using WCA-VMD.FWEO demodulation and envelope spectrum analysis are performed for the component to extract bearing fault characteristic frequency.The validity of the proposed method in extracting fault characteristic frequency is verified by comparative experiments and the bearing fault Diagnosis is realized. |
