Study On Embedded Intelligent System Based Online Detection For Broken Rotor Bars In Induction Motors

Because of their simple structure, low price and high reliability, squirrel-cage induction motors are widely used in many aspects. But mechanical faults such as breakage of rotor bars often happen owing to various stresses. Therefore online detection on induction motors is necessary.The method adopted in this paper is called Motor Current Signature Analysis (MCSA). It detects broken rotor bars according to the amplitude of the (1-2s)f1 frequency component in stator current signal. This paper proposed a novel method, which is based on continuous and subdivided Fourier Transform (FT), wavelet packet decomposition (WPD), BP artificial neural network (ANN) and 'energy-fault' diagnosis mode for the first time.Under steady load condition a method based on continuous and subdivided FT is proposed. The stator current signal under operation condition is analyzed with continuous and subdivided FT. A novel arithmetic is proposed for the first time to counteract the fundamental component. The problem that the (1-2s)f1 frequency component is submerged by the fundamental component is solved completely.WPD is adopted because the spectrum of the stator current cannot be gained correctly by FT when the load torque is fluctuant. The WPD coefficients of stator current signal under operation condition are used to judge whether the load torque is fluctuant and to read the fluctuant points. The signal's energy in the range of steady state is used to calculate the amplitude of the fundamental component, which is selected as the input of BP ANN together with the energy of WPD coefficients in the range of steady state of special node. The result of ANN is the amplitude of the (1-2s)f1 frequency component. The information of fault degree is gained from two amplitudes abovementioned. Under steady condition this method and the above method based on continuous and subdivided FT can testify each other to increase the detection veracity.An intelligent online detector based on embedded PC system for broken rotor bars is manufactured successfully. This system has advantages of high integration level, small volume, low power, high stability and reliability. Moreover, it has the same program protocol with ordinary PC. Experiment results show the accuracy of the proposed method for broken rotor bar detection in squirrel-cage induction motors under both steady and fluctuant load condition.