基于遗传算法的项目决策优化模型研究

发布时间：2018-06-16 13:33

本文选题：遗传算法 + 网络计划优化　；参考：《吉林大学》2013年硕士论文

【摘要】：随着经济的发展，工程项目越来越大型化，复杂化，依靠传统的方法已经很难得到好的网络决策计划。网络计划优化主要包括：工期优化、费用优化、资源优化三个方面。项目管理人员追求的目标即是：安排合理的进度计划以使整个项目所花费的资金最少、工期最短、资源最均衡，这是决定项目获利与否以及获利多少的关键。然而这些目标的优化问题，目标之间通常是相互冲突的，约束条件也是相冲突的，且目标解不唯一，甚至不存在最优解。项目决策优化问题的核心是网络计划技术，但解决这些问题的传统的数学规划等方法存在很多缺陷如：方法针对性太强，，不能广泛应用于各实际问题。在处理工作逻辑关系复杂的问题时效率低，优化效果大打折扣，所以为了更加科学、合理的进行项目决策，本文选用智能化算法遗传算法来研究项目决策优化问题和网络计划优化问题。遗传算法，提供的是求解问题的一种通用框架，具有较好的全局搜索性能，易于并行化。它可以用来有效地解决那些非线性的、不连续的、不可微不可导、多峰、多目标的问题，而且遗传算法本身并不依赖于问题的具体领域，非常适合处理离散优化组合问题，它具有很广泛的可行解的表示，不需要辅助信息，具有群体搜索的特征和内在的启发式随机搜索特征，而且可扩展性高，易于和其他的方法结合使用，具有很高的鲁棒性(Robust)，易于广泛推广使用。该算法在处理大型复杂系统优化问题上已经取得了显著的成果，其所表现出来的独特的优越性和健壮性，是其他方法所无法比拟的。本文基于遗传算法对项目工期费用优化和工期资源优化分别进行了应用研究，基于时间费用的两种常见关系类型即连续型和离散型，分别设计了不同的遗传算法优化方法；对工期固定资源均衡和资源有限工期最短两个问题也分别建立了遗传算法求解模型，并且在最后给出了算法实例进行验证，得出优化结果，得到一系列最优解。本文在模型的构建以及求解算法上的研究为项目管理中的目标优化问题提供了一种新颖的可操作性强的思路与方法。项目管理人员可根据实际情况对优化方案进行比较选择，达到提高经济效益的最终目的，具有极强的现实指导意义。
[Abstract]:With the development of economy, engineering projects are becoming more and more large-scale and complicated. It is difficult to get good network decision plan by traditional methods. Network planning optimization includes three aspects: duration optimization, cost optimization and resource optimization. The goal pursued by the project manager is to arrange a reasonable schedule to make the whole project spend the least amount of money, the shortest duration and the most balanced resources, which is the key to determine whether the project is profitable or not and how much. However, the optimization problems of these objectives are usually conflicting with each other, and the constraints are also conflicting, and the solution of the target is not unique, or even the optimal solution does not exist. The core of the project decision optimization problem is the network planning technology, but the traditional mathematical programming method to solve these problems has many defects, such as: the method is too targeted to be widely used in practical problems. In order to deal with the complex problem of working logic relationship, the efficiency is low, the optimization effect is greatly reduced, so in order to make the project decision more scientifically and reasonably, In this paper, an intelligent genetic algorithm is used to study project decision optimization and network planning optimization. Genetic algorithm, which is a general framework for solving problems, has good global search performance and is easy to be parallelized. It can be used to solve nonlinear, discontinuous, non-differentiable, multi-peak and multi-objective problems, and genetic algorithm itself is not dependent on the specific domain of the problem, so it is very suitable to deal with discrete optimization combinatorial problems. It has a wide range of representations of feasible solutions, does not require auxiliary information, has the characteristics of group search and inherent heuristic random search, and is highly scalable and easy to use in conjunction with other methods. It has high robustness and is easy to be widely used. The algorithm has achieved remarkable results in dealing with large-scale complex system optimization problems, and its unique superiority and robustness can not be compared with other methods. In this paper, based on genetic algorithm, the cost optimization and resource optimization of project are studied, and the two common relationship types of time cost are continuous and discrete. Different genetic algorithm optimization methods are designed respectively, and the solving models of genetic algorithm are established for the two problems of fixed time limit resource balance and resource limited time limit, and an example is given to verify the algorithm. The optimization results are obtained and a series of optimal solutions are obtained. In this paper, the construction of the model and the research of solving algorithm provide a novel and operable method for the goal optimization problem in project management. According to the actual situation, the project manager can compare and select the optimization scheme to achieve the ultimate goal of improving economic benefits, which has a strong practical significance.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TU712

【参考文献】