机械制造与自动化2025，Vol.54Issue(3)：301-305,5.DOI:10.19344/j.cnki.issn1671-5276.2025.03.058
基于GNN和蒙特卡罗法的电力采购系统围标串标行为辨识算法
Identification Algorithm of Bidding Behavior in Power Procurement System Based on GNN and Monte Carlo Method
摘要
Abstract
Influenced by the diversity of bidding methods and the complexity of node relationship in power procurement system,the identification accuracy is usually dis satisfied due to neglecting the consideration of bidding pattern regularity when identifying the behavior of bidding collusion.To solve this problem,an identification algorithm of bidding behavior in power procurement system based on GNN and Monte Carlo method is proposed.The network diagram is constructed by combining the entity and cooperation relationship within the power procurement system,and the attention mechanism aggregation function is introduced.The embedding layer is used to transform the embedded representation of node/edge feature vectors,capturing the structural attribute information in the original power procurement system.A state vector is assigned to each bidder and the state transition probability is calculated.With Monte Carlo method,the bid-winning behavior is simulated through random sampling.Taking the bid price difference,bid winning frequency and bidding mode as the key characteristics of bidding collusion,the risk index is calculated,and combined with the judgment result of identification threshold,the bidding collusion behavior is identified from the power procurement transaction process,and the identification accuracy of the proposed method is tested.The final test results show that the logarithmic loss function value of the identification result is low and the identification accuracy is obviously high.关键词
电力采购/围标串标/行为辨识/招投标/GNN算法/蒙特卡罗法Key words
electric power procurement/string bidding/behavior identification/bidding/GNN algorithm/Monte Carlo method分类
信息技术与安全科学引用本文复制引用
吴勇,潘晓华,杨轶俊,叶雪峰,刘福权..基于GNN和蒙特卡罗法的电力采购系统围标串标行为辨识算法[J].机械制造与自动化,2025,54(3):301-305,5.基金项目
南方电网数字平台科技(广东)有限公司信息化建设项目 （广东）
南方电网数字平台科技(广东)有限公司开发项目(21000320220301022100673) （广东）
