| Abstract |
Vector control of AC machines provides high performance and uncoupled control of flux and torque. But wrong electric parameters can induce torque ripple and instability. They decrease dynamic behavior of the system. Thus, unbiased and accurate estimation of parameters must be performed. This paper presents a comparative study of identifiers based on deterministic or stochastic optimization tools: Simplex, Simulated annealing and Genetic. It gives a short explanation of their behavior and a presentation of their main characteristics. These algorithms are tested on benchmarks. The influence of the measurement noise and the effect of the initial estimated vector , on the final estimated parameters of a simulated induction machine, is studied. Finally, these tools are applied on a real case. |
