The optimization of production systems processes is driven by the international competition and the manufacturing technological advances. In recent years, more attention has been given to assess the potential advantages of using the innovative technology to elaborate different scheduling solution procedures in order to balance production efficiency and obtain high levels of productivity. Computerized scheduling systems for process operations can be used to achieve high system utilization: to identify and eliminate weaknesses and avoid unscheduled failures and resource waste. In this study, we extend the general scheduling framework to model a specific machine-scheduling problem that characterizes real industrial frameworks with more complex functional demands: precedence-constrained tasks, individual processing times of operations on different machines and communication cost between tasks that are not assigned to the same machine. The specificity of this problem generates an increased complexity and needs higher computational effort compared to a classical task-scheduling problem. The problem is modeled as a Mixed Integer-Linear Programming (MILP) formulation and is solved using a Particle Swarm Optimization approach - a heuristic population-based global optimization method that performs well in difficult multi-objective optimization problems arising in computer science and engineering. The proposed workflow task scheduling and allocation algorithm was tested on several randomly generated test instances. Experiments have shown valid scheduling results: minimum schedule lengths were obtained, while task-precedence constraints are fulfilled. We also noticed good computational experience even when problem size was increased.
Paper
Full text
A particle swarm-based procedure for task allocation in cyber-physical systems
Semantic Scholar · Computer Science · 2023
Abstract
The optimization of production systems processes is driven by the international competition and the manufacturing technological advances. In recent years, more attention has been given to assess the potential advantages of using the innovative technology to elaborate different scheduling solution procedures in order to balance production efficiency and obtain high levels of productivity. Computerized scheduling systems for process operations can be used to achieve high system utilization: to identify and eliminate weaknesses and avoid unscheduled failures and resource waste. In this study, we extend the general scheduling framework to model a specific machine-scheduling problem that characterizes real industrial frameworks with more complex functional demands: precedence-constrained tasks, individual processing times of operations on different machines and communication cost between tasks that are not assigned to the same machine. The specificity of this problem generates an increased complexity and needs higher computational effort compared to a classical task-scheduling problem. The problem is modeled as a Mixed Integer-Linear Programming (MILP) formulation and is solved using a Particle Swarm Optimization approach - a heuristic population-based global optimization method that performs well in difficult multi-objective optimization problems arising in computer science and engineering. The proposed workflow task scheduling and allocation algorithm was tested on several randomly generated test instances. Experiments have shown valid scheduling results: minimum schedule lengths were obtained, while task-precedence constraints are fulfilled. We also noticed good computational experience even when problem size was increased.