基于数据的油藏生产过程设定点优化模型及算法研究

Optimization and control the oil reservoir production process have a wide application prospect, it can produces a great economic benefit. However, at present, optimization and control methods for oil reservoir production process are mostly based on the large scale steady-state model. Previous work do not provide any optimization procedure that systematically takes into account the interactions of an integrated oil and water production system and simultaneously optimizes the oil produced and water injected rates. In petroleum fields, injection and production rates are the most abundant data. In this research, an optimization problem for oil reservoir production process is formulated, where water injection rates and oil production rates are optimized simultaneously to maximize the future economic return of the reservoir asset. At present, there is no reliable guidance for how to develop the appropriate intelligent optimization algorithm and choose the suitable constraint-handling technique when solving a particular set point optimization problem. The performance of algorithm mainly depends on two components. One is the constraint-handling technique, and the other is the intelligent optimization algorithm. Motivated by these observations, this project aims to study intelligent optimization algorithms using diversified design in order to improve their global search capabilities. To accelerate the convergence speed of the algorithms, the selection and design of individual adaptive criteria will be guided by the domain knowledge inherent to the issue as well as the information obtained from evolutionary feedback. We will focus on the research of intelligent optimization algorithm-based constrained optimization method for oil reservoir set point, the method will provide a more practical and effective technology for set point optimization of oil reservoir production process and the optimization and control of general complex production process.

油藏生产过程优化控制理论与方法蕴藏着极大的经济效益且具有广阔的应用前景。目前对油藏生产过程的优化与控制大都基于大规模稳态模型，而对生产过程中注水率和产出率的设定点优化研究得较少。基于油藏生产过程中积累的注水率和产出率历史数据，本项目拟建立以注水率和产出率为决策变量的油藏生产过程设定点优化数学模型。目前，在利用智能优化算法求解设定点优化模型时，设计高效的智能优化算法和选择合适的约束处理技术没有可靠的方法和理论依据，且算法性能由两部分共同决定。本项目拟研究基于多样性设计的智能优化算法，以提高算法的全局搜索能力；根据问题自身领域知识和进化反馈信息指导选择和设计自适应个体比较与选择准则，以加快算法的收敛速度。研究基于智能优化算法的油藏生产过程设定点优化方法期望为油藏生产过程优化乃至更一般的复杂生产过程优化与控制提供实用有效的技术手段。

油藏注水生产过程建模与优化方法蕴藏着极大的经济效益且具有广阔的应用前景。本项目针对油藏注水生产过程优化建模、设计智能优化算法求解优化模型展开研究，获得了一系列的研究成果，包括：基于历史数据，建立以注水率和产出率为决策变量的油藏生产过程设定点优化模型；以“约束优化智能算法 = 约束处理技术 + 智能算法”的研究框架，提出一种双层迭代通用框架，即在外层迭代中，首先利用修改拉格朗日乘子法将原约束优化问题转换为含界约束的优化子问题，修改惩罚因子和乘子罚参数以及判断算法是否收敛，在内层迭代中，利用改进人工蜂群算法和改进灰狼优化算法求解转换后的子问题；结合改进粒子群优化算法和自适应约束处理技术，提出一种混合方法用于解决约束优化问题；在智能算法设计方面，分析了灰狼优化算法的勘探和开采能力与控制参数a和位置更新方程有关，提出了几种改进的灰狼优化算法；为了避免灰狼优化算法、鲸鱼优化算法和正弦余弦算法陷入局部最优，提出了基于光学透镜成像反向学习策略、折射反向学习策略和随机反向学习策略；在正弦余弦算法中引入自适应惯性权重以协调算法的全局搜索和局部搜索能力。以上研究成果，为油藏生产过程建模、优化与控制提供了一套完整的理论和方法，对一般复杂生产过程优化和控制具有重要的理论意义和应用价值。