Research On Bearing Fault Prediction Based On Full Vector Extreme Learning Machine

The current machinery and equipment toward a more efficient,integrated and large-scale development,so how to ensure the safe and stable operation of equipment has great significance.With the development of computer technology and sensor technology,fault prediction theory and technology are also developed to provide a better solution to the above problems.But the traditional way of information collection is to use a single sensor to collect single-channel information for analysis,because the information is not comprehensive,and the reliability is not very good.Full-spectrum technology combines the two orthogonal channels of the signal,which can more fully reflect the equipment running accurately.The prediction algorithm has always been a hotspot in artificial intelligence research.In recent years there have been a series of new algorithms,such as extreme learning machine with good accuracy and reliability,and it also has a small amount of calculation,the characteristics of fast.In this paper,the full spectrum technique is combined with the neural network and the Extreme Learning Machine,and the rolling bearing is used as the object to carry on the fault prediction research,the research content includes the following several aspects:(1)The full vector hilbert demodulation method is used to extract the characteristic frequency of the rolling bearing.Through simulation and case analysis,this method can effectively extract the characteristic frequency and lay the theoretical foundation for the neural network training sample extraction.(2)A fault prediction model based on full vector fuzzy neural network is proposed and verified.The results show that,because of the defects of fuzzy neural network,the prediction results are unstable,which can easily lead to misdiagnosis.(3)A fault prediction model based on full vector wavelet neural network is proposed and verified.The results show that the neural network can be improved by wavelet basis function,which can improve the effect of neural network in rolling bearing fault prediction.The wavelet neural network still does not solve the defects/shortcomings of traditional neural network.It still uses the gradient descent learning method.So it is easy to fall into the local extreme value and affect the prediction model effect;(4)Aiming at the shortcomings of the traditional neural network,the bearing failure prediction model of the full vector extreme learning machine is proposed.The example shows that it can fit the expected output well,the error degree is small,and the prediction of the rolling bearing fault was very good stabilized and accurated.Extreme learning machines do not need to set a large number of parameters like traditional neural networks.They only need to set the number of neurons in the hidden layer.So it's computationally small and fast,with good real-time performance and usage value.