Dynamic game analysis of coupled medium- to long-term and spot electricity markets under high renewable penetration

As the share of renewable energy in power systems continues to increase and demand-side volatility and uncertainty intensify, achieving coordinated decision-making and intertemporal alignment between medium- to long-term electricity market and electricity spot (ES) market has become a key challenge for resource allocation and risk management under high renewable penetration. To address this issue, this study develops a dynamic game model that couples medium- to long-term electricity and electricity spot markets and explicitly characterizes the decision-making behavior of multiple market participants, including wind generators, conventional generators, and the grid company. ES market bidding is formulated as a Bellman equation and solved using Approximate Dynamic Programming with an online actor–critic reinforcement learning algorithm. In addition, an Input Convex Neural Network combined with inverse optimization is used to capture the reaction mapping from contract parameters to spot market revenues. On the basis, the ADMM algorithm is further applied to coordinate contract signing and ES bidding strategies across multiple participants and to obtain the corresponding market equilibrium. Numerical results show that: uncertainty in renewable generation is shifting electricity spot price formation from demand-side dominance toward supply-side dominance, giving rise to a characteristic “load-follows-resource” pattern; system load still plays a decisive role in shaping the generation mix and the scale of contract lock-in, highlighting the indispensable capacity and balancing value of conventional generators under high-load conditions.