Machine learning application in batch scheduling for multi-product pipelines: A review
Renfu Tu, Hao Zhang, Bin Xu, Xiao‐Ying Huang, Yiyuan Che, Jian Du, Chang Wang, Rui Qiu, Yongtu Liang
Abstract
Batch scheduling is a crucial aspect of pipeline enterprise operation management, especially in the context of market-oriented operation. It involves three main tasks: quickly preparing batch plans, accurately tracking interface movement, and monitoring operations in real time. Traditionally, multi-product pipeline batch scheduling is mainly performed by building simulation models or optimization models, which are further solved by corresponding conventional mathematical methods. This traditional approach becomes inefficient when applied to large-scale systems. The rapid development of machine learning has brought about a significant transformation in batch scheduling research. This paper first reviews the current state of batch scheduling technology, and suggests that applying machine learning to it is a promising direction for future development. Then, it summarizes the progress of machine learning applications in batch planning, interface movement tracking, and operational condition monitoring, while pointing out their limitations. Finally, in response to the separation of the production, transportation, and sales processes of refined oil, this paper offers application recommendations for five points: oil supply or demand prediction and pipeline state prediction, batch planning, batch interface movement tracking, mixed oil development monitoring, and pipeline operation condition identification, to provide guidance for better application.