A Model-Free Direct Control Solution for Multiphase Electric Drives Based on Gradients of Electric Variables

The environmental situation requires the use of more competitive electric drives. In this regard, multiphase systems provide several advantages over conventional three-phase drives, due to their higher number of phases. However, to exploit their benefits, high-performance regulation techniques are necessary to satisfy their higher complexity. Focusing on the control of multiphase drives, model predictive control is employed as an interesting option thanks to its flexibility and suitable usage of the dc-link voltage. Unfortunately, this regulation technique has historically required the estimation of the electrical parameters since a machine model is employed to estimate future currents. To reduce the dependence on the machine parameters, model-free predictive schemes have been considered as a potential alternative. Nevertheless, according to the current state-of-art, model-free proposals for electric drives generate a considerable increase in the computational cost or entail the online parameter estimation. Looking for a model-free predictive technique with a reduced computational burden and a simple implementation, this work proposes a version of this direct strategy based on the comparison between the normalized gradient of the current response and control actions. Simulation results for a nine-phase permanent magnet synchronous machine validate the performance of the proposed model-free direct control (MFDC).