基于减法聚类及自适应模糊神经网络的短期电价预测

提出了基于Takagi-Sugeno模型的自适应模糊神经网络的短期电价预测方法。首先采用减法聚类方法确定自适应模糊神经网络的结构,然后利用混合学习算法训练该网络的前件参数和结论参数,最后将影响未来日电价的相关因素输入到训练好的自适应模糊神经网络中进行电价预测。以美国加州电力市场公布的1999年负荷与电价数据进行模型训练和预测,结果表明采用该方法所建立的预测模型具有较高的预测精度。     

基于遗传算法和径向基函数神经网络的短期边际电价预测

文章分析了影响电价的主要因素及电价的变化特点,讨论了电价预测模型中必需引入的影响电价的因素。在比较常用的几种电价预测方法的优缺点后,作者采用径向基函数神经网络(radial basis function neural networks,RBF)建立短期边际电价预测模型,用递阶遗传算法(HGA)同时训练RBF网络结构和参数。并以美国New England ISO公布的2002年历史电价数据进行训练和测试,与传统的BP网络预测模型相比较, 测试结果证明该模型的预测精确度是令人满意的。     

Day-ahead price forecasting in restructured power systems using artificial neural networks

Over the past 15 years most electricity supply companies around the world have been restructured from monopoly utilities to deregulated competitive electricity markets. Market participants in the restructured electricity markets find short-term electricity price forecasting (STPF) crucial in formulating their risk management strategies. They need to know future electricity prices as their profitability depends on them. This research project classifies and compares different techniques of electricity price forecasting in the literature and selects artificial neural networks (ANN) as a suitable method for price forecasting. To perform this task, market knowledge should be used to optimize the selection of input data for an electricity price forecasting tool. Then sensitivity analysis is used in this research to aid in the selection of the optimum inputs of the ANN and fuzzy c-mean (FCM) algorithm is used for daily load pattern clustering. Finally, ANN with a modified Levenberg–Marquardt (LM) learning algorithm are implemented for forecasting prices in Pennsylvania–New Jersey–Maryland (PJM) market. The forecasting results were compared with the previous works and showed that the results are reasonable and accurate.     

基于动态计量经济学模型的短期电价预测

电力市场中的电价受众多因素影响,单变量时间序列法已很难提高短期电价的预测精度。针对该问题,文中运用时间序列模型的动态计量方法来预测短期电价。首先建立电价和电量的一般自回归分布滞后模型;然后对电价和电量的时间序列数据进行预处理;在通过平稳性和协整性检验后,建立误差修正模型,最终由Eviews 5.0估计出模型的参数。利用此模型对澳大利亚新南威尔士州电力市场的短期电价进行预测,结果表明此模型具有较高的预测精度。     

含误差预测校正的ARIMA电价预测新方法

在电力市场中，准确的电价预测是各市场参与方共同关心的重要问题。已经提出多种理论和方法尝试提高电价预测精度，然而由于影响电价的因素十分复杂，实践证明靠建立单一的电价预测模型来提高预测精度是非常困难的。该文在分析电价波动特性和现有预测方法的基础上，首次提出结合误差预测校正电价预测来提高预测精度的新思路。在建立常规电价预测模型的基础上，对预测后的残差形成的随机序列也迭代地建立预测模型，并用预测的误差修正电价预测结果。该文采用ARIMA方法建立电价预测和误差预测模型，并用加州电力市场的历史数据建立基于ARIMA的日平均电价预测模型，预测结果表明所提方法能明显改善预测精度，而且方法简捷明了，能够推广到小时电价预测、负荷预测和其它预测领域。     

电力市场短期边际电价的分时重构混沌相空间预测

为了实现高精度的电力市场短期边际电价预测,该文对市场边际电价时间序列数据分时段聚类进行了相空间重构,并分别计算分形维数和提取最大Lyapunov指数,经分析得出了边际电价分时序列数据的演化具有混沌特征,由此提出了短期边际电价的分时重构混沌相空间预测算法,相比目前通常采用的单一时间序列混沌预测算法,该算法具有相空间嵌入维数少和模型参数配置灵活的特点,通过电力市场短期边际电价预测实例验证,结果表明该算法比单一时序混沌预测算法在预测精度上有显著提高.     

