摘要
由于电力市场价格的不确定性,供需侧管理在电力市场中遇到了许多困难。电力供应商可以通过了解电力市场价格变化的信息,在短期预测其合理报价时获得更多优势。因此,近年来,电力市场对价格预测的研究变得更加重要。根据预测框架,预测技术可分为三类,即统计模型、时间序列方法和基于人工智能(AI)的方法。因此,为解决上述问题,提出一种基于混合人工神经网络的短期电价预测混合方法。同时,本研究开发了一种基于互信息(MI)和神经网络(NN)相结合的特征选择技术,用于选择输入变量子集,这些子集对电价预测具有重要影响。通过结合人工协同搜索算法(ACS)和人工神经网络(ANN)进行,进一步提高了预测的精度。通过比较所提出的混合预测方法与混合SVM和混合ANN方法的相关性和精度,并通过粒子群优化(PSO)CSA算法对混合SVM和混合ANN方法的参数进行了优化。开发的ANN-ACS模型在电力市场具有鲁棒性。在电价预测的情况下,它提供了比其他AI方法更高的预测精度和简单性,在冬季、春季、夏季和秋季的MAPE值分别为4.58%、1.2%、2.62%和3.79%。
Supply and demand side management has encountered many difficulties in the electricity market due to the uncertainty of electricity market prices.Electricity suppliers can gain more advantages in predicting their reasonable offers in the short term by knowing information about price changes in the electricity market.Therefore,the study of price forecasting in the electricity market has become more important in recent years.According to the forecasting framework,forecasting techniques can be classified into three categories,namely,statistical models,time series methods,and artificial intelligence(AI)-based methods.Therefore,in this paper,a hybrid method for short-term electricity price forecasting based on hybrid artificial neural networks is proposed to address the above issues.Also,a feature selection technique based on a combination of mutual information(MI)and neural network(NN)is developed in this study for selecting subsets of input variables that have a significant impact on electricity price forecasting.The prediction accuracy is further improved by combining the artificial collaborative search algorithm(ACS)and artificial neural network(ANN)for this purpose.By comparing the correlation and accuracy of the proposed hybrid forecasting method with the hybrid SVM and hybrid ANN methods,and by optimizing the parameters of the hybrid SVM and hybrid ANN methods with the particle swarm optimization(PSO)CSA algorithm.The developed ANN-ACS model is robust in the electricity market.In the case of electricity price forecasting,it provides higher forecasting accuracy and simplicity than other AI methods,with MAPE values of 4.58%,1.2%,2.62%and 3.79%in winter,spring,summer and autumn,respectively.
作者
马会领
曲尧
王星凯
MA Huiling;QU Yao;WANG Xingkai(SDIC Power Holding Co.,Ltd.,Beijing 100034,China)
出处
《自动化与仪器仪表》
2023年第8期250-256,共7页
Automation & Instrumentation
基金
国投电力软科学研究项目,名称:发电侧参与现货市场规范化流程及报价策略研究(000052-21XB0029)。
关键词
人工神经网络
电力市场
电价预测
互信息
向量机
artificial neural network
electricity market
electricity price forecast
mutual information
vector machine
