节点文献
基于智能优化算法的施工项目风险预测与网络计划优化研究
Study on Risk Prediction and Network Planning Optimization for Construction Item Based on Intelligent Optimization Algorithms
【作者】 李万庆；
【导师】 张世英；
【作者基本信息】 天津大学 ， 管理科学与工程， 2004， 博士
【摘要】 本文在研究和分析影响施工项目管理风险的不确定性因素及不确定性信息处理方法的基础上,对人工神经网络、遗传算法等智能优化算法进行了理论探讨;设计提出了基于遗传算法的人工神经网络算法、实现了两种智能优化算法的优势互补;结合遗传算法和模拟退火算法各自的优点,提出了引入退火机制的混合遗传算法。文中,依据风险因素与风险结果之间的对应关系,通过改进的人工神经网络建立了从风险因素到风险结果之间的映射关系,给施工项目风险预测提供了一种先进的预测方法。在对网络计划优化技术深入研究的基础上,建立了兼顾项目收益最大和质量最优的工期—净收益—质量多目标优化模型。设计了基于混合遗传算法的网络计划多目标优化的实现程序,结合风险预测中得出的工期和费用风险值对多目标优化结果进行修正,使之更符合复杂多变的施工环境。本文将粗糙集理论引入到网络计划优化目标如何实现的问题中,得出了与施工企业实际措施相一致的若干条决策规则,用以指导以后现场施工管理人员进行管理,确保最终目标的实现。从而为施工项目管理实现完全智能化提供了新的思路。
【Abstract】 This thesis gives a theoretical discussion to Artificial Neural Network, GeneticAlgorithms and other intelligent optimization algorithms based on the research andanalysis of the uncertain factors and uncertain information disposition method whichinfluence the construction item’s risk management. Devises and introduces anArtificial Neural Network Algorithms grounded on the Genetic Algorithms to fullyuse the advantages of themselves. Then combines with the advantages anddisadvantages of both Genetic Algorithms and Simulation Anneal algorithms, putsforward the Hybrid Genetic Algorithms using the Simulation Anneal mechanism.Between the line, according to the connection of risk factors and risk result, themapped relax of them is built up by advancing the Artificial Neural Network, thusoffering a advanced prediction method. The multi-objective optimal modelconsidering the maximal pure value and quality of construction project is made badeon a deep study on Network Planning Optimization. Design the program achievementof Network Planning multi-objective optimization based on Hybrid GeneticAlgorithms Considering risk value of times and cost predicted by above improvedANN, this multi-objective optimal value must be corrected in order that the correctedvalue is much more consonant with complicated construction fact than the optimalvalue.The rough sets theory is introduced in the problem of how the Network Planningoptimization can be achieved, a series of decision rules is obtained according with thereal measure of construction enterprise. Thus guiding the construction managementpersons in the spot to manage so as to realize the ultimate object and offering a newmethod for the full intellective construction item management.