Wavelet Transform Applied Research In The Motor Performance Test

This paper focuses on the applications of wavelet analysis theory in the domain of motor test. Taking DC motor test as an example, a creative idea of applications of some advanced signal processing tools, such as time-frequency analysis, wavelet, and instantaneous frequency (IF) evaluation, in motor testing field is presented. The paper systematically summarizes basic theories of time-frequency analysis and wavelet transform based on which an in-depth research of effective IF evaluation based on wavelet ridge and wavelet curve is carried out. In the end a fast T-n characteristic testing system with neither torque nor speed sensor is successfully developed.After carefully comparing various existing motor performance testing systems and PCs and digital instruments based automatic testing systems, an effective testing system with PC and data acquisition card is proposed. The fast T-n characteristic testing system with neither torque nor speed sensor implemented in the paper is an excellent example of this new type. In the hardware level, the system acquires motor armature current by WB current sensor and PCI-1202 card. Hence WB current sensor, peripheral driving circuits, PC-1202 card and its interfaces are respectively discussed in detail in the paper. As a perfect example of the new pattern testing system previously mentioned, the system has distinct advantages because it fully exploits abundant resources and rapid speed of PCs.This paper synoptically and comprehensively introduces basic theory of wavelet analysis. Due to the distinct superiority of wavelet transform (WT) and wavelet packet transform (WPT) in time-frequency analysis of non-stationary signals, instantaneous frequency law can be obtained directly from amplitude of WPT coefficients. Thus an indirect T-n characteristic testing method can be implemented with satisfying precision. The paper also studies fast arithmetic and programming of discrete time quadrature WT and WPT and takes a deep look into their additional applications in motor testing, including denoising, filtering and so on.Starting from basic theory of time-frequency analysis and wavelet analysis of asymptotic signals, this paper investigates several IF law extraction methods and presents an especially effective one using wavelet ridge. This method can rapidly extract speed law from starting current of DC motor with fairly good precision, and thus the torque-speed characteristic can be acquired. Simulations and experiment results show that this system is provided with high speed and satisfactory precision besides getting rid of both torque and speed sensors. Therefore it is particularly fit for those occasions for which torque and speed sensors are difficult to amount. Indeed similar method can be generalized to AC motor performance testing after properly adaptation.