【作者】 王鹏；

【导师】 高欣； 高迎欣；

【作者基本信息】 华北电力大学(北京) ， 应用统计硕士（专业学位）， 2019， 硕士

【副题名】以模糊粒化支持向量机法在大唐电力的应用为例

【摘要】 电力工业是中国国民经济的支柱产业之一。随着经济形势的不断变化,中国电力工业目前正处于调整期,增长速度正在放缓。面对新形势,如何正确判断和评估电力行业上市公司的实际价值尤为重要。本文首先阐述了该主题的背景和意义。事实表明,电力行业上市公司的合理估值可以为投资者,管理者和决策者提供决策参考,并基于国外现代企业价值理论,系统地总结国内外的研究现状。从宏观经济环境,产业政策和电力行业的特点出发,研究影响电力行业估值的相关因素,在估值和资本市场不断发展的过程中,选择EVA经济增加值模型对大唐电力这家电力上市公司进行基本面分析,同时考虑到支持向量机具有良好的建模理论,无需系统性地探求数学模型,基于其对数据拟合运用结构风险最小化原则且具备优良的训练效率。本文对电力上市公司每日开盘股票与信息粒化结合进行回归预测判别。同时对支持向量机与信息粒化理论作了较为深入的概括,为原始数据优选核函数,构建遗传算法、粒子群优化算法和网格搜索算法三种参数寻优算法进行数据回归效果比对,并对其归一化处理。在此之后采用模糊粒化支持向量机模型对大唐电力进行股票回归预测,预测结果说明该模型对电力上市公司有很好的预测结果,本文还运用了梯形区间二型模糊系统对上述预测结果进行深化处理并且取得了更好的预测结果,本文可以为政策制定者和战略执行者提供有效地数据支撑和预测估值,对企业的战略性调整和我国电力行业进军海外提供重要的理论依据。更多还原

【Abstract】 Electric power industry is one of the pillar industries of China’s national economy.With the continuous change of economic situation,China’s electric power industry is currently in a period of adjustment,and the growth rate is slowing down.Faced with the new situation,how to correctly judge and evaluate the actual value of Listed Companies in the power industry is particularly important.Firstly,the background and significance of this topic are elaborated.The facts show that the reasonable valuation of Listed Companies in power industry can provide decision-making reference for investors,managers and decision-makers.Based on the foreign modern enterprise value theory,this paper systematically summarizes the research status at home and abroad.Starting from the macroeconomic environment,industrial policy and the characteristics of the power industry,this paper studies the relevant factors affecting the valuation of the power industry,and preliminarily understands the valuation objects at the macro and micro levels.In the process of valuation and capital market development,EVA economic value-added model is selected to analyze the fundamentals of Datang Electric Power Company.Considering that the support vector machine has good modeling theory,there is no need to systematically explore the mathematical model.Based on the data fitting,the principle of structural risk minimization is applied and the training efficiency is excellent.In this paper,the combination of daily opening stock and information granulation of power listed companies is predicted and discriminated by regression.At the same time,the theory of support vector machine and information granulation is summarized in depth,and three parameter optimization algorithms,genetic algorithm,particle swarm optimization and grid search algorithm,are constructed to optimize the kernel function for the original data,and the data regression effect is compared and normalized.After that,the stock regression prediction of Datang Electric Power Company is carried out by using the Fuzzy Granulation Support Vector Machine Model.The forecasting results show that the model has good forecasting results for the listed electric power companies.This paper also uses the trapezoidal interval two-type fuzzy system to deepen the forecasting results and achieve better forecasting results.This paper can provide policy makers and strategic executives with better forecasting results.Effective data support and predictive valuation provide important theoretical basis for strategic adjustment of enterprises and for China’s power industry to enter overseas.更多还原