膜计算多模态优化算法及在汽油生产调度中的应用研究

The complicated optimization problems encountered in the process of gasoline production and scheduling are with large-scale, discontinuous, combinatorial, nonlinear, and constrained, etc. property. It is of the most important theoretical and applied significance to research the membrane computing based multi mode optimization algorithms and application in gasoline producing and scheduling. Membrane computing (MC) is a class of powerful computing model abstracted from the way that the living cells process chemical compounds, energy and information in their closed membrane structures. Single/multi objective multi mode optimization algorithms based on MC with a self learning system will be presented in this project, which will be fulfilled by possessing a certain multi mode membrane structure, a self learning system and new computational rules abstracted from biological phenomena in living cells and tissues, and introducing some computing mechanism or systems of other computational intelligence algorithms (e.g. EC) and mathematical programming methods. The proposed algorithms will be applied in complicated optimization problems in the process of gasoline production and scheduling. And the practical application in refineries will be verified by simulation experiments. This project will provide some new intelligent optimization theory and methods for solving complex engineering optimization problems, and has a wide range of potential application.

汽油生产调度问题是具有大规模、非连续性、组合性、非线性、约束等特性的复杂优化问题。研究基于膜计算多模态优化算法汽油生产调度问题具有重要的理论和应用价值。膜计算是一类计算能力强大的，模拟活细胞在封闭的生物膜结构中处理化合物、能量和信息的方式的并行分布式理论计算模型。本项目拟以膜计算理论为基础，通过将活细胞及其组织的一些生物学现象抽象为新的计算规则、多模态膜结构及自学习机制，并引入其他智能算法（例如进化计算等）和数学规划方法中的寻优计算策略，提出新的具有自学习机制的单目标、多目标膜计算多模态优化算法，将所提算法用于求解复杂的汽油生产调度问题，并通过仿真实验研究应用中的问题。本项目的研究将为复杂工程优化问题的解决提供新的智能优化理论与方法，具有广阔的应用前景。

针对具有大规模、非线性、组合性、有约束、难以数学公式描述等特性的复杂优化问题，本课题组以膜计算多模态优化算法研究为核心，并研究将算法应用于汽油生产过程中的生产调度问题。此外，课题组对不确定性分析、基于数据建模方法、图像处理等进行了应用研究，迄今为止，课题组发表了9篇科研论文，其中SCI索引论文2篇，另有5篇科研论文在投和待投。.本课题组所做研究包括单目标膜计算多模态优化算法及其在汽油调合、燃料电池模型参数辨识问题中的应用；汽油生产中催化裂化反应再生过程的支持向量回归、神经网络、小波神经网络模型及其最优化；锂电池物理性能、电量管理研究；自适应膜计算优化算法、智能优化算法在桥式吊车、汽车防抱死系统PID控制器参数辨识问题中的应用；量子信息熵连续控制器设计及仿真研究；卡尔曼滤波混合支持向量机算法在目标追踪问题中的应用。.本课题的研究将为复杂问题的建模、优化、控制器设计，提供新的理论和方法，课题的案例研究显示出其重要的理论与应用价值。