基于小波分析与广义自回归条件异方差模型的短期电价预测

由于电价波动具有非线性及波动集群现象,因此提出了一种基于小波分析和广义自回归条件异方差模型相结合的短期电价预测新方法。首先应用小波分解原理将电价序列分解成低频部分和高频部分,在此基础上对各子序列分别建立广义自回归条件异方差模型并进行预测;然后利用小波理论对各子序列的预测结果进行重构,实现对原始电价序列的预测;最后以美国加州电力市场历史数据为例进行了验证,结果表明本文方法是可行和有效的。     

A Statistical Approach for Interval Forecasting of the Electricity Price

Electricity price forecasting is a difficult yet essential task for market participants in a deregulated electricity market. Rather than forecasting the value, market participants are sometimes more interested in forecasting the prediction interval of the electricity price. Forecasting the prediction interval is essential for estimating the uncertainty involved in the price and thus is highly useful for making generation bidding strategies and investment decisions. In this paper, a novel data mining-based approach is proposed to achieve two major objectives: 1) to accurately forecast the value of the electricity price series, which is widely accepted as a nonlinear time series; 2) to accurately estimate the prediction interval of the electricity price series. In the proposed approach, support vector machine (SVM) is employed to forecast the value of the price. To forecast the prediction interval, we construct a statistical model by introducing a heteroscedastic variance equation for the SVM. Maximum likelihood estimation (MLE) is used to estimate model parameters. Results from the case studies on real-world price data prove that the proposed method is highly effective compared with existing methods such as GARCH models.     

基于可信度加权组合的电价波动模式日前预报方法

准确的日前电价预测对电力市场参与者的优化决策具有重要意义。目前,大多数日前电价预测方法并不区分每天电价的波动模式而采用统一模型进行预测,当被预测日的波动模式与历史数据出现较大差异时无法保证预测的准确性。根据不同的日波动模式采用相似历史数据进行分类建模是解决此问题的有效途径,这就需要建立针对历史数据不同波动模式的分类识别模型和针对未来波动模式的日前预报模型。为此,文章提出一种针对分类预测的电价日波动模式日前加权组合预报方法。第一,采用K-means算法对日电价序列进行聚类分析,在分析聚类结果特性的基础上提取反映每日波动模式差异的特征向量,利用支持向量机分类（support vector machine for classification, SVC）方法建立电价数据日波动模式的识别模型;第二,利用多种常规方法建立日前电价预测模型对日前电价进行预测,并将预测结果输入日波动模式识别模型得到对应的模式识别结果;第三,根据多个方法波动模式预测结果对历史数据表现出来的不同精度,设计了基于可信度的组合机制,实现考虑预测准确性的加权组合预测,从而得到最终的日波动模式预测结果。利用美国PJM电力市场电价数据进行的仿真分析表明,提出的日前电价波动模式预测方法能得到较为准确的模式预测结果;利用电价波动模式日前预报进行分类预测的精度相对统一预测有显著提高。     

基于特征加权模糊聚类的电力负荷分类

电力用户的负荷分类为电力系统和电力部门的系统规划、负荷预测、分时电价等研究提供了基本的指导工作。利用基于变分模态分解（variational mode decomposition,VMD）和模糊C均值聚类算法（fuzzy C-means,FCM）实现电力负荷的分类研究,针对FCM中欧氏距离的特征权重唯一的问题,利用基于特征加权的模糊聚类方法,提出基于特征加权的VMD-FCM聚类算法。根据电网实测负荷数据,VMD算法可对数据的固有模态有效分解,结合FCM算法引入的权重系数,显著提高了算法收敛速度和聚类准确度。对聚类结果分析表明：所提VMD-FCM聚类方法能够有效区分不同负荷类型,具有实际应用价值,从而为电力系统的设计规划提供指导作用。     

