With the wide deployment of Industry IoT, massive amount data are collected using smart sensing technologies from the plant floor on a real-time basis. The potential applications of advanced AI and machine learning techniques on the production systems is extremely promising and exciting. However, desultory deployments of AI and machine learning programs in the production system could only lead to very unsatisfactory system performance. The throughout understandings and domain knowledge on manufacturing process and system are the key to the successful applications of machine learning techniques on smart manufacturing systems.
In this research, we combine our system models and domain knowledge with state-of-the-art machine learning techniques to solve the most imperative problems in today's manufacturing industry, which include maintenance decision-making, makespan prediction, production scheduling and energy control etc. The research aims to develop an integrated control framework that deliver optimal control decisions and coordinates different levels in production systems, e.g. machine-level, process-level, and system-level, to achieve higher efficiency and lower costs for the whole system. Specifically, the following works are conducted in this research:
- Establishment of system models for complex-structured production systems and derivation of useful system properties
- Paving the path for knowledge-guided applications of advanced data-driven methods in smart manufacturing systems
- Real-time preventive and corrective maintenance decision making using deep reinforcement learning techniques
- Product makespan prediction and scheduling and in production systems with highly customized products
- Development of coordinated multi-level control framework for smart manufacturing systems
- Enhancing the interpretability of machine learning models used in above areas