Optimal Artificial Neural Network-based Maximum Torque Control for Permanent Magnet Synchronous Machine in Modular Cascade Machine System

This paper presents Artificial Neural Network (ANN)-based Maximum Torque Control for the Permanent Magnet Synchronous Machine (PMSM) in a Modular Cascade Machine (MCM) System. Several ANNs with different numbers of hidden layers and the number of neurons per hidden layer are trained offline to predict the d-axis and q-axis components of the stator reference current in rotor reference frame. The operating points for ANNs training, validation, and testing are determined using Maximum Torque per Ampere (MTPA) algorithm, Current Voltage Limited Control (CVLC), and Maximum Torque per Voltage (MTPV) algorithm across the full speed and torque range within the safety limits. The reference operating points determined from these approaches ensure increased efficiency, torque density, and speed range compared to conventional field-oriented control algorithms. The developed control scheme with Optimal ANN based on the validation and testing performance is implemented on a 50 kW Toyota Prius 2004 model PMSM with the speed profile inspired by electric vehicle motion on an inclined plane. Comparative Simulation results obtained from the conventional Lookup Table approach indicate that the developed control has superior and reliable torque tracking capability with reduced oscillations in the terminal voltage and current originating due to the sudden change in motion parameters.