The short-term life prediction model of gearbox based on chaotic neural network

Since faults of gearbox occurred randomly during the normal status, chaos theory was chosen to analyze the nonlinear characteristics of vibration acceleration signals for gearbox. The short-term prediction model of chaotic neural network for gearbox life was proposed based on chaotic time series. In the model, the chaotic time series phase space was reconstructed as the input vectors of neural network, and the predictable step of gearbox was set as the output vectors of neural network, then the short-term life of gearbox was obtained. The results of the simulation on the vibration acceleration signals of the test-gearbox showed that the model is more effective and accurate compared with the traditional neural network prediction methods.