Proximal policy optimization-based speed regulation for low-power marine electric motors: modeling, control design, and experimental validation

Abstract

Nowadays, a lot of low-power marine electric motors are utilized in ship propulsion systems; these motors must meet three criteria: energy-saving, precise speed control, and steady operation. To address the adaptability and disturbancerejection flaws of the former, a Data-Driven Control Strategy based on Proximal Policy Optimization (PPO) is presented. Create a dynamic state-space model for the marine motor that incorporates the required mechanical and electrical properties together with realistic load variations. Stable velocity control and low power consumption are the goals of the PPO control system's reward function. The performance of a PPO controller and a conventional PID approach under different reference-tracking and disturbance situations will be experimentally validated using a MATLAB/Simulink simulation platform. The aforementioned findings demonstrate that the PPO approach has enhanced load-disturbance rejection and decreased the system's rise time, overshoot, and steady-state error. The aforementioned quantitative investigations verify that the PPO controller outperforms conventional techniques in terms of both control efficiency and transient and steady-state response characteristics. The aforementioned findings demonstrate that reinforcement learning can be used to dynamically modify the speed of marine motors, resulting in the realization of a more reliable and effective marine propulsion system.