Study On Fault Diagnosis Of Gearbox Based On Particle Filter

The modern production is constantly moving in a high-performance, large-scale, highlyautomated direction.Gear box plan an ts of the failkure frequency and the detection ofimportant role in the machinery and ethe fault and the fault will be recognisede and have greatsignificance. However, in the process of the operationg of the gear box ,there is backroundnoise, making the collected vibrationg signals are often drowned in the noise, making itimpossible to identify the fault. To be able to accurantely diagnoise faules ,the collectedvibration signal pre-processing, and in order to diagnose fault,collected signal is processed forone thing to enhance the Signal-to-Noise ratio.Particle filtering is a new model-based state estimation technique. Studying the principleof particle filter in-depth then use it to the noise reduction of gear vibration accelerationsignals. The signal model and noise statistics should be known when use particle filtertechnology to denoise. In this paper, it’s realized as follows: establish time series AR model ofthe vibration acceleration signal.The coefficients of dynamic equationgs required to determinethe coefficients in the modeling process. FPE criteria for model order and the least squaremethod to calculate the estimated parameters are used in the establishment of the AR model.In this paper, the test object is JZQ250, according to the expertimengtal environment andbackground, as well as the failure of the gear fallues in the rorm vibrationg characteristics,theuse of the coefficients and experiment project on condition testing and fault diagnosis of thegearbox is designed.Finally, we can see the value of noise dater characteristics are superior to the former. Asan adaptive pattern recognition technology, neural network has been widely applied in thefield of fault pattern recognition and the theoretical research is mature. In this paper, we usedatas of denoised to diagnose faults by BP neural network, and the diagnosis results show thatthe data denoised have a good effect.