The smooth and healthy development of the power market is an important goal of power market management. During the operation of the power market, the operation risk plays a crucial role in the safe, reliable and stable development of the power market. The article starts from the deviation probability that exists in the new energy access to the southern regional power market, combines with the southern regional power market clearing model, and constructs the power market transaction risk evaluation index system. It also utilizes the cloud entropy method to solve the index weights, and then combines the cloud model with the material element topology model to construct the material element topology cloud model for evaluating the transaction risk level of the southern regional power market. Based on Stacking integrated learning, a variety of machine learning algorithms are introduced to construct an early warning model for power trading risks. The study shows that the comprehensive score of the power market transaction risk in the southern region is 1.52, and the overall risk rating is “low”, with a low leakage rate of the power market transaction risk warning, the average value of which is only 2.05%. Relying on the access of new energy in the new power system, combined with the power market transaction risk assessment and early warning model, the accurate early warning of power market transaction risk can be realized, laying a foundation for ensuring the stability of power market transactions.
