Reducing induction motor identified parameters using a nonlinear Lasso method

Reliable induction motor modeling is critical in power system planning and operation. This paper considers the identifiability of induction motor parameters, with a particular emphasis placed on using subset selection and shrinkage methods to allow the identification methods to focus on the most significant parameters. The proposed approach is validated using experimental data and the results found are compared to those of a recently proposed method based on sensitivity analysis.

► An extended Lasso method was applied to the induction motor identification problem.
► The obtained results show that the proposed method is capable to determine the order of importance of the parameters.
► The Lasso ordering causes less error for the validation data after fixing a similar number of parameters compared with a sensitivity based approach.
► The Lasso method takes all the outputs into consideration along with the sensitivities, as it uses the Jacobian and the error vector both, while sensitivity of each output can be different for the same parameter.