Cognitive edge computing node parameter optimization method and device based on reinforcement learning

The invention provides a cognitive edge computing node parameter optimization method and device based on reinforcement learning. The method comprises the following steps: determining a partial observable Markov decision model based on a frequency band use state of a current time slot of a main user side, and determining a belief probability corresponding to the main user side in each time slot in the future and an observation probability and reward corresponding to a secondary user side in each time slot in the future by using the partial observable Markov decision model; and based on the belief probability, the observation probability and the reward corresponding to each time slot in the future, and the target state probability of the secondary user side, constructing a Bellman optimization model, and based on the Bellman optimization model, maximizing an average reward within a preset future time slot range in a corresponding target state, and determining a corresponding cognitive edge computing node parame