Research On Gearbox Fault Diagnosis System Based On AR Model

In recent years,the world's wind power industry has developed by leaps and bounds.In 2016,the global installed capacity of wind power generation reached more than 54.6 GW.The huge number of wind turbines brought huge economic benefits and brought difficulties in maintenance and repair.Wind turbines have a bad working environment,and gearbox failure is the main cause of wind turbine failure.Once the fault causes the shutdown,it will bring huge economic losses.Therefore,monitoring the working state of the gearbox and detecting possible failures can prevent problems before they occur.Aiming at the problem of fault diagnosis of fan gearbox,Burg method and FPE method for parameter estimation of AR model are further improved.In the process of AR parameter calculation,the average forward prediction error and backward prediction error are used to calculate the reflection coefficient.The new reflection coefficient is used to calculate the AR parameters of each order.The final prediction error can be reduced in the short data section collected.Improve the accuracy of prediction.The minimum average prediction error is used to determine the optimal order in the FPE criterion,which improves the phenomenon of higher order determination in the FPE criterion.In the aspect of fault identification,AR parameter model and power spectrum analysis are combined to locate and identify faults accurately.Firstly,AR parameter models of different fault types and power spectrum of normal operation are established,and then the tolerance range of parameters under normal conditions is determined.In the final fault identification,the fault type is identified by matching AR parameter model,and the fault location is determined by power spectrum analysis.The two methods are combined to locate the fault point.In addition,according to the performance requirements of gear box vibration signal detection system,the hardware and software parts based on ARM platform are designed.The hardware includes ARM core board,signal acquisition module,communication module,man-machine interface module,data storage module,JTAG interface module and so on.The software part includes the transplantation of system software platform,software program design and man-machine interface design.The algorithm of AR model and the software and hardware of the system constitute the fault diagnosis system.It is verified that the system can identify faults accurately and has practical application value.