Towards Reliable Deep Reinforcement Learning for Industrial Applications: A DDPG-based Algorithm with Improved Performance

Articles in Press

Document Type : Research Article

Authors

Department of Electrical Engineering, Faculty of Engineering, University of Isfahan, Isfahan, Iran

Abstract

This paper proposes Improved Model-Based Deep Deterministic Policy Gradient (IMB-DDPG), a novel reinforcement learning algorithm designed to overcome three critical challenges in industrial DRL applications: (1) poor sample efficiency requiring excessive real-world trials, (2) safety risks from unstable policies during training, and (3) difficulty scaling to high-dimensional continuous control spaces. Building on DDPG's strengths for continuous control, IMB-DDPG introduces four key innovations: (i) a Virtual Environment (VE) for data-efficient learning, (ii) a Simulation Rate (SR) mechanism adapting model reliance dynamically, (iii) a Simulated Experience Buffer (SEB) preventing divergence, and (iv) a Performance Threshold (PT) for fail-safe operation. Evaluated on Cart-Pole benchmark via OpenAI Gym python library, IMB-DDPG demonstrates faster convergence than standard DDPG while maintaining performance degradation under sensor malfunctions or communication losses. These improvements derive from the algorithm's unique ability to simultaneously leverage real-world data and model-generated experiences, reducing physical trial costs while ensuring operational safety. The results establish IMB-DDPG as a practical solution for industrial control systems where reliability and data efficiency are paramount, particularly in applications like chemical process control and precision robotics that demand stable operation amid sensor/communication failures.

Keywords

Main Subjects

AUT Journal of Mechanical Engineering

Articles in Press, Accepted Manuscript
Available Online from 18 September 2025

Files

Share

How to cite

Statistics