Variable Predictive Model Based Class Discriminate In The Application Of Rolling Bearing Fault Diagnosis Research

Rolling bearing is an element of mechanical equipment which is used most andeasily damaged. When a rolling bearing is broken down, it can cause the failure of thewhole mechanical equipment. So it is very important to monitor and diagnosis theworking conditions of rolling bearing. There are many methods for fault diagnosis ofrolling bearing，but the most important is diagnosis method of bearing vibration signal.In fact, the process of rolling bearing fault diagnosis is essentially a patternrecognition. The pattern recognition process is firstly to extract the feature value ofvibration signal, and then recognition of characterist ic value can be conductedthrough classifier.In this paper, an approach of rolling bearing based on VPMCD is proposed. Themain research contents of this paper are as follows:1.To pattern recognition for rolling bearing is first for the extraction ofcharacteristic value of bearing signals, as for the acquisition of rolling bearingvibration signal is often non stationary and therefore the need for signal processingLCD (local characteristic scale decomposition) talked about in this paper is anadaptive processing method for non stationary signal. As a result, experiments showthat the decomposition rate and the devoicing ability of LCD are good.2. The precision of pattern recognition is also determined by the benefits andweaknesses of the pattern recognition methods. In the field of pattern recognition,VPMCD (variable predictive model based class discriminate) is a new method. Paperstudied the basic principle of VPMCD method, and the VPMCD method application inthe rolling bearing fault diagnosis are studied, the experimental results verify theeffectiveness of the VPMCD method3. This paper also according to the drawback in the parameter estimation processof the VPMCD, the VPMCD approach is improved，its parameter estimation approach,least square method, has been replaced by ridge regression. The ridge regression caneliminate the influence of the multicollinearity among the independent variables. Atlast, it can get more precision model parameters and improve the precision of patternrecognition. The analysis results from the rolling bearing vibration signals of differentworking conditions demonstrate the good effect of this method.4. In the case of small training samples, VPMCD may not be the most can reflectthe intrinsic relationship among the variables of the model or not precise model parameters. This paper proposes the use of genetic simulated annealing algorithm tooptimize the model weights. The optimal weight matrix and the prediction on the testsample value matrix are fused, and with the discriminate function to achieveclassification and recognition. Experiments are carried out through analysis of bearingdata; the results show that the method is effective.