【作者】 魏智军；

【导师】 马军海；

【作者基本信息】 天津大学 ， 管理科学与工程， 2013， 博士

【摘要】 在我国的电力体制下，电力行业由国家垄断经营，因此，不少电力企业存在生产效率和经营业绩比较低下的问题。为提高电力企业的效益，就需要运用有效的绩效评价方法，找到改善经营管理和提高竞争力的途径。内蒙古电力集团(有限)公司是内蒙古自治区国有独资的大型电网企业，也是作为全国唯一的省属独立电网公司，因此，要根据它所具有的自身特点，采取与其它电网公司有所区别的方法，对其进行效率评价。首先，根据NMG电力集团独特的定位特点，提出一套新的效率评价指标体系。该指标体系包括5个输入指标和1个输出指标。再根据集团公司下属的9个地方电业局2011年的数据，采用伽玛概率分布函数法将其标准化。为了更加科学全面地评价各地方电业局运行情况，不仅要考虑电业局的当前现状，还需要兼顾以后的发展潜力。因此，用各局的6个评价指标的历史(1987年-2011年)数据和拟合多元状态空间模型和向量ARMA模型方法，预测出各指标2015年的未来值。本文选取了一种简单、直观、改进的全排列多边形图示指标方法，分别以6个效率评价指标2011年的现值和2015年的预测值，计算了9家电业局的排名。首先计算各指标之间秩相关系数和互信息指数，在三角形面积计算中还考虑了其边长的相关性。结果表明WH电业局和鄂尔多斯电业局的运营效率要好于其它电业局，同时，改进方法计算出的效率要高于原始方法。为了提高效率评价的精度，进一步将其它领域的先进函数估计方法引入到效率评价中。这就需要建立新的评价样本。首先将各指标划分为5个等级，用最优设计法选取30个样本进行输入，而样本输出的效率评价值则通过Delphi方法确定。再采用规则集成、随机森林、随机梯度Boosting、支持向量、人工神经网络、自适应样条和线性回归6种机器学习方法得到效率评价结果。最后，运用组合预测的思想，对6种方法的效率评价值，进行算术平均，最终给出9家地方电业局的排名。结果表明，EEDS电业局和WH电业局相比于其它电业局更具效率。因此，可通过对这2家电业局的优势和特点的总结，并尽快向整个集团推广，从而提高整个集团的效益。在分析电力金融市场风险的种类、特点基础上，以灰色系统、多元线性回归、VaR和马尔科夫链等方法度量各种风险，提出风险管理流程和办法。并以SWOT模型和层次分析法分析NMG风能发电现状。更多还原

【Abstract】 In our country electric power industry is run by state monopoly. As a result, manyelectric power enterprises have poor production efciency and operating performance.In order to improve the benefits of the electric power enterprises, efective performanceevaluation methods are taking into accounting. NMG Power Group Co.,Ltd, a whollystate-owned power grid company in Inner Mongolia autonomous region, is the onlyprovincial State grid corporation alone. Efciency evaluation must be diferent withother power grid companies according to its distinct characteristics.First of all, a new set of efciency evaluation indexes was put forward on theground of the unique positioning of NMG Power Group Co.,Ltd. The indexes systemincluded five input and one output indicators. Based on2011datum of nine groupcompany subordinates, gamma cumulative probability distribution function was usedto standardization. In order to be more scientific, development potential would be paidattention to as the same as the current situation. With history datum of six indicatorsfrom1987to2011, Multiple State Space and Vector ARMA models were fitted andpredicted values of2015were calculated.Full Permutation Polygon Synthesis Index, an intuitive method and its improvedformula were chosen. Indicators of2011and2015were separately applied to orderefciencies of nine subordinates. Rank correlation coefcients and mutual informationcoefcients were evaluated and edge correlations were considered to compute trianglearea. Results showed that the operating efciency of WH power and EEDS’s werebetter than others. The efciencies of the improved method were higher than ones ofthe original method.To improve the accuracy of efciency evaluation, the advanced function estimationmethods in other fields were quoted and samples needed to be established. Each indexwas divided into five levels and30input samples were chosen by Optimal Design. Theoutput efciencies of them were evaluated through Delphi method. Rule Ensemble, Sixalgorithms, Random Forest, Stochastic Gradient Boosting, Support Vector, ArtificialNeural Network, Multivariate Adaptive Spines Regression and Linear Regression, wereapplied to obtain efciencies. The average of six values was used as final ranks in thelight of combination forecast. It was showed that EEDS power and WH power had more efciency than others. These two bureaus should be paid more attention to analyze andpopularize.Based on the analysis of the type and characteristics of risk in electric power mar-ket, Grey System, Multiple Linear Regression, VaR and Markov Chain were employedto measure all kinds of risk. Risk management processes were put forward. The statusof wind power in NMG were analyzed by SWOT and Analytic Hierarchy Process.更多还原