基于时变Copula的供电公司多期购电组合优化模型  

A Multi-Period Electricity Purchasing Model for Power Supply Company

作　　者：张宗益[1] 亢娅丽[1] 郭兴磊[1] 

机构地区：[1]重庆大学经济与工商管理学院,重庆400030

出　　处：《管理工程学报》2013年第1期147-152,共6页Journal of Industrial Engineering and Engineering Management

基　　金：国家自然科学基金资助项目(70941029)

摘　　要：针对供电公司在多市场、多阶段的购电组合问题,构建了以供电公司期末收益最大、以多期一致性风险度量函数测度的风险最小的多期购电组合优化模型。其中考虑到电力日前期货市场和实时市场电价序列的尖峰、厚尾特性和两市场电价序列间的相关结构,采用时变SJC Copula-(GARCH-T,GARCH-GED)模型拟合电价序列,并进行模拟。算例结果表明,多期投资组合决策优于单期连续决策,也优于整体购电决策。This paper studies the power supply company＇s electricity purchasing problems in an open market （e. g. medium and long- term contract market and spot market） where power can be freely traded. Owing to the high volatitity of electricity price, the power supply company must purchase electricity from different markets to spread price risk based on financial investment portfolio theory. Most of previous literatures mainly research ona power supply company＇s single-phase decision-making process. However, the decision-making process is a typical multi-phase problem. As the decision made in one stage may influence the decision and profit in the next stage, we need to consider the multi-phase circumstance in order to optimize the risk of accumulated return. It is important to estimate and simulate the loss distribution function of asset portfolio when measuring the risk of accumulated return since the relationships among multiple variables are characterized by their joint distribution. Therefore, we need to carefully analyze the correlation between electric assets before considering the power market＇s portfolio strategy. Most of the previous literatures assert that price sequence follows a normal distribution and measures the correlation between electric power assets using linear correlation coefficient. Other studies show that thick tail and heteroscedasticity are typical phenomena in the price return series data. This strong ~ nonlinear characteristic between power assets makes it difficult to use linear correlation coefficient to properly describe their relationships. Fortunately, Copula function can be used to describe the non-linear correlation structure between assets. ~ Based on the above analysis, we establish amulti-phase power purchasing model which takes the maximum expected accumulated return and the minimum risk measured by dynamic coherent risk function as objectives. We adopt the time-variation SJC Copula - （GARCH-T, GARCH-GED） model to fit and simulate a series of power prices, by consider

关 键 词：投资组合 时变COPULA 多期一致性风险测度 

分 类 号：F830.33[经济管理—金融学]