A novel application of the fast simulated diffusion algorithm for dynamical parameter identification of brushless motor drive systems

The application of the fast simulated diffusion (FSD) algorithm as a practical method of motor dynamical parameter identification with a multiminima cost function is presented. The effectiveness of this parameter optimization technique in obtaining global minimum estimates of the motor shaft friction and load inertia is demonstrated in the simple case of a high performance brushless servodrive system (BLMDS) over a response surface with embedded local extrema. The restriction of the search procedure to quantized step sizes in parameter space, due to noisy `local´ minima proliferation near the global extremum is examined. The accuracy of the FSD search method in acquiring the global minimum parameter estimates with remote initialization is demonstrated by the use of step response current feedback, responsible for genuine local minima plurality, in the objective function formulation. The necessity for a good physical model of the BLMDS with nonlinear inverter delay operation as part of the identification strategy is briefly discussed. Frequency and phase coherence of the step response BLMDS current feedback with model simulation, using returned FSD parameter estimates, attest to FSD convergence accuracy which validates this parameter identification method