Intelligent Control Suite 03

Improving EMS efficiency with Deep Reinforcement Learning

  • Goal

    To develop an energy-efficient control system with various control parameters in vehicle EMS(Energy Management System) that controls temperatures of multiple target areas

  • Our Approach

    Compared to conventional PID controls, a more energy-efficient control algorithm is accomplished using MakinaRocks’ Intelligent Control solution. The solution provides application-specific algorithms based on the state-of-the-art Reinforcement Learning (RL), which can be trained on data collected from physics-based simulators. The trained RL model will be further tuned with sim2real technology.

  • Results

    Expected to save energy up to 20%, compared to existing rule-based control algorithms