强化学习技术在工业产品质检调度中的实践

针对工业产品质检调度问题,讨论了模拟退火算法和Q-Leaning强化学习算法的实践过程。首先描述了本次研究存在的问题,其次抽象出了质检顺序矩阵和质检时间矩阵并进行问题求解。在实际应用中,选择模拟退火算法还是Q-Learning算法取决于问题的特性和需求。若问题具有全局搜索需求,则模拟退火算法可能更适合;若问题可以建模为强化学习问题,则Q-Learning可能是更好的选择。以Q-Leaning对问题进行实践求解,得到工业产品质检调度甘特图,可为实际工业产品质检排程提供参考。

基于ARM-Linux的爬壁机器人控制器研究

针对人工船舶除锈的效率低、危险性高、成本高等问题,研制出爬壁机器人进行船舶除锈。采用了上、下位机的分布式控制方案,对基于ARM-Linux的爬壁机器人控制器进行研究,包括遥感器、比较器及其组成电路。通过引用强化学习算法Q-Leaning算法,实现了爬壁机器人的强化学习循迹,改进了传统PID等算法无法针对环境进行最优化动作策略选择的缺点,提高爬壁机器人在不同环境下循迹的准确性。实验结果表明,基于ARM-Linux的爬壁机器人控制系统性能较好,可以满足控制器要求。

Implementation of AUV local planning in strong sea flow field based on Q-learning

This article investigated the implementation of AUV local planning under the strong sea flow field by integrating Q learning with fuzzy logic method.The dynamics of AUV under the sea flow was analyzed in detail,and thus fuzzy logic behaviors were defined,including a fuzzy behavior which was defined to resist the sea flow by giving an extra angle towards sea flow.This behavior was complemented by two other behaviors,the moving-to-goal behavior and collision avoiding behavior.The recommendations of these three behaviors were integrated through adjustable weighting factors to generate the final motion command for the AUV.And Q-learning was used to adjust the peak point of fuzzy membership function to increase adaptability.Simulation results showed that it improves the adaptability of AUV under different sea flow greatly.