Identification of Abnormal Bidding Behavior Types of Power Suppliers in Electricity Market Based on Multiple Isolated Forests

Abstract

With the steady development of the electricity market, China's electricity market is increasingly showing a large-scale, multi-agent and other characteristics. There are many types of market violation risks and it is difficult to prevent them. Aiming at all kinds of violations in the electricity market, in order to ensure fair competition in the market and prevent and resolve market risks caused by abnormal bidding behaviors in the market, this paper proposes a method for identifying the types of abnormal bidding behaviors of power producers in the electricity market based on multiple isolated forests. Based on the identification index of multi-dimensional abnormal bidding behavior type proposed in this paper, this method first encodes the characteristics of different types of abnormal behavior in the market bidding process, and establishes the numerical characteristic table of abnormal behavior. Secondly, the multiple isolated forest algorithm proposed and constructed in this paper is used to identify the types of potential abnormal behaviors in the bidding process of e-commerce. The example part takes the actual bidding data of a power producer in the US PJM spot market as the research object for anomaly identification and analysis verification. The example results show that the method can effectively identify the abnormal bidding behavior types of the power producer in the market, verify the effectiveness of the method, provide decision support for market supervision, and help the power market run healthily and smoothly.