Monthly electricity demand forecasting based on a weighted evolving fuzzy neural network approach

This research develops a weighted evolving fuzzy neural network for monthly electricity demand forecasting in Taiwan. This study modifies the evolving fuzzy neural network framework (EFuNN framework) by adopting a weighted factor to calculate the importance of each factor among the different rules. In addition, an exponential transfer function (exp(−D)) is employed to transfer the distance of any two factors to the value of similarity among different rules, thus a different rule clustering method is developed accordingly. Seven factors identified by the Taiwan Power Company will affect the power consumption in Taiwan. These seven factors will be inputted into the WEFuNN to forecast the electricity demand of the future. The historical data will be used to train the WEFuNN. After training, the trained model will forecast the future electricity demands. Finally, the WEFuNN model is compared with other approaches, which are proposed in the literature. The experimental results reveal that the MAPE for WEFuNN model is 6.43% which is better than the MAPE value for other approaches. Thus, the WEFuNN model is more accurate in forecasting the monthly electricity demand than the other approaches. In summary, the WEFuNN model can be practically applied as an electricity demand forecasting tool in Taiwan.     

基于R/S分析法的电价预测模型

在自由竞争的电力市场中,准确的电价预测对于电力市场所有参与者具有重要意义.针对电价突变性的特征给电价预测结果带来误差的问题,本文提出了一种基于R/S分析法的BP神经网络电价预测模型.运用R/S分析法对电价序列之间的关联性和相似性进行修正,并采用BP神经网络模型对电价进行预测.通过实验,验证了用R/S分析法修正后的数据进...     

考虑分时电价的居民用户短期用电量分类预测及修正方法

为了更好地利用智能电网中的用户用电信息,提高短期用电量预测精度,针对居民用户提出一种考虑分时电价的分类短期用电量预测及修正方法。首先,通过模糊聚类将用户按用电行为分类,将电价、用电量和加权气象日期影响因素作为预测模型输入量。然后,针对各类用户的用电特点,经仿真对比选择相适应的BP、Elman、LSTM神经网络算法构建预测模型。最后,运用修正算法对误差较大的峰谷值进行修正,将修正后的分类预测结果相加以获得整体预测值。以广东省云浮市某小区为例对该方法进行仿真分析,并与随机森林、CART等算法进行对比。实验结果证明所提方法具有更高的预测精度。     

Electricity Price Forecasting Based on AOSVR and Outlier Detection

Electricity price is of the first consideration for all the participants in electric power market and its characteristics are related to both market mechanism and variation in the behaviors of market participants. It is necessary to build a real-time price forecasting model with adaptive capability; and because there are outliers in the price data, they should be detected and filtrated in training the forecasting model by regression method. In view of these points, this paper presents an electricity price forecasting method based on accurate on-line support vector regression (AOSVR) and outlier detection. Numerical testing results show that the method is effective in forecasting the electricity prices in electric power market.     

Short‐term electricity load and price forecasting based on clustering and next symbol prediction

Short‐term electricity load and price forecasting is an important issue in competitive electricity markets. In this paper, we propose a new direct time series forecasting method based on clustering and next symbol prediction. First, the cluster label sequence is obtained from time series clustering. Then a lossless compression algorithm of prediction by partial match version C coder (PPMC) is applied on this obtained discrete cluster label sequence to predict the next cluster label. Finally, the whole time series values of one‐step‐ahead can be directly forecast from the predicted cluster label. The proposed method is evaluated on electricity time series datasets, and the numerical experiments show that the proposed method can achieve promising results in day‐ahead electricity load and price forecasting. © 2014 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

