Reinforcement learning (RL) algorithms enable automated controller configuration without plant-specific parameter knowledge and allow consideration of multiple objectives in a single training phase. Emerging developments in the domain of electric drives feature a holistic design approach and exhibit efficient drive operation after only few minutes of learning. The exploration required for this includes random actions to collect informative data during the training phase, which may lead to unsafe state reactions and thus drive damage. Accordingly, this contribution focuses on safety-guaranteed RL-based direct torque control (DTC) for a modulator-driven permanent magnet synchronous motor, i.e., the applicable voltage domain is considered as a continuous control set (CCS). A safeguarding algorithm is derived to secure the online training phase, preventing any voltage with unwanted consequences from being applied to the motor, e.g., to avoid overcurrent. The proposed CCS-RL-DTC is applied and evaluated on a real-world test bench, where comprehensive functionality and performance investigations are conducted with concern to the safeguard, the online training phase and the resulting torque-tracking behavior.

Advances in the field of reinforcement learning (RL)-based drive control allow formulation of holistic optimization goals for the data-driven training phase. The resulting controllers feature efficient drive operation without the necessity of an a priori known plant model but, so far, conduction of the corresponding training phase in real-world drive systems has been applied only sparsely due to safety concerns. This contribution targets the challenging problem of self-learning torque control for a permanent magnet synchronous motor assuming a finite control set (FCS), i.e., the direct selection of switching actions instead of a modulator-based setup. In order to allow a secure and effective online training with real-world drive systems, the RL controller is monitored by a safeguarding algorithm that prevents application of unsafe switching actions, e.g., such that result in overcurrent. The accruing amount of measurement data is handled with the use of an edge computing pipeline to outsource the RL training from the embedded control hardware. The inference of the utilized artificial neural network in hard real time is realized with the use of a reconfigurable FPGA architecture. The resulting RL-based algorithm is able to learn a torque control policy in just ten minutes which has been validated during comprehensive real-world experiments.