An Optimization Approach for Automata Strategies in Industrial Simulation
Yuhan Zhang, Zhendong Chen, Maolin Pan, Yang Yu · 2024
In industrial contexts, managing time and resources typically involves complex systems with various operational components. This paper introduces a state machine model, optimized for simulating multi-component collaborative processes in industry, which is based on the State Chart XML (SCXML) standard and referred to as SCXML automata. The execution of these automata is treated as Markov decision processes, providing a framework for optimization. The primary goal is to develop strategies ensuring safe operation of such automata while adhering to time and cost constraints. A detailed representation of these automata is presented, including grammar, semantics, and strategies, and methods for strategy generation are adapted from single to multi-component systems. The approach involves the analysis of simulation data from such automata using adaptive reinforcement learning techniques, leading to the generation of strategies that approach optimality. Results demonstrate the capability of this approach in finding near-optimal strategies within a specified number of iterations. This method holds significant research value in the field of industrial flexible production lines, offering a means to generate strategies for multi-layered automata, thereby enhancing process efficiency.