Integrated methodology for developing forecasting models for MPC/NMPC in the architecture of cyber-physical manufacturing systems

Abstract

The article proposes an integrated and reproducible methodology for developing mathematical models of technological objects, with a focus on their application to cyber-physical systems, industrial digital twins, and model predictive control (MPC/NMPC) algorithms in digital manufacturing environments. Unlike traditional approaches that treat modelling as an autonomous identification task, the proposed approach formally links control objectives, technological constraints, uncertainty structure, and computational feasibility requirements within a single predictive loop. Particular attention is paid to the structural and parametric optimization of the model from the standpoint of numerical stability and real-time operation.