PWM spectrum analysis and optimization for model based condition monitoring and control

This paper identifies several PWM optimisation possibilities for model based condition monitoring and control, with particular reference to induction motors. Future exploration and implementation of these possibilities via spectral analysis simulation is justified in terms of the complexity of these optimisations, and the lack of sufficiently general analytical tools. Detailed simulation results are presented for the traditional analogue sine-triangle algorithm and a commercial device with symmetrically sampled digital space vector modulation and third harmonic injection. Close agreement between simulated and measured results for the commercial device indicate the sufficient level of detail in the simulations. The application of a field programmable gate array and a software tool for rapid firmware prototyping in future practical investigations is briefly described. Simulated and measured results are presented for a prototype firmware PWM algorithm proposed to replace the commercial device. A case study of PWM optimisation for noninvasive online numerical impulse response measurement is also presented. The theory and implementation of direct PWM injection of pseudorandom binary sequences for numerical impulse response based system identification is discussed. Simulated and measured numerical impulse responses are presented for a simple three phase load example.