Current Signature Analysis And Health Sensing System Implementation For AC Motor

Stator and rotor fault possess a high proportion in motor failure modes The effective monitoring and real-time fault diagnosis for motor healthy state are available measures to avoid catastrophic consequences caused by rotor and stator failure. Under this background, motor current signature analysis methods and feature signal processing algorithms used for rotor and stator fault diagnosis are summarized systematically. Based on the analysis of the shortcomings of these existing methods, a view is indicated that the state cannot be detected effectively depending on the traditional single algorithm, and two simple fault diagnosis algorithms are put forward. Finally, the methods are verified by experiments. The main contents and innovation work are as follows:(1) The features, causes and fault symptoms of the state with broken rotor bar and stator short circuit of AC motor faults are analyzed deeply. The mechanisms of different fault characteristic frequencies are explored from the perspective of electromagnetism.(2) Basing on the multi-loop mathematical model of asynchronous motor, a simplified mathematical model and a simulation model of two phase coordinates are set up by coordinate transformation. Through changing load signal properties or bringing in different sizes of fault resistance in rotor, stator winding in simulation model, the formal, overload operation, stator or rotor asymmetric fault states of power frequency and frequency conversion motor are simulated. Feature of three phase current in different states are analyzed. Furthermore, diagnostic characteristic parameter and its change rule are clear and definite.(3) After comparing and analyzing the disadvantages of the current signal feature extraction algorithms including the fast Fourier transformation method, phase difference of three phase current method and Park vector transformation method, current signal envelope analysis method and amplitude demodulation method are put forward. The detection features are reflected in the normalized amplitude figure and modulation signal spectrum, which realized the separation of power frequency component and different failure frequency component. Simulation results showed that the combination of vector transportation and amplitude demodulation can break the limitation of imperfect detection by single algorithm. Besides, the method can be used in state monitoring of frequency conversion motor, the rotor fault, stator fault and their composite fault diagnosis.(4) AC motor health perception system architecture is designed, and the system software platform is established at the same time. An intelligent human-computer interaction system is accomplished, which integrates functions of current signal perception, data collection, data analysis, signature extraction, test and diagnosis, result display and so on. It realizes on-line and off-line diagnosis of motor states. The validity of the methods and the diagnosis system is verified by their application of the actual equipment.