Automated Parameter Identification of SPMSMs Based on Two Steady States Using Cloud Computing Resources

This paper proposes a novel offline parameter identification method of surface permanent magnet synchronous machines (SPMSMs) suitable for large-scale industrial applications based on a cloud/edge computing architecture. Measurement data coming from the drives are collected and stored in a cloud application in which an offline parameter identification is performed using Adaline neural networks (AdNNs). In order to overcome the rank-deficiency issue and minimize the estimation errors, an automated procedure is proposed to choose two optimal SPMSM steady states among the ones stored in the cloud with which to feed the AdNNs. The method has been validated using a hardware-in-the-loop setup using data obtained by means of the simulation of a SPMSM drive. The results achieved show good accuracy of the parameter estimations.