Detection And Diagnosis Of Motor Controlling System Fault Based On Wavelet Transform

Subject investigated of this paper is the method of fault diagnosis of wavelet transform which is often used in large-scale in-phase no-brush excitation motor in large-scale production enterprise. Because of its rolling commulate set circumgyrating with the motor's armature together, motor often suffered centrifugal force and vibration .It is often damaged and destructed, and cause that system can't work in order and get in trouble. This kind of fault can't be detected directly and discriminate ahead. This paper introduces a kind of the method to detect fault with wavelet transform, in order to let system recognize the fault ahead which is already connotative and existent during the work process, and make the judge within a short time, then it send out the warning to system. It makes system to produce a corresponding method to deal with. In this way, not only protecting the controlled object, but also it avoid the economic losses. So it is very significance to improve the system security and reliability.Capital work of this thesis is: First, we gather the sampling signal from in-phase no-brush excitation motor of falling to the groud experiment simulation device. Second, we analyzed the simulate dates with three means based on theory of wavelet transformFirst, we adopt the convenient and shortcut MATLAB 1-D wavelet toolbox , and then analyzed signal with multiresolution , selected the db8 as mother wavelet, 6 as scale, and sampling date is 91, get out a result.Second, we use programming language of MATLAB, selected the dmey as mother wavelet, 5 as scale, and sampling date is 91, get out a result.Third, we use MATLAB programming language to deal with the date, which is based on DFT transform spectrum analysis, and get out a result.Comparing the wavelet transform with spectrum , we can draw a conclusion.(1) Spectrum can reflect the signal character entirely and detect the fault of equality efficienly.(2) Wavelet transform can analysis the signal character entirely and detect the fault with multiresolution , it is similar to spectrum analysis .(3) It is no influence for spectrum analysis on abrupt and brief signal of fault, it is to say that the fault can't be detect with spectrum.(4) It is definitely for wavelet transform to point out the band range and the moment of occurrence for abrupt and brief signal, it is to say that the fault can't be detect with the theory of wavelet transform, which can't be proved in this experiment.The results show that the theory is right.