A Tidal Phenomena Control Method for Shared Bicycle System Based on Dynamic Electronic Fence and Reinforcement Learning

Abstract

The present invention faces the imbalance between supply and demand of the shared bicycle system, and proposes a tidal phenomenon control method of the shared bicycle system based on dynamic electronic fence and reinforcement learning, including the following steps: (1) determining the state information of the electronic fence group; (2) Determine the scheduling action of the electronic fence; (3) determine the behavior and interaction of the agent; (4) determine the current benefits that can be obtained by taking action a; (5) determine the reinforcement learning environment in the electronic fence scheduling system; (6) determine the The agent state transition rules of the DQN neural network; (7) Construct the DQN neural network and perform forward calculation; (8) Use the random exploration strategy to select each output action; (9) Training and parameter update of the DQN neural network model (10) Judging whether the training of the DQN neural network is completed; (11) Inputting the initial moment and the initial state of the electronic fence group into the trained neural network to obtain the control strategy of the electronic fence.

Description

(1) Technical field [0001] Aiming at the imbalance between supply and demand of the urban shared bicycle system, the present invention proposes a tidal phenomenon control method of the shared bicycle system based on dynamic electronic fence and reinforcement learning. This method is based on the premise of limited number of bicycles and dynamic electronic fence, and uses reinforcement learning method to scale the electronic fence and induce pedestrians to park their bicycles to a reasonable position before the tide phenomenon comes. This method aims to improve customer satisfaction and bicycle utilization, and establishes a reinforcement learning model based on Deep Q Network (DQN) to optimize the control and decision-making of the tidal phenomenon of shared bicycles, thereby alleviating or solving the tides in the shared bicycle system. Phenomenon. The method belongs to the field of intelligent transportation. (2) Background technology [0002] In order to alleviate urban...

AI Technical Summary

Problems solved by technology

[0005] The present invention proposes a tidal control scheme based on an electronic fence based on a shared bicycle system with reinforcement learning as the core. By regulating and inducing the parking position of pedestrians, the automatic allocation of the bicycle position before the arrival of the traffic peak is realized, thereby solving the problem caused by the tidal phenomenon. supply and demand imbalance

Images