The Application Of Modern Signal Processing In The Motor Vibration Analysis

With the wide use of the motor in the modern industrial production, the vibration is themost important problem in operation of the motor, which affects the stability and productionof equipment, so motor vibration analysis gradually attract people’s attention. Sensor collectsthe mixed-signal which is consists of the other source of vibration signals mixed together in avariety of ways which is very complex, so we adapt the modern approach blind sourceseparation technique to separate the vibration signal. Blind signal separation technology isdefined when the source signals and transmission channel is unknown, recover theindependent signals from the observed signal only, which is extensively used in voiceprocessing, signal processing, image processing. So many researchers pay attention to it andits basic theory and research applications is further deepened.The object of this paper is motor vibration acquisition signal based on the inverter andapplication of fast independent quantitative analysis (Fast ICA) algorithm to separate themotor acceleration, speed, displacement vibration signal which under the control ofconventional PID and neural networksPI αDβ. Power spectrum was analyzed, andcharacteristics of the frequency causing the motor vibration were summarized, moreover,comparison of the two controllers’ properties were made.First, the article described significance of the development of motor vibration analysis,summed up foreign and domestic development status of the blind source separation and thenintroduced the traditional ways to analyze motors: the time domain analysis, signalpreprocessing, frequency domain analysis.Second, based on the theory of blind source separation, the mathematical model,stability and uncertainty of the separation, the independence criterion and optimizationalgorithm, the article highlights the FastICA algorithm，it’s a fast algorithm of independentanalysis, which is suit for observed variables with a large number of sampling points, andwhich is suitable for separation of the motor vibration signal.Finally, the paper designs the motor vibration signal hardware environment, andvibration data acquisition scheme, plenty of charts were used to show the motor acceleration、speed、displacement data curve and power spectrum under the control of conventional PIDand neural networkPI αDβ, analyzing the vibration character of the motor and making acontrast between the two controllers.