化工生产过程基于模型自适应的蒸汽系统实时优化方法研究

Steam system is an important part of the utility system of the chemical process. As the expansion and transformation of the production scale and the process, there are many problems such as large energy consumption, high cost and low energy efficiency in the steam system. Real time optimization (RTO) is an important way to reduce costs, improve production efficiency, and realize energy saving for the industrial enterprises. Accurate and reliable model is the prerequisite for the successful implementation of real-time optimization. But the RTO approaches require dealing with plant-model mismatch, which can lead to infeasibilities as well as to suboptimal solutions.This project addresses the characteristics of chemical process, uses the technology such as intelligence information proceeding, engineering thermodynamics, process modeling and intelligent optimization algorithm, and studies the modeling methods and real time optimization theory which subject to the process and the equipment constraints of the steam system for energy saving and economic benefits. The main contents include energy-using process modeling of the key steam equipment based on process mechanism and operation information, steam system modeling based on mixed integer nonlinear programming model, model adaptation using multi-source information, real time optimization of steam system based on co-evolutionary algorithm and GPU. Furthermore, the developed techniques will be applied in the ethylene plant, and then the real time optimization framework of steam system will be established. Thus the management and operation level of the enterprise’s energy system is improved and energy-saving and cost-reducing can be achieved.

随着化工生产企业规模和工艺流程的扩大和改造，作为其公用工程的重要组成部分的蒸汽系统存在能耗大、成本高、能量利用效率低等问题。实时优化是工业企业降低成本、提高生产效率、实现节能降耗的重要手段。准确可靠的模型是实时优化成功实施的前提，而实际的应用中往往存在模型失配的问题，使得优化问题无解或者收敛到次优解。为此，本项目针对化工生产过程蒸汽系统的特点，综合应用智能信息处理、工程热力学、工业过程模型化和智能优化算法等方法和技术，研究面向节能降耗和企业经济效益的蒸汽系统建模和实时优化方法。主要研究内容包括：融合工艺机理和过程运行信息的关键蒸汽设备用能过程建模；基于混合整数非线性规划模型的蒸汽系统建模；基于多源信息的模型自适应方法；基于协同进化算法和GPU加速的实时优化方法。并通过在乙烯装置蒸汽系统上的工业示范应用，构建完整的蒸汽系统实时优化理论和实践体系，提升企业能量系统管理和运行水平，实现节能降耗。

能源是经济可持续发展和社会稳定的重要支撑，目前全球区域能源短缺现象极为严重。各国政府及企业都认为节能降耗是可持续发展的重中之重，尤其对传统的石油化工行业影响更为深远。工业企业是能源消耗大户，作为其公用工程的重要组成部分的蒸汽系统存在能耗大、成本高、能量利用效率低等问题。利用大数据技术、工业过程模型化、智能优化、智能硬件等技术实现传统工业智能制造升级是工业企业降低成本、提高生产效率、实现节能降耗的重要手段。为此，本项目按照计划任务书开展研究工作，完成的研究内容包括：采用融合工艺机理和过程运行信息的方法建立了关键蒸汽设备如抽凝式透平、减温减压器等的用能过程过程模型；以抽凝式透平抽汽量、电泵/透平泵开备、减温减压器蒸汽流量为优化变量，建立了蒸汽系统混合整数非线性规划模型；基于多源信息，采用递归偏最小二乘建模方法实现了透平做功过程模型的在线更新，保证了优化模型的准确性；采用协同粒子群优化算法实现了蒸汽系统的优化，应用基于偏好的最优历史知识库更新机制，提高了算法的求解效率；利用GPU加速方法，减少了协同粒子群优化算法的求解时间，加速比达到10倍以上，满足了实时优化的要求。项目实施期间，发表了论文11篇，其中SCI、EI检索论文8篇；申请国家发明专利1项，申请登记软件著作权1项；以项目研究内容为研究方向，培养博士研究生1名，培养硕士研究生2名；为企业培养了一批蒸汽系统实时优化的人才。