The invention relates to the field of intelligent driving, in particular to a parking strategy based on deep reinforcement learning. In the prior art, the traditional automatic parking system is poorin effect based on a traditional path planning algorithm. The invention provides a parking method and system based on deep reinforcement learning; the method and the system can obtain a parking planning route through a deep reinforcement learning algorithm; a tuple is formed by a vehicle observation state, a vehicle prediction action and a reward function; the parking planning method based on thetuple has the characteristic of extracting features based on product characteristics, so that the required parameters are few. In addition, the coefficient is not required to be adjusted based on an objective function (distance + steering + collision); according to the parking strategy in the invention, the features are extracted by adopting a deep reinforcement learning mode, and the strategy hasthe beneficial technical effects of being high in overall planning time, fast in response to the outside and the like.