【作者】 郭钧；

【导师】 钟登华；

【作者基本信息】 天津大学 ， 管理科学与工程， 2004， 博士

【摘要】 本文进行了天津港口发展综合评价的理论与方法的深入研究,从系统分析的角度,提出港口一体化发展理论观点,依据从指标体系的构建、指标定义、分析、量化、预测、指标评价到综合评价的研究方案,提出港口吞吐量相关分析预测法、指标优化综合评价法,引入综合评价指标体系法、黑箱模型法、人工神经网络法、主成份分析法、层次分析法、模糊综合评价法等系统工程理论和信息分析方法,深化和改进港口发展综合评价的理论与方法,对港口发展提供了有力的决策支持,具有广阔的应用前景。论文主要研究工作和研究成果如下:(1)系统构建港口发展综合评价指标体系,提出港口一体化发展理论观点和指标选择原则,按照概念模型和层次分析原理,建立递阶层次结构模型,进行指标定义分析,探讨指标数据规范化、标准化、归一化方法,建立指标模糊评价标准集,解决了指标体系中定性和定量指标并存条件下的指标量化问题,为港口发展综合评价理论与方法研究提供了理论基础。(2)对港口发展综合评价的重要指标—港口吞吐量进行了深入研究,提出港口吞吐量的相关分析预测法及其改进BP人工神经网络算法。分析了吞吐量的传统预测方法在实用中存在的问题,提出港口吞吐量受指标体系内部另一指标—港口建设规模强烈相关影响,并可据此进行预测分析的观点,结合黑箱模型理论,通过定性研究和定量检验,证明了两种指标之间存在显著的正相关关系,建立天津港口吞吐量的相关分析预测法模型,结合天津港口吞吐量实际数据,提出改进神经网络实现港口吞吐量相关预测的简化算法。从而为港口吞吐量预测提供了一种新的方法,也为研究港口发展综合评价指标间的相互关系提供了新的途径。(3)引入主成份分析法进行港口发展指标评价。在主成份分析基础上进行指标评价,确定关键指标、计算港口发展指标评价指数,使用该方法实现了天津港口发展指标评价,从而为指标评价提供了一种新的手段。(4)提出港口发展指标优化综合评价法,集指标体系、指标定义分析、指标量化、指标评价、指标优化、指标求权、综合评价于一体,建立港口发展指标优化综合评价法模型,提出指标优化综合评价法程序,以此为基础完成天津港发展综合评价。利用该方法进行多指标、多层次、多方法、多目标的港口发展综合评价,优化了指标,降低了指标相关性,简化了评价过程,提高了综合评价的准确性、适应性和实用性,为港口发展综合评价提供了新的途径和方法。更多还原

【Abstract】 Focusing on research about theory and methods of comprehensiveevaluation for Tianjin port development, the paper introduces andpresents comprehensive evaluation factors system method, Principalcomponents analysis method, analytic hierarchy process (AHP) method,fuzzy comprehensive evaluation method, improved neural networks,correlation analysis port throughput prediction method, factors optimizingcomprehensive evaluation method, and etc, to research the theory andmethods on comprehensive evaluation of port development . By the stepsof that, from setting up factors system of comprehensive evaluation forport development, analyzing factors, measuring factors, forecasting keyfactor, evaluating factors, to the studies of comprehensive evaluationmethods. The primary research contents and results are presented asfollows.(1) Setting up factors system of comprehensive evaluation for portdevelopment. After research background and situations about theory andmethods of comprehensive evaluation for Tianjin port development havebeen analyzed, Principles to select factors of comprehensive evaluationfor Tianjin port development are provided, Factors are fixed on,Hierarchy frame chart of factors system of that is drawn according toconception’s model and analytic hierarchy method, The factors areanalyzed including their situations and development trend by the numbers;factors measure ways are studied too including factors specificationmethod, factors standardization method and factors unitized method. Forthe qualitative factors and quantitative factors exist together in the system,and fuzzy evaluation standard collections of factors are presented to solvethe problem on factors measure method. It presents a very good basis andway on comprehensive evaluation of Tianjin port development to set upthe factors system and present the factors measure method.(2) A correlative analysis forecasting method of port throughput andimproved BP neural network arithmetic are presented in the paper.Traditional port forecasting methods are analyzed firstly to be comparedbetween their advantages and disadvantages, their applications, limitedconditions and problems have been studied by case research. By theBlack Box theory, one factor, port construction investment, is fixed on asinput sequence of the Black Box, and port throughput is fixed on asoutput sequence of the Black Box too, by the qualitative analysis andcheckout of general error, t checkout of and F checkout of the model,distinct correlation between port construction investment and portthroughput is confirmed, which change a traditional method to depend onport hinterland direct build method of port throughput cause and resultforecasting . According to the Black Box Model, a model of own adaptiveincrement monadic operator linearity regression is presented to crate aport throughput forecast model of time and cause, by the time sequenceforecast method to predict port investment, and the paper studies how tomake use of amendable nerve net method to carry through the factorscorrelation forecast method of port throughput, then a balance researchbetween that way and a predigesting arithmetic method of factorscorrelation forecast method is finished too. It presents a new way forresearch of prediction of Tianjin port throughput, and a new way to studythe relationship between factors of Port comprehensive evaluation.(3) Factors evaluation for Tianjin port development is finished.Basing on Principal components analysis method to evaluate factors,confirm key factors and calculate index of factors evaluation, Whichpresents a new way for factors evaluation of port development.(4) Factors optimizing comprehensive evaluation method ispresented. The method model is set up. Multi-information analysismethods are combined together, according to demands of comprehensiveevaluation of Tianjin Port, a optimizing evaluation sequence is chosen tobe applied in Tianjin port’s comprehensive evaluation. This is a kind ofmulti-factors, multi-hiberarchy, multi-routes, multi-objects method,which presents a new effective way for comprehensive evaluation ofPort development.更多还原