Intelligent agent exploration method adaptive to half-sine period

The invention discloses an intelligent agent exploration method of a self-adaptive half-sine period, an intelligent agent explores an environment according to values of exploration factors, and the values of the exploration factors comprise a half-sine variation function, an exponential decay function and a constant. The size of the exploration factor changes in a half-sine periodic manner along with the increase of the number of learning curtains and has an exponential decay trend; when the intelligent agent finds a new state, the exploration factor is immediately adjusted to a larger value when the curtain is not ended, and the intelligent agent is encouraged to explore the state space more randomly; the intelligent agent exploration method is combined with the Q-learning algorithm, and the exploration strategy of the intelligent agent exploration method can significantly increase the convergence speed of the value function while improving the state space exploration capability of the intelligent agent. 本发明