基于自注意力编码高斯混合网络的电力市场异常行为检测OA

Against the industry backdrop of new energy integration into the market and the regular operation of the spot market,the types of market participants have become more diverse,the number of participants has significantly increased,and the frequency of transactions has continuously risen.The operating environment of the electricity market has become increasingly complex,and the difficulty of detecting abnormal behaviors of market entities has further increased.Traditional detection methods rely on manual review and rule-based judgment.When dealing with massive,high-frequency,and high-dimensional transaction data,their efficiency and accuracy can no longer meet the requirements.To address the above issues,this paper proposes an algorithm for detecting abnormal behaviors of market entities based on a deep self-attention encoding Gaussian mixture network.Firstly,considering the participation of various types of generating units in market bidding,detection indicators for abnormal behaviors of market entities are constructed.Secondly,a method for detecting abnormal behaviors based on a deep self-attention encoding Gaussian mixture network is proposed.The expression network and the improvement and reconstruction of the loss function of the self-attention encoding Gaussian mixture model are theoretically elaborated in detail,and the abnormal suspicion index and the detection process for abnormal behaviors are proposed.Finally,through experimental simulations,the performance of the proposed method in detecting abnormal behaviors is verified,and suggestions for the application scenarios of the algorithm are provided.