Laser interference detection parameter adaptive control method based on reinforcement learning

Abstract

The application discloses a laser interference detection parameter adaptive control method based on reinforcement learning, comprising the following steps: obtaining relevant data to form an interference state vector, and constructing an interference optical physics model and a safety constraint model; constructing a sub-target generation unit to generate a target optimization direction set; constructing a hierarchical reinforcement learning structure to output an adjustment control quantity sequence of a detection parameter; constructing an external reward generation unit and an internal reward generation unit, introducing a reward consistency mapping unit to correct the internal reward, and generating an updated hierarchical reinforcement learning structure; periodically updating the interference state vector to realize a laser interference detection parameter adaptive control closed loop. The application utilizes hierarchical reinforcement learning to combine optical physics and safety constraint modeling, realizes adaptive optimization control of a laser interference detection parameter, and has the advantages of high measurement precision, high intelligent degree of parameter adjustment, and safe and stable operation.

Images