【作者】 王小雅；

【导师】 闫庆友；

【作者基本信息】 华北电力大学 ， 技术经济及管理， 2014， 博士

【摘要】 2008年以来,在经济危机及世界经济的多重影响下,电力需求增速快速回落,使电网公司的资产效益受到严重的影响,迫使电网公司变革资产管理方式。资产全寿命周期管理是电网公司转变管理方式、提升管理水平、提高营运效率、提升资产质量、延长设备使用寿命、优化资产成本效益的重要手段。另外,资产全寿命周期管理也是智能电网发展的必然选择。但开展资产全寿命周期管理是一项系统工程,实现总体目标将面临很多困难和挑战,其中之一就是缺乏风险意识。对外部影响环境的研究不够,处于“被动”决策的情况较多,因此,正确认识和把握外部环境的变化趋势,提高应对外部环境变化的能力,是开展资产全寿命周期管理的必然要求。面对动态和不确定的复杂的外部环境的影响,电网公司应该如何把握其外部环境的变化趋势,并采取相应有效的应对措施,使其快速适应外部环境的变化；或者采取某种方式对其外部环境做出相应的调适,使其向有利于资产全寿命周期管理目标实现的方向发展。本文构建了电网资产全寿命周期管理外部环境预警模型,旨在解决“如何把握外部环境的变化趋势”这一问题。论文中涉及的主要内容如下：本文首先介绍了资产全寿命周期管理理念、整体框架模型、评价指标体系及资产全寿命周期管理的特点；接着对外部环境的概念、特点、一般研究方法与调适方法进行了概述。在此基础上,提出了电网资产全寿命周期管理外部环境的概念和主要分类,引出研究电网资产全寿命周期管理外部环境的必要性和构建合适的外部环境预警模型的必要性。在了解电网资产全寿命周期管理模型及评价指标的基础上,提出了影响资产全寿命周期管理的关键外部环境因素的识别流程,设计了电网资产全寿命周期管理外部环境关键影响因素调查问卷；同时构建了基于传递熵的TOPSIS外部环境影响识别模型,编写了相应的Matlab工具箱,对影响资产全寿命周期管理的关键外部环境因素进行识别；并分别对规划设计、采购建设、运行检修、退役处置阶段的关键外部环境的影响机制进行了深入分析。识别出电网资产全寿命周期管理关键外部环境以后,采用客观指标替代法这一环境度量方法将定性的外部环境因素转化为与其相关的定量指标,得到了替代指标体系；接着,对替代指标体系与电网资产全寿命周期管理三个目标(安全、效能、成本)评价指标之间的相关性进行了分析,剔除掉不相关的指标。在指标筛选的基础上,构建了电网资产全寿命周期管理外部环境预警指标体系。在预警指标体系的基础上,分别构建了基于主成分分析和预警指数的外部环境预警模型和基于PSO-BPNN算法的外部环境模型,但由于主成分模型不能将外部环境与资产全寿命周期管理目标联系起来,而PSO-BPNN算法的粒子群维度太大,容易受样本数据的影响。因此,本文将两种预警模型结合在一起,构建了基于PCA-PSO-BPNN算法的外部环境预警模型,将外部环境因素与资产全寿命周期管理目标SEC联系在一起,通过BPNN算法的自学习机制对外部环境的影响程度进行预警,以便决策者能够及时捕捉到外部环境变化的信息以及其所带来的风险和机会。采用Matlab软件和2008年至2013年的季度数据对上述外部环境预警模型进行了仿真模拟,通过不同模型预测结果及其误差的比较分析,验证了该模型的有效性,并利用训练好的模型对2013年第三季度的外部环境进行了预警,结果表明当前的外部环境处于轻警状态,应采取相应的措施规避其不利影响。分别针对规划设计、采购建设、运行检修、退役处置阶段所涉及的各个外部环境因素提出相应的规避风险和利用机会的应对策略,以期对资产全寿命周期管理工作有所帮助。最后对本文的不足与下一步的工作方向进行了总结。更多还原

