全球能源互联网2025，Vol.8Issue(1)：110-123,14.DOI:10.19705/j.cnki.issn2096-5125.2025.01.012
面向混合增强智能的省级现货市场报价样本集增强及市场力识别技术
Provincial Spot Market Quote Sample Set Enhancement for Hybrid Augmented Intelligence and Market Force Recognition
摘要
Abstract
With the advancement of electric power spot market reform,there is an urgent need to conduct market power identification on the quotation behavior of provincial spot market participants to encourage healthy competition.Considering that in the initial stage of the market,the number of samples derived from the spot market practice in which market members exercise market power to obtain excess profits is relatively small compared to the samples from the normal trading behavior,there is a sample imbalance problem.In this paper,first of all,the market power label tagging was labeled based on experts'experience.In this paper,the samples of spot market practice with market power labels based on experts'experience were labeled and used to train the Random Forest algorithm for market power identification,and then enhance the sample imbalance to improve the accuracy of market power identification by using the sample sampling method.In addition,it is considered that as the market matures,the criteria for judging market power are not static,and in order to cope with the changing criteria for market power,the"human intelligence"and the"machine intelligence"were combined in this paper.The results of the algorithm show that after the market power label is changed by"human intelligence",and"machine intelligence"can still recognize the market power samples effectively,which reflects the advantages of the hybrid augmented intelligence proposed in this paper.关键词
混合增强智能/多维度市场力样本集/样本不平衡/随机森林算法Key words
hybrid augmented intelligence/multidimensional market power sample set/sample imbalance/random forest algorithm分类
动力与电气工程引用本文复制引用
宁龙飞,刘飞宇,王蓓蓓,郑亚先..面向混合增强智能的省级现货市场报价样本集增强及市场力识别技术[J].全球能源互联网,2025,8(1):110-123,14.基金项目
国家电网有限公司科技项目(涵盖极端形态的省级日前电能市场推演场景智能构建技术研究,SGZJJH00DKJS2310199).Science and Technology Foundation of SGCC(Research of Provincial Day Ahead Power Market Simulation Scenario Intelligent Construction Technique Covering Extreme Pattern,SGZJJH00DKJS2310199). （涵盖极端形态的省级日前电能市场推演场景智能构建技术研究,SGZJJH00DKJS2310199）
