基于DDPG算法的桥式吊车自适应防摆控制
Adaptive Anti-sway Control of Bridge Cranes Based on DDPG Algorithm
辛增淼 1万思成 2高永锹 2王天雷 2郝晓曦 1邱光繁3
作者信息
- 1. 五邑大学 机械与自动化工程学院,广东 江门 529020
- 2. 五邑大学 电子与信息工程学院,广东 江门 529020
- 3. 江门市蒙德电气股份有限公司,广东 江门 529040
- 折叠
摘要
为解决桥式吊车控制中存在的台车定位速度慢和负载摆动大等问题,提出一种基于深度确定性策略梯度(DDPG)的控制策略.将桥式吊车设为强化学习的智能体;通过设定状态量及其误差作为智能体观测目标,设计合适的惩罚与奖励函数;智能体通过对控制系统的实时运行情况进行响应来生成控制动作.仿真结果表明,与传统控制算法相比,基于DDPG强化学习的算法在定位速度和摆角抑制方面的表现都优于传统方法,展现出较传统控制算法更高的性能和应用潜力.
Abstract
To overcome the challenges of slow trolley positioning and pronounced load swinging in overhead crane control,a control strategy based on Deep Deterministic Policy Gradient(DDPG)is proposed.Initially,the overhead crane is conceptualized as an agent within the framework of reinforcement learning.Subsequently,by establishing state variables and their associated errors as the observation targets of the agent,appropriate penalty and reward functions are meticulously designed.Furthermore,the agent generates control actions by responding dynamically to the real-time operational conditions of the control system.Simulation results reveal that,in comparison with traditional control algorithms,the algorithm based on DDPG reinforcement learning exhibits superior performance in terms of positioning speed and swing angle suppression.
关键词
桥式吊车/强化学习/深度确定性策略梯度/自适应控制/防摆控制Key words
Bridge cranes/Reinforcement learning/Deep Deterministic Policy Gradient/Adaptive control/Anti-swing control引用本文复制引用
出版年
2025
NETL
NSTL
万方数据