【Abstract】 Since2008, the electricity demand growth fell back quickly under the multiple influences of economic crisis and world economy. It made the asset profit of the Grid Corporation been seriously affected. Therefore, the asset management mode of Grid Corporation had to been changed. Life Cycle Asset Management (LCAM) is an important means to change the asset management mode, enhance the management level, improve operational efficiency and asset quality, prolong equipment life, and optimize asset cost-effectiveness. Furthermore, LCAM is also the inevitable choice to the Smart Grid development. But it is a systematic project to carry out LCAM. It will be faced with many difficulties and challenges to achieve the overall goal of LCAM. One is the lack of risk awareness. In other words, it is inadequate to study the external environment. And there were many passive decision-making situations. For the reason, it is an inevitable requirement to carry out the LCAM to correctly understand and grasp the development trend of the external environment and improve the ability to respond to the external environment.Facing with the complex dynamic and uncertain external environment, what ways should be to grasp the changing trends of external environment and take appropriate and effective measures to quickly adapt to the change of external environment with the Grid Corporate. Possibly, which means should be to adjust and adapt the changes of external environment and make the external environment conducive to achieve the goals of LCAM. This paper builds an early-warning model with the external environment of LCAM to be designed to address the problem that how to grasp the trend of the external environment. The followings are the main elements involved in the thesis.This thesis firstly described the concept, the overall model framework, the evaluation index system and features of LCAM. Then it outlined the concept, characteristics, the general research methods and response strategies of external environments. On this basis, the concept and main classifications of the external environments with LCAM were proposed. The necessity was proposed to research the external environment of LCAM and to build a suitable model for the external environment early-warning.On the basis of understanding the overall model and evaluation indicators of LCAM, it was proposed the identification process of the key external environments with LCAM. The questionnaire with the external environments of LCAM was designed. Simultaneously, the external environments identification model based on the transfer entropy and TOPSIS was constructed. The Matlab toolbox corresponding with the identification model was built to identify the key external environments Influence LCAM. The external environment Influence mechanisms were deeply analyzed separately in Planning and Design Phase, in Procurement and Construction Phase, in Operation and Maintenance Phase, in Retried and Disposal Phase.After the key external environments had been identified, the substitute means by objective indicators was used to transfer the qualitative external environments into the quantitative indicators. The external environment alternative index system of LCAM was received. And it contained22categories external environments and56indicators. Then, the correlation analysis between alternative indicators and management objectives was conducted to select the lager correlation indicators. And all the lager correlation indicators were constituted the external environment early-warning index system of LCAM.After the early-warning index system was presented, the early-warning models based on the principle component analysis and based on the PSO-BPNN were separately constructed. But the model based on the principle component analysis (PCA) can’t make the external environment link with the management goal of LCAM. And the particle dimensions of PSO-BPNN algorithm were too numerous. Meanwhile, the accuracy of PSO-BPNN algorithm was too vulnerable to be affected by the sample data. Therefore, the thesis presented the external environment early-warning model based on PCA-PSO-BPNN algorithm by integrated the above two models. The new model can link the external environment and the management goal indicator of SEC (Security-Efficiency-Cost). And the new model complied the early warning with the external environment by the self-learning mechanism of the BPNN algorithm. It can help the decision-makers timely capture the changing information of the external environment and the risks and opportunities from the external environments changes.The above external environment early-warning models were simulated by the Matlab software and the quarter data from2008to2013. The validity of the PCA-PSO-BPNN model was verified by comparing the predictions and errors of the above models. On the basis, the external environment status in the third quarter of2013was predicted by the trained model. The result showed that the current external environment be classified in the slight alert status. And the appropriate measures should be taken to avoid its adverse effect.Respectively, the corresponding strategies to avoid risks and take advantage of opportunities was presented separately in Planning and Design Phase, in Procurement and Construction Phase, in Operation and Maintenance Phase, in Retried and Disposal Phase. The purpose was to help to make the decision-makers to achieve to the management goal of LCAM.Finally, the shortcomings and the future work direction of this thesis were summarized.更多还原