Differential Evolution and Hybrid Petri-Net Based Optimization Approach for a Gas Compressor Station Start-up

カテゴリ: 研究会(論文単位)

論文No: IS16007

グループ名: 【C】電子・情報・システム部門 情報システム研究会

発行日: 2016/05/17

タイトル(英語): Differential Evolution and Hybrid Petri-Net Based Optimization Approach for a Gas Compressor Station Start-up

著者名: Malamura Evgeny(Waseda University),Murata Tomohiro(Waseda University)

著者名(英語): Evgeny Malamura(Waseda University),Tomohiro Murata(Waseda University)

キーワード: differential evolution|petri net|hybrid systems optimization|gas compressors|pipelines|differential evolution|petri net|hybrid systems optimization|gas compressors|pipelines

要約(日本語): In addition to the normal steady-state operation, the cycle of many production processes includes large transition modes such as start-up, shut-down, or product changeover which occur occasionally. Although transition modes often consume a large amount of time and resources, thereby the optimization of such transition procedures is highly desirable. The optimization of large transitions is a complicated task since it often requires the consideration of both, continuous and discrete controls such as discretely switched pumps and valves. Such processes are often modeled as hybrid systems with interacting continuous and discrete dynamics.In this paper we present an integrated Differential Evolution and Hybrid Petri Net based approach for the optimization of a hybrid system start up. In particular the problem of a start-up of a natural gas compressor station with parallel pumps is addressed in the case-study.The approach uses a heuristic optimization algorithm - Differential Evolution (DE) for computing optimal values of the continuous input parameters (output pressures and flow through compressors of the station), while the cost function is evaluated by embedded Hybrid Petri Net (HPN) simulation of the hybrid process. Evolutionary optimization technique is used here because, in contrast to other optimization schemes, DE is not affected by discontinuities of the cost function which may arise from the autonomous switching of the embedded hybrid model. The HPN model includes a continuous part, corresponding to continuous gas flow in pipes and a discrete part that simulates transitions between distinct operational states of a compressor station (e.g. number of compressors working) and pipes (in/out of service).The main advantage of the proposed approach is a reduction of complexity in the optimization problem formulation for a system with non-linear constraints and hybrid discrete-continuous parameters.

要約(英語): In addition to the normal steady-state operation, the cycle of many production processes includes large transition modes such as start-up, shut-down, or product changeover which occur occasionally. Although transition modes often consume a large amount of time and resources, thereby the optimization of such transition procedures is highly desirable. The optimization of large transitions is a complicated task since it often requires the consideration of both, continuous and discrete controls such as discretely switched pumps and valves. Such processes are often modeled as hybrid systems with interacting continuous and discrete dynamics.In this paper we present an integrated Differential Evolution and Hybrid Petri Net based approach for the optimization of a hybrid system start up. In particular the problem of a start-up of a natural gas compressor station with parallel pumps is addressed in the case-study.The approach uses a heuristic optimization algorithm - Differential Evolution (DE) for computing optimal values of the continuous input parameters (output pressures and flow through compressors of the station), while the cost function is evaluated by embedded Hybrid Petri Net (HPN) simulation of the hybrid process. Evolutionary optimization technique is used here because, in contrast to other optimization schemes, DE is not affected by discontinuities of the cost function which may arise from the autonomous switching of the embedded hybrid model. The HPN model includes a continuous part, corresponding to continuous gas flow in pipes and a discrete part that simulates transitions between distinct operational states of a compressor station (e.g. number of compressors working) and pipes (in/out of service).The main advantage of the proposed approach is a reduction of complexity in the optimization problem formulation for a system with non-linear constraints and hybrid discrete-continuous parameters.

原稿種別: 英語

PDFファイルサイズ: 1,069 Kバイト