The Fault Diagnosis Of A Gear Transmission System Based On Wavelet Analysis And The Current Signature Of The Induction Motor

With the development of modern industry, the equipments become greater, more integrated, speedier, more automated and more intelligent. In order to monitor the condition of some key equipments and secure their running, condition monitoring and fault diagnosis of mechanical system develop. The assignment is to monitor the running condition of mechanical equipment, to diagnosis and find the fault of mechanical equipment, to provide effective measures for correcting the fault, and instruct the management of equipment and maintenance.It is rather complicated for failure form of gear transmission, which also has various fault forms. The gear transmission has been widely used in vehicle, machine tool, aviation, navigation and so on, so it is very important to monitor the running condition of gear transmission. Vibration resulting from the gear transmission includes a lot of fault information and is used widely in diagnosing its faults. However, because of the difficulty of installing sensor in practice, it is not easy to obtain ideal vibration signal. In past few decades, the speed sensorless control technique of induction motor has been successfully used in industrial production, therefore utilizing currents signature of induction motor to monitor and diagnosis gear transmission becomes a new focus. Since the sensor of current is installed easily, we can gain the phase current signatures. The dynamic character of external load can be reflected by the signal of current in induction motor, by which we can predict the dynamic character of external load. Hence, we get the foundation of fault diagnosis. In this paper, by arbitrary rotating reference (d, q), we set up small-signal model of the association motor-load and give the corresponding transfer function, therefore it is approved theoretically that the current signal of induction motor can reflect the dynamic character of external load.Gear vibration signals behave very complexly, such as broken gear teeth, tooth root cracks, and so on. All these signals change over time and perform non-stationary, and traditional Fourier spectrum analysis has been unable to meet the demand. At present, the methods of time-frequency analysis can analyze the non-stationary signals efficiently, such as wavelet, short-time FFT, etc. In this paper, wavelet analysis method is adopted to deal with the non-stationary signals because of its simple algorithm and flexible application.Before using wavelet analysis, we should overcome the additional frequency components and boundary effect generated from decomposition of wavelet. For the additional frequency components, we eliminate the frequency by adding all detailed signal together; for boundary effect, we adopt a new sub-symmetric method of signal extension which will be presented later. From the simulation of new methods, the effect can be approved.At last, based on small-signal model of the association motor-load in arbitrary rotating reference (d, q), VC++language is used to simulate the running of motor. When the load is different, we can see that the signal of current reflects the character of load very well. The spectrum analysis of simulating data also reflects the frequency character of external load. To sum up, the current signals include all the characters of external load. In addition, by using the wavelet to analyze the single phase current signature of induction motor, we can diagnose the fault of mechanical system.