基于ARMA-GARCH模型的电价预测与研究
发布时间：2018-08-04 08:17
【摘要】：电价在电力市场中的重要性不言而喻,它既能反映出电力市场中的供求关系,又能调节和控制电力市场的交易,所以电价作为电力市场竞争效率的核心部分,电价的确定在电力市场中对市场各参与方来说都是最重要的部分。随着电力市场的改革浪潮席卷全球和电力市场打破垄断、相互竞争局面的形成,电力市场各参与方便十分注重电价的预测。因为准确的电价预测对市场各参与方来说具有十分重要的意义,他们在电力市场竞争中做出相关决策时,可以将预测的电价作为参考依据,以便在电力市场交易中处于有利地位,因此,如何根据电力市场中的历史电价数据和电价的相关特点准确预测出未来电价,已然成为国内外学者研究的热点,故准确的电价预测也变得越来越重要。电价具有与其他商品不同的特征,由于电价受众多因素的影响,使得电价具有均值回复、较强的波动性、较强的跳跃性、以及价格尖峰和杠杆效应等特征,电价的这些特征加大了对电价预测的难度。目前,国内外学者已相继提出多种预测方法,主要有时间序列法、人工神经网络法、基于小波理论分析法和组合模型预测法等,本文主要分析基于时间序列方法建模的模型。本文针对系统电价的特征和不同市场的特点,运用GARCH模型、TGARCH模型、EGARCH模型和PARCH模型对PJM电力市场、MISO电力市场(包括MISO中的三个节点市场)以及New England电力市场一共6个市场的电价序列分别构建了预测模型。在模型估计时假设残差分别服从正态分布、学生t分布和广义误差分布,从而比较不同电力市场下不同模型的预测精度,通过比较分析得出,由于不同市场的电价数据特征不同,GARCH族模型的预测精度也会有所不同。很难单一的从某方面来说GARCH族模型中哪个模型的预测效果更佳,根据不同电力市场的特点以及假设残差的不同分布,GARCH模型、TGARCH模型、EGARCH模型和PARCH模型分别适合不同的电力市场电价预测。
[Abstract]:The importance of electricity price in electricity market is self-evident. It can not only reflect the relationship between supply and demand in electricity market, but also regulate and control the transaction of electricity market, so electricity price is the core part of the competitive efficiency of electricity market. The determination of electricity price is the most important part for all participants in the electricity market. With the reform of the electricity market sweeping the world and the electricity market breaking the monopoly and the formation of mutual competition, the electricity market is very convenient to participate in the prediction of electricity price. Because accurate electricity price prediction is of great significance to all participants in the market, when they make relevant decisions in the competition of the electricity market, they can use the electricity price forecast as a reference basis. In order to be in a favorable position in electricity market transaction, therefore, how to accurately predict the future electricity price according to the historical electricity price data and the relevant characteristics of electricity price in the electricity market has become a hot spot of domestic and foreign scholars. Therefore, accurate electricity price prediction is becoming more and more important. Electricity price has different characteristics from other commodities. Because electricity price is influenced by many factors, electricity price has the characteristics of average return, strong volatility, strong jump, price spike and leverage effect, etc. These characteristics of electricity price increase the difficulty of forecasting electricity price. At present, many prediction methods have been put forward by domestic and foreign scholars, such as time series method, artificial neural network method, wavelet theory analysis method and combination model prediction method, etc. In this paper, the modeling model based on time series method is mainly analyzed. This paper aims at the characteristics of system electricity price and the characteristics of different markets. Using the GARCH model, the EGARCH model and the PARCH model are used to predict the price sequence of the miso electricity market (including the three node markets in MISO) and the six markets in the New England electricity market. In the estimation of the model, it is assumed that the residuals are divided into normal distribution, student t distribution and generalized error distribution, so that the prediction accuracy of different models in different power markets is compared. The prediction accuracy of GARCH model will be different because of the different characteristics of electricity price data in different markets. It's hard to single out, in some ways, which model of the GARCH family has the best prediction effect. According to the characteristics of different power markets and the different distributions of the assumed residuals, the TGARCH model and the PARCH model are respectively suitable for different electricity market price forecasting.
【学位授予单位】：重庆师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F426.61;F726
[Abstract]:The importance of electricity price in electricity market is self-evident. It can not only reflect the relationship between supply and demand in electricity market, but also regulate and control the transaction of electricity market, so electricity price is the core part of the competitive efficiency of electricity market. The determination of electricity price is the most important part for all participants in the electricity market. With the reform of the electricity market sweeping the world and the electricity market breaking the monopoly and the formation of mutual competition, the electricity market is very convenient to participate in the prediction of electricity price. Because accurate electricity price prediction is of great significance to all participants in the market, when they make relevant decisions in the competition of the electricity market, they can use the electricity price forecast as a reference basis. In order to be in a favorable position in electricity market transaction, therefore, how to accurately predict the future electricity price according to the historical electricity price data and the relevant characteristics of electricity price in the electricity market has become a hot spot of domestic and foreign scholars. Therefore, accurate electricity price prediction is becoming more and more important. Electricity price has different characteristics from other commodities. Because electricity price is influenced by many factors, electricity price has the characteristics of average return, strong volatility, strong jump, price spike and leverage effect, etc. These characteristics of electricity price increase the difficulty of forecasting electricity price. At present, many prediction methods have been put forward by domestic and foreign scholars, such as time series method, artificial neural network method, wavelet theory analysis method and combination model prediction method, etc. In this paper, the modeling model based on time series method is mainly analyzed. This paper aims at the characteristics of system electricity price and the characteristics of different markets. Using the GARCH model, the EGARCH model and the PARCH model are used to predict the price sequence of the miso electricity market (including the three node markets in MISO) and the six markets in the New England electricity market. In the estimation of the model, it is assumed that the residuals are divided into normal distribution, student t distribution and generalized error distribution, so that the prediction accuracy of different models in different power markets is compared. The prediction accuracy of GARCH model will be different because of the different characteristics of electricity price data in different markets. It's hard to single out, in some ways, which model of the GARCH family has the best prediction effect. According to the characteristics of different power markets and the different distributions of the assumed residuals, the TGARCH model and the PARCH model are respectively suitable for different electricity market price forecasting.
【学位授予单位】：重庆师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F426.61;F726
【相似文献】
相关期刊论文 前10条
1 李娜;李郁侠;王丽霞;杨亚刚;;电价与负荷的相关性分析及其在电价预测中的应用[J];武汉大学学报(工学版);2009年04期
2 张建,马光文,杨东方,过夏明,王立明;边际电价预测的三时点模型[J];水电能源科学;2003年02期
3 潘e，
本文编号：2163215
本文链接：https://www.wllwen.com/jingjilunwen/guojimaoyilunwen/2163215.html
