| With the deterioration of the environment and the shortage of fossil energy,people have more eager to pursue a safe and reliable clean energy to solve the problems of energy shortage and environmental pollution.As a clean and abundant new energy source,wind power has been agreed by everyone.After more than ten years of development,the proportion of wind turbines in my country has been increasing year by year.With the increase in operating years,many early wind turbines have experienced various failures in key components,and these failures will cause the stability of the wind turbine to continue to decrease.Thus,this paper introduces local mean decomposition into the fault diagnosis of wind turbine gearboxes,and studies in detail the de-drying ability of wind turbines based on local mean decomposition for nonlinear and non-stationary signals.At the same time,multi-scale entropy and energy are used to extract the feature vector of the fault signals.And building a particle swarm optimization least squares support vector machine fault diagnosis model in order to realize the accurate identification and diagnosis of different fault vibration signals in the wind turbine.At first,this paper elaborates the background,purpose and meaning of the topic selection,and then,introduces the research status at home and abroad in three aspects:noise reduction processing methods,feature extraction methods and fault diagnosis methods of rotating machinery equipment vibration signals,summarizes the defects encountered in the process of fan pattern recognition,and presents the research ideas and content of this paper.Secondly,this paper introduces the basic structure and working principle of wind turbines precisely and several typical failures of wind turbines and the causes of failures are explained in detail,including the types and failures of gears,rolling bearings and high and low speed bearings in the gearbox,the main failure types of generators and impellers in wind turbines and the reasons for the failure.Thirdly,in order to deal with the non-linear and non-stationary characteristics of the vibration signal of the wind turbine gearbox,the traditional local mean decomposition method is prone to modal aliasing problems.Therefore,this paper proposes an improved LMD method for noise reduction processing of vibration signals.The verification is carried out with various types of signals collected by the rolling bearing of the fan,and the analysis of the collected signals proves the practicability of the method in this paper.At the same time,the basic difference between empirical mode decomposition and local mean decomposition is introduced.Then,this paper introduces the basic principles of multi-scale entropy in detail,combines multi-scale entropy and energy values as feature vectors and explains the basic principles of least squares support vector machine(LSSVM).Aiming at the problem that the penalty coefficient and the width of the radial basis kernel function are variables,and the different values will directly affect the accuracy of fault diagnosis,this paper uses particle swarm optimization(PSO)to optimize the parameters of LSSVM and builds the fault diagnosis model of particle swarm optimization least squares support vector machine.Finally,taking the wind turbine rolling bearing data as an example,applying it to the fault diagnosis model proposed in this paper ground on improved LMD and PSO-LSSVM,and performing simulation analysis from three different operating states of rolling bearings.It is verified that the method proposed in this paper can accurately extract the feature vector of the fault signal,which has high fault diagnosis accuracy and the value of engineering application. |
