Improving a Hybrid Algorithm for APCS Hierarchical Architecture Design via Hyperparameter Optimization

Abstract

Automated process control systems, widely used in industry, have a hierarchical multi-level tree-based architecture and are built using specialized hardware and software. Constructing such a control system from off-the-shelf industrial components with predetermined characteristics is a complex combinatorial optimization problem with multiple constraints. A hybrid method for constructing a tree-based structure is considered, combining a deterministic construction algorithm and an ant colony metaheuristic for device selection. The effectiveness of the metaheuristic algorithm depends on its hyperparameters, which are proposed to be optimized. This paper presents a general problem formulation, defines optimization criteria, describes the solution method, and provides several numerical examples.