【作者】 谢颖；

【导师】 黄文杰；

【作者基本信息】 华北电力大学（北京） ， 技术经济及管理， 2008， 博士

【摘要】 多年来,政府投资项目为我国的经济发展做出了巨大的贡献,国家每年都要投资数千亿元进行建设。但政府投资项目普遍存在着“三超”、“主体缺位”及腐败等顽疾。如何提高政府投资项目的投资效益和建设效率已经成为近年来倍受关注的问题。依照当前投资体制改革的精神,在非经营性政府投资项目中要推行“代建制”。在代建项目中,将政府对项目的行业管理职能与项目管理职能分离,将政府类项目法人对建设项目的所有权与具体的工程管理权分离,从而提高政府投资项目的建设效率,实现政府投资效益。要达到以上的目的,在推行代建制的政府投资项目中,政府投资人的工作重点须发生重大的转移,政府投资人的管理工作须重新进行准确的定位。本论文以政府投资人对代建项目的管理为研究对象,从政府投资人的角度研究代建项目管理的相关理论、方法及有关问题。论文分别就项目管理公司的信用分析、代建项目的政府采购活动、代建项目中政府投资的风险分析、代建项目投资的智能化预测、代建项目建设市场的公共规制,以及代建人的项目管理绩效评价等问题进行了研究。对代建项目中政府投资人与代建人的权利、义务以及各自的管理工作进行界定,并制定了代建项目实施过程的基本工作流程。建立了基于数据挖掘技术和支持向量机技术的项目管理公司信用评价混合模型。通过主成分分析有效地实现了降维。通过ICA实现高阶去相关,分离出信用属性“源”。提出了代建制中履约保函、职业责任保险以及履行保证保险等信用风险抵减技术。分析了项目管理招标中的委托人选择机制及其不完全信息的博弈特征。建立了项目管理招标人之间的报价博弈模型,以及项目管理招标人与投标人的混合策略模型,得出了不同背景下博弈人的最优策略。建立基于模糊聚类和灰色关联分析的项目管理招标综合评标模型。针对代建项目投资的非高斯性变化特征,分析了蒙特卡洛模拟技术的不足,分别建立了基于马尔科夫链、聚类分析以及多属性渐进算法的建设投资风险评估模型。建立基于高阶去相关的代建项目建设投资属性重构模型,引入独立分量分析实现建设投资属性之间的高阶去相关和去冗余;建立基于ICA和ANN技术的建设投资智能化预测模型。对于代建项目的非线性建设投资回归问题,在小样本条件下,建立最小二乘支持向量机的建设投资的预测模型。分析代建项目建设市场存在的市场失灵现象,以及政府内生性、外生性和体制性失灵问题。探讨了代建项目的公共规制内容、手段。对代建项目的委托代理特征进行分析,建立了代建项目委托代理合同的监督模型、工程承包合同的监督模型,以及项目管理公司与承包商的合谋与防范模型。构建了代建人的项目管理绩效评价指标体系,建立了基于模糊识别的项目管理绩效综合评价模型,引入权广义距离之和构造隶属函数。更多还原

【Abstract】 The government investment projects have important economic and social repercussions in China. The substantive capital was invested in governmental investment projects by country, such as infrastructure and other public works, amounting to numerous billions every year. However, it is general that the performance of some government projects is unsatisfied because of“body devoid”, cost-over-run, poor quality and peculation, and so on.It has attracted worldwide attention that how to improve the investment returns and construction efficiency of government projects in recent years. According to the spirit of investment system reform, construction-agent system should be proposed in non-commercial government investment projects. In construction-agent system project, the efficiency of the project construction is improved and investment returns is achieved by separating the function of industry management from project management of government, separating the ownership from the management right of projects.To realize these objectives, the focus of government should be transferred and the management should be re-positioned accurately during the implementing the construction-agent system. The objective of this article is to estabilish a series of management methods for government investors in construction-agent projects. Theory, methods and related issues were studied from the point of government. Credit analysis of project management enterprises, government purchasing of construction-agent projects, risk analysis of government investment, intelligent forecasting of construction-agent projects investment, public regulation of construction-agent market and project performance evaluation of construction-agent manager were studied in this article.The right, obligation, the management of government and agent were defined and the primary flow of the construction-agent projects was made.The credit evaluation hybrid model based on the data mining technique and SVMs was established. The dimension was descended effectively by the way of PCA. Moreover, the sources of the credit attributes were acquired based on ICA. What is more, the credit risk mitigation methods, such as performance guarantee, professional liability insurance, guarantee insurance and so on, were proposed.The choice mechanism of trustor and the game of incomplete information in the bidding process of project management were analyzed. The bids game model of tenders, the hybrid strategy model of tender and the bidder were set up and the best strategy of bidders who have different background was found. The comprehensive evaluation model of project management bids was proposed based on the fuzzy clustering and grey association analysis,The limitation of Monte Carlo simulation technology was pointed out on the basis of researching on the construction-agent project’s non-gauss variation features, and the construction investment risk evaluation models were established respectively based on Markov theory, clustering analysis and gradient algorithm.The reconstruction model of construction-agent projects based on the higher decorrelation is established, higher decorrelation and data redundancy among construction investment attributes were realized by introducing independent component. The intelligent forecasting model of construction investment is established based on ICA and ANN technology. The forecasting model of construction investment was established for the nonlinear construction-agent projects in the condition of small sample.The analysis for phenomenon of market failure, government endogenousness, exogenousness and system failure in the market of construction-agent projects were analyzed. The contents and methods of construction-agent projects also were discussed. The character analysis of agency about construction-agent projects was made, the supervision models of agency contracts and engineering contracts about construction-agent projects, the collusion and prevention model between management companies of projects and contractors were established.The index system of agents’project management performance was established, a comprehensive evaluating model was set up based on fuzzy recognition theory.更多还原