Analysis of Risk Characterization and Identification Techniques for Electricity Spot Settlement Based on Multilayer Perceptron Machine

Abstract

Abstract In this paper, the structural characteristics of the perceptron neural network and the calculation method of the hierarchical relationship of the MLP neural network model are first studied. Then the aspects of two-part settlement, generation-side settlement, and customer-side settlement in the electricity spot settlement mechanism are studied, and the importance of these mechanisms for the operation of the electricity market and risk identification is pointed out. Following that, the effectiveness of the risk identification model is assessed and analyzed. This paper examines market performance indicators, the impact of the dual-track mechanism, and the time characteristics of the price index to characterize risk. The results show that in Guangdong, for example, the price in the day-ahead market is much higher than the supply-demand equilibrium price most of the time, and the maximum difference can be as high as 0.662 yuan/(kW-h). For the entire month, the real-time market’s average price is RMB 0.546/(kW-h) and it is RMB 0.063/(kW-h) higher than the day-ahead market. The importance of this study lies in its role in managing and responding to risks for electricity market operators and participants.