Five phase multi-layer drive with fault tolerant neuro-fuzzy features for safety critical applications

This article presents a method for control of a multiphase motor drive based on the use of a cognitive network that is capable of modelling nonlinearities of the system in fault conditions. To obtain a faithful model of the drive, a neural network (trained with the results of simulated faults) is used in the control algorithm in order to predict the system performance and adapt the control. A new multiphase motor with variable characteristics, particularly suited to applications in which a very wide variability of the speed range is required, will be described and presented in its structure. The technique to remedy failure adopted for this drive will be presented and described in its characteristics. The effectiveness of the control technique lastly will be highlighted by simulations that put it in comparison with classical solutions that employ PID controllers.