To address the challenges posed by varying heat generation modules in Ground Control Stations (GCS) during various work modes, a cooling system has been developed. This research introduces an Adaptive Variable Channel Control (AVCC) cooling approach using the Deep Reinforcement Learning Soft Actor-Critic (SAC) algorithm. The primary contributions of this paper include: (1) the design of a distributed cooling module featuring multiple cooling fans, which enables a variable channel cooling structure
(2) the development of a multi-module temperature control platform that simulates the heat generation conditions of each module under six work modes, providing a training environment for the cooling control algorithm
(3) the formulation of a model-free control method based on the SAC algorithm, AVCC, to optimize the cooling efficiency and endurance of the GCS. Finally, the effectiveness of the AVCC method was evaluated under maximum load conditions, contrasting it with a rule-based policy. The experimental results indicate that the AVCC method is able to cool the modules B (voltage stabilizer), Q (battery), E (charger), G (data transmission), D (voltage stabilizer), C (picture transmission), H (computer), and F (voltage stabilizer) from 78.9 °C, 58.5 °C, 88.3 °C, 88.3 °C, 79.2 °C, 88.9 °C, 77.5 °C, and 78.8 °C, respectively, to 50 °C in 280 s, improving the cooling efficiency by 40.4% and decreasing the energy consumption by 42.2% compared to the rule-based approach. The proposed distributed cooling module and AVCC method improve cooling efficiency and reduce energy consumption in the GCS. This study provides a valuable reference for the control and design of cooling systems in electronic equipment cabinets, especially those with similar shapes, sizes, cooling methods, number of heat sources and environmental conditions.