Answers are grounded in this paper and cite their sources.
As contemporary industrial enterprises and manufacturing facilities expand in scale, the judicious scheduling of numerous in-house equipment clusters has become a critical factor in enhancing operational efficiency and reducing production costs. Traditional static algorithms merely sustain the baseline functionality of factory apparatuses, thereby failing to unlock the latent synergy among integrated equipment. In response to the advent of artificial intelligence and the maturation of machine learning algorithms, innovative solutions have emerged to address such operational scheduling challenges. This paper presents an algorithm predicated on a deep reinforcement learning-based optimization strategy, integrated with the localized time-of-use electricity tariff framework, to diminish the operating expenses of dryer clusters within the facility. Initially, an objective function aimed at minimizing total operating costs was constructed, alongside the formulation of constraint conditions derived from the operational protocols of the dryer clusters. Subsequently, under a reinforcement learning paradigm, the dryer clusters were modeled by delineating the state and action spaces and formulating a reward function contingent upon the tariff schedule and the degree of completion of gas-drying tasks. Ultimately, a neural network employing a double-DQN architecture was developed and utilized to resolve the scheduling strategy through model training. Empirical results from multiple training iterations with dryer clusters validated that the proposed algorithm can reduce electricity expenditures without necessitating alterations to the existing equipment, while also providing recommendations for the optimal number of additional standby dryers to further reduce operational costs.
No takes yet. Share an insight, caveat, or question.
Feng et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: