基于TD3算法的热管堆功率控制算法优化研究  

Optimization Reserach of Heat Pipe Reactor Power Control Algorithm Based on TD3 Algorithm

作　　者：宋霄森 余刃[1] 毛伟[1] 殷少轩 SONG Xiaosen;YU Ren;MAO Wei;YIN Shaoxuan(Naval University of Engineering,Wuhan 430033)

机构地区：[1]海军工程大学,武汉430033

出　　处：《舰船电子工程》2023年第8期104-109,共6页Ship Electronic Engineering

摘　　要：核反应堆功率控制仍多采用PID控制算法,但其控制参数往往难以选择,也难以在不同功率水平下均保持最优的控制效果。论文针对热管冷却核反应堆的功率控制设计PID控制算法,并基于深度强化学习TD3算法来实现对PID控制器的参数寻优。对比基于试凑法和参数寻优选取的PID参数的控制效果,采用深度强化学习TD3算法寻优参数可以获得更快速稳定的控制效果。PID control algorithm is still used in power control of nuclear reactors,but it is difficult to select the control parame⁃ters,and it is difficult to maintain the optimal control effect under different power levels.In this paper,the PID control algorithm is designed for the power control of heat pipe cooled nuclear reactor,and the parameters of the PID controller are optimized based on the deep reinforcement learning TD3 algorithm.Compared with the control effect of PID parameters selected based on trial and error method and parameter optimization,the deep reinforcement learning TD3 algorithm for parameter optimization can obtain more rap⁃id and stable control effect.

关 键 词：热管冷却核反应堆 深度强化学习 TD3算法 功率控制 

分 类 号：TL362.5[核科学技术—核技术及应用]