In the air traffic system, constructing a unified supply-demand situations modeling framework for different decision-making stages is critical for efficient multi-level and multi-stage collaborative decision-making. Accordingly, a heterogeneous-Agent-based unified deduction model of supply-demand situations in air traffic network is developed, referred to as the heterogeneous Agent model. First, the theoretical analysis demonstrates that the completeness of the airspace network structure has a decisive impact on the accuracy of delay characterization, and clarifies the functional relationship between network node completeness and prediction error. Then, by integrating the Agent interaction mechanism with fluid queuing theory, a unified dynamic multi-element coupling framework covering flights, airports, and airspace is constructed. Three types of heterogeneous Agents (flight, airport and sector) are defined to establish state transition and congestion/delay propagation mechanisms. Based on historical Automatic Dependent Surveillance-Broadcast (ADS-B) trajectory data, sector service time is calibrated, and the main input parameters of the sector fluid queuing system are determined, enabling the cross-level mapping and parallel deduction of system operating states across multiple levels. Finally, using China-wide flight operation data at the flight-season scale covering 250 airports and 287 sectors as the sample, the model is validated in three scenarios: flight schedule configuration, next-day flight planning, and sudden capacity degradation. The results show that the heterogeneous Agent model achieves higher delay prediction accuracy than existing methods in all scenarios. Capable of integrated “flight-airport-airspace” supply-demand situations analysis across strategic, pre-tactical, and tactical decision-making stages, and providing a reliable, accurate, and efficient decision-support for planning, evaluation and operational management of air traffic system.

Reinforcement Learning (RL) techniques are being studied to solve the Demand and Capacity Balancing (DCB) problems to fully exploit their computational performance. A locally generalised Multi-Agent Reinforcement Learning (MARL) for real-world DCB problems is proposed. The proposed method can deploy trained agents directly to unseen scenarios in a specific Air Traffic Flow Management (ATFM) region to quickly obtain a satisfactory solution. In this method, agents of all flights in a scenario form a multi-agent decision-making system based on partial observation. The trained agent with the customised neural network can be deployed directly on the corresponding flight, allowing it to solve the DCB problem jointly. A cooperation coefficient is introduced in the reward function, which is used to adjust the agent’s cooperation preference in a multi-agent system, thereby controlling the distribution of flight delay time allocation. A multi-iteration mechanism is designed for the DCB decision-making framework to deal with problems arising from non-stationarity in MARL and to ensure that all hotspots are eliminated. Experiments based on large-scale high-complexity real-world scenarios are conducted to verify the effectiveness and efficiency of the method. From a statistical point of view, it is proven that the proposed method is generalised within the scope of the flights and sectors of interest, and its optimisation performance outperforms the standard computer-assisted slot allocation and state-of-the-art RL-based DCB methods. The sensitivity analysis preliminarily reveals the effect of the cooperation coefficient on delay time allocation.