基于非参数GARCH模型的电价预测

基于非参数条件异方差估计理论,提出了一种改进的电价曲线预测方法。文中从实际电价曲线出发,针对条件方差函数建模,并采用非参数估计方法确定其模型。另外,在非参数估计中,针对条件标准差不可测困难,引入了迭代估计算法,通过不断修正作为输入量的条件标准差估计值来提高条件方差函数的估计可信度。在研究加州电力市场2000年日前电价时间序列波动特性的基础上,对Humb节点的日前电价时间序列进行建模并模拟预测。试验结果表明,文中所提模型能够更好地体现电价时间序列波动集群性这一特征,利用非参数估计所确定的模型提升了尖峰电价的预测效果。     

Next-Day Peak Electricity Price Forecasting Using NN Based on Rough Sets Theory

Electricity price forecasting is an essential task for market participants in deregulated electricity market. This paper proposes an approach for next-day peak electricity price forecasting, since it is important for risk management and bidding strategy. In the proposed method, neural network (NN) is employed as the forecasting method, and its learning data is selected by using rough sets. Moreover, the creating method of learning data based on temperature fluctuation is also proposed for generation of new learning data in order to efficiently learn. This method is examined by using the data of Pennsylvania-New Jersey-Maryland (PJM) electricity market and The independent electricity system operator (IESO) market. From the simulation results, it is observed that the proposed method is useful for next-day peak electricity price forecasting. Copyright © 2009 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.     

基于CEEMDAN分解的深度学习分位数回归电价预测

电力市场中电价预测的准确性对于供应商竞价策略的制定至关重要。针对电价预测问题,提出一种基于完备经验模态分解的深度学习分位数回归电价预测方法。首先,采用自适应噪声的完备经验模态分解方法对电价序列进行分解,得出各个模态分量;然后,采用深度学习中空洞因果卷积神经网络预测模型在不同分位数下对各个分量进行预测,并将预测结果重构;最后,对预测结果采用核密度估计得到电价的概率密度函数。经过对美国电力市场PJM的实际数据进行仿真验证,所提出的组合预测方法相比于其他分位数回归方法,不仅具有更高的预测准确度,且可以为供电商提供更多有效信息。     

基于模糊最小二乘支持向量机的系统边际电价预测方法

提出了基于模糊最小二乘支持向量机的系统边际电价（system marginal price,SMP）预测方法。为了减少样本数据中孤立点对回归性能的影响,将模糊隶属度的概念引入到最小二乘支持向量机中的同时,采用网格搜索和交叉验证的方法寻找最佳参数组合,使系统边际电价算法性能达到最佳。以美国加州电力市场的实际数据作计算实例,分别采用标准三层BP神经网络和模糊最小二乘支持向量机进行系统边际电价预测,结果表明基于模糊最小二乘支持向量机的系统边际电价预测的方法有效提高了预测精度。     

基于非参数GARCH的时间序列模型在日前电价预测中的应用

电力市场中电价序列具有较强的波动性、周期性和随机性,以致经常出现价格尖峰,这在很大程度上影响了电价预测的精度。提出了一种基于小波变换和非参数GARCH(generalized auto regressive conditional heteroskedasticity)模型的时间序列模型对日前电价进行预测。利用小波变换将历史电价序列分解重构概貌序列和细节序列,分别建立累积式自回归滑动平均(auto-regressive integrated moving average,ARIMA)模型进行预测,采用非参数GARCH模型对电价序列预测残差的随机波动率进行建模,从而提高对价格波动性的预测能力和ARIMA模型的预测精度。将该模型应用于美国宾夕法尼亚—新泽西—马里兰(Pennsylvania-New Jersey-Maryland,PJM)电力市场的日前电价预测。算例结果表明,非参数GARCH模型可以更好地拟合电价序列剧烈波动的特性,该模型能够提高电价的预测精度。