Research On Rolling Bearing Fault Diagnosis Based On Volterra Level

Rolling bearings are the basic components of most rotating equipment and are also the most vulnerable mechanical parts.Choosing the right fault diagnosis method plays an important role in reducing production costs and increasing productivity.The Volterra series model has a clear physical meaning that reflects the essential characteristics of the system and is applicable to the study of most non-linear systems and has been applied to many fields.In this paper,the Volterra time-domain model is applied to the fault diagnosis of rolling bearings,and the value of the Volterra time-domain model based on the hybrid particle swarm algorithm is demonstrated through simulation and experimental investigation.The main research elements of this paper are as follows:Considering that the particle swarm algorithm has outstanding search capability and uses randomly changing particle velocity in the solution space to make the motion random,thus achieving the feature of reducing the computational difficulty and finding the global optimal solution faster,the Volterra time-domain kernel identification method based on the particle swarm algorithm is proposed.The method has been studied by simulation and is found to be effective in the presence and absence of noise.Experimental results show that the Volterra timedomain model can significantly represent the characteristics of the system and can effectively distinguish between the two states.In view of the excellent characteristics of genetic algorithms and global optimisation properties,a genetic algorithm based Volterra time domain kernel identification method is designed.The method is applied to Volterra level model identification,and the simulation study results show that the method has excellent identification effect.Finally,the Volterra timedomain model of rolling bearings in both normal and fault conditions is constructed,and the state differences can be accurately distinguished by observing the system characteristics.The problems of the particle swarm algorithm and the genetic algorithm in Volterra time domain kernel identification are analysed,and an effective combination of the particle swarm algorithm and the genetic algorithm is proposed.The inertia weights and individual learning acceleration factors in the particle swarm algorithm are adjusted,and on the basis of this idea,the selection,crossover and variation ideas of the genetic algorithm are embedded into the particle swarm algorithm to compensate for the problems existing in the operation of the two algorithms,and the hybrid particle swarm algorithm is proved to be more effective through simulation studies.The performance is better than particle swarm and genetic methods.It is proved that the hybrid swarm algorithm is an important practical application of the Volterra time-domain kernel identification method.After combining the Volterra level model to complete the fault diagnosis,a support vector machine fault classification method based on the Volterra time domain kernel is proposed.The GPSO method is used to identify the kernel as a feature vector for the different orders of the rolling bearing’s time domain,and then the bearing data is input into the support vector machine for fault classification.The experimental results show that using the first three orders of the kernel as a feature vector can obtain more information about the rolling bearing faults,and that the non-linear variation characteristics of the system are not only reflected in the first order kernel,but also in the higher order kernels such as second and third order.The effectiveness of this fault classification method is also demonstrated through the study.