| Induction motors are a kind of rotating equipment that convert electrical energy into mechanical energy.They play an irreplaceable role in various fields because of their simple structure,strong durability and low cost.However,induction motors still fail frequently under severe working conditions,which can damage itself at the least,bring huge economic losses,and even cause casualties.Therefore,faults detection and identification of induction motors have important theoretical and practical significance.Broken rotor bar and stator winding interturn short-circuit faults account for a relatively large proportion of all motor faults,about 10% and 15%,respectively.At present,there are many methods for stator and rotor faults detection and identification and the development is gradually mature.However,most methods require professional and technical personnel to analyze and identify its graph and spectrum results in order to accurately diagnose motor faults.It is difficult to realize automatic identification of stator and rotor faults.Based on these,the research content of this article is as follows:(1)Aiming at the fault characteristics of broken rotor bars are easy to be submerged by the fundamental wave and difficult to identify,the filter Park vector ring broken rotor bar fault detection method based on corrected FFT is proposed.First,the band-pass filter is used to filter out the influence of other harmonics and noise.Then the corrected FFT technology is used to accurately calculate the frequency,amplitude and initial phase of the fundamental component of the Park vector,and the fundamental component of the Park vector is filtered out by the constructed function.Finally,it is judged whether the rotor has broken bar fault by analyzing the Park vector ring of the residual signal.The performance of the proposed method is tested through simulation and experiment and the results verify the feasibility and effectiveness of the proposed method.(2)Aiming at the interturn short-circuit fault,the negative sequence component extraction Lissajous fault detection method based on corrected FFT is proposed.The fundamental components of voltage and current signals are extracted by the corrected FFT and the influence of other harmonics and noise is eliminated and the negative sequence components are formed.Then,the analysis results are presented in the form of Lissajous figure to detect the interturn short-circuit fault.The simulation and experimental results consistently prove that the proposed method can detect interturn short-circuit fault of the motors under light load,half load and full load.(3)In order to realize the automatic analysis and identification of the graphs and spectrums of the detection results,the motor stator and rotor faults identification method based on machine vision is proposed.The first way extracts the contour and shape features of the images through the histogram of oriented gradient and forms feature vectors firstly.Then some of the feature vectors are used as training samples to train support vector machine to obtain a fault classification model.The remaining feature vectors are used as test samples and input into the fault classification model to realize automatic fault identification.The second way extracts the texture features of the images through the local binary patterns and the gray level co-occurrence matrix,and forms feature vectors.Then a part of the feature vectors are used to train the random forest to obtain the fault classification model and the remaining feature vectors are input into the fault classification model to obtain the fault categories of the motor.Finally,simulation and experimental results verify that these two ways are effective and feasible.63 figures,1 table and 79 references are included in this thesis. |
PDF Full Text Request