Co-Scheduling of Energy and Production in Discrete Manufacturing Considering Decision-Dependent Uncertainties

TL;DR

This paper introduces a new co-scheduling model for energy and production in discrete manufacturing that accounts for decision-dependent uncertainties, offering a solution suitable for real-time control and cost reduction.

Contribution

It presents a novel energy-production co-scheduling model incorporating decision-dependent uncertainties and a linearization method, along with a C&CG-based algorithm for efficient solving.

Findings

Significantly reduces production costs in real-world tests.

Achieves better frequency regulation performance.

Provides a convergent algorithm with manageable complexity.

Abstract

Modern discrete manufacturing requires real-time energy and production co-scheduling to reduce business costs. In discrete manufacturing, production lines and equipment are complex and numerous, which introduces significant uncertainty during the production process. Among these uncertainties, decision-dependent uncertainties (DDUs) pose additional challenges in finding optimal production strategies, as the signature or the shape of the uncertainty set cannot be determined before solving the model. However, existing research does not account for decision-dependent uncertainties (DDUs) present in discrete manufacturing; moreover, current algorithms for solving models with DDUs suffer from high computational complexity, making them unsuitable for the real-time control requirements of modern industry. To this end, we proposed an energy-production co-scheduling model for discrete manufacturing, taking into account decision-dependent uncertainties (DDUs). Subsequently, we devised a method for transforming the constraints associated with DDUs into linear form ones. Finally, we designed a novel algorithm based on the column-and-constraint generation (C&CG) algorithm and undertook a theoretical analysis of its performance of convergence and algorithmic complexity. A real-world engine assembly line was used to test our model and algorithm. Simulation results demonstrate that our method significantly reduces production costs and achieves better frequency regulation performance.