电力大数据环境下异常交换行为的识别  

Abnormal Interchange Behavior Identification in Electric Power Big Data Environment

作　　者：胡晶 HU Jing(School of Information and Intelligent Transportation,Fujian Chuanzheng Communications College,Fuzhou 350007,China)

机构地区：[1]福建船政交通职业学院信息与智慧交通学院,福建福州350007

出　　处：《山东电力高等专科学校学报》2025年第1期39-42,共4页Journal of Shandong Electric Power College

基　　金：福建船政交通职业学院科教发展基金项目“基于大数据技术的海量数据分类查询算法的研究”(项目编号:Z202311033)。

摘　　要：针对电力大数据环境下异常交换行为,采用孤立森林算法对电力交换数据进行异常检测,并对检测性能进行评估。为进一步提高异常检测的精确性,引入生成对抗网络算法,学习并逼近真实数据分布,将判别器的输出作为异常分数,从而实现无监督异常检测,并采用特征匹配策略、小批量判别机制等优化策略提升性能和稳定性。实验结果表明,基于孤立森林和生成对抗网络的模型能够有效地进行异常交换行为识别,并且优于其他经典异常检测算法。In allusion to the abnormal interchange behavior in the electric power big data environment,the isolation forest algorithm is used to detect the anomalies in the electric power interchange data and evaluate the detection performance.To further improve the accuracy of anomaly detection,the generative adversarial networks algorithm is introduced,which can learn and approximate the distribution of real data and take the output of the discriminator as the anomaly score,thus achieving the unsupervised anomaly detection.The optimization strategies,such as feature matching strategy and small batch discrimination mechanism,are adopted to improve the performance and stability.The experimental results show that the model based on the isolation forest and generative adversarial networks can effectively perform abnormal interchange behavior identification,and outperforms other classical anomaly detection algorithms.

关 键 词：电力大数据 异常交换行为识别 孤立森林 生成对抗网络 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]