基于强化学习的转台控制参数自整定方法研究

Self-Tuning Method of Turntable Control Parameters Based on Reinforcement Learning

为保证转台能够获得较好的精度及动态性能，需要对其控制参数进行调节。为了避免传统上转台控制参数需要专业人员长时间调节的不足，提出一种基于强化学习的转台控制参数自整定方法。该方法通过引入强化学习自整定回路对控制参数进行调节，根据转台性能指标建立评价函数，依照评价函数结果指导参数调节的方式。该方法通过离线方式对转台控制参数进行调节，避免了强化学习调节实时性不足对转台运行造成的影响，同时具有能够自动调节转台控制参数的优势，有效提高了生产效率。

To ensure that the turntable can achieve better accuracy and dynamic performance, the control pa- rameters need to be adjusted. In order to avoid the shortage that traditional turntable control parameters requi- ring professionals to adjust for a long time, a self-tuning method of turntable control parameters based on rein- forcement learning is proposed. This method adjusts the control parameters by introducing reinforcement learn- ing self-tuning loop. The evaluation function is established according to the performance index of the turntable, and the way of parameter adjustment is guided according to the evaluation function result. This method adjusts the control parameters of the turntable in an off-line manner. It avoids the influence of insufficient real-time learning on the operation of turntable. At the same time, it has the advantages of automatically adjusting control parameters of the turntable and effectively improving the production efficiency.