Optimization of Short-term Scheduling for Crude Oil Operations Based on Deep Reinforcement Learning
Hou Yan
Yang Jiajia
Teng Shaohua
Zhu Qinghua
Abstract：This paper aims to address the problem of suboptimal pipeline transfer rates in short-term crude oil scheduling using a decomposition approach that converts discrete pipeline transfer rates into a continuous range.A novel decision generation method is introduced to avoid direct search in the continuous transfer rate domain,thereby maintaining algorithm performance.A crude oil scheduling method based on the Soft Actor-Critic(SAC)algorithm is proposed,by reasonably designing state features,action space,and reward function.Five objectives are considered including pipeline mixing costs,tank bottom mixing costs,distiller switching tank costs,charging tank usage costs,and energy consumption costs.Case analysis shows that the SAC-based scheduling method improves single-objective optimization by 1.2%to 77.8%compared to existing methods.
Keywords：crude oil short term schedulingdeep reinforcement learningcombination optimizationsoft actor-critic(SAC)
Publication Date：2026-02-28
Online Publishing Date：2026-03-19(First online date of this platform, not the publication date of the document)
Pages：9( 155-163 )