融合学习策略与导向果蝇机制的气味源主动定位方法研究

工业生产过程中常发生由有害气体泄漏引起的火灾或爆炸事故,利用载有气体传感器的移动机器人实时监测并搜索定位泄漏气体源是预防重大事故的有效方法,而高效的搜索策略是保证机器人快速准确定位气味源的关键因素.现有的气味源搜索算法存在定位成功率不高和对气味源定位不准的问题,本文提出一种将仿生果蝇算法和学习策略相融合的气味搜索策略.针对传统果蝇算法易陷入饱和收敛的问题,提出一种新的导向果蝇极值更新方式;针对寻优不精的问题,进一步提出一种基于学习策略的导向果蝇气味源搜索算法(OCGFOA).仿真实验结果表明OCGFOA算法完成定位速度更快且离泄漏气味源位置更近,其定位效果更能满足对危险气味源定位的要求;最后,在物理场景下进行气味源主动定位验证实验,证明本文所提算法在实际场景下也具有可行性.

Study on active odor source localization method based on learning strategy and guided fruit fly mechanism

During the process of industrial production, fire or explosion accidents caused by harmful gas leakage occur frequently. Using mobile robots with gas sensors to real-time monitor and locate the source of leaking gas is an effective way to prevent major accidents, and an efficient search strategy is a key factor to ensure that the robot can quickly and accurately locate the source gases. The existing algorithm to search the source gases has the problems of low positioning success rate and inaccurate positioning of the source gases, and this paper proposes a gas search strategy that combines the fruit fly optimization algorithm (FOA) and the learning strategy to improve the success rate and accuracy of robot positioning. Aiming at the problem that the traditional fruit fly algorithm is easy to fall into saturation convergence, a new guide to the extreme value update method of fruit fly is proposed. Aiming at the problem of poor optimization, the opposite learning and the Cauchy distribution random are added, and a novel oriented fruit fly odor source search algorithm based on the learning strategy oriented center guided FOA (OCGFOA) is proposed. The comparative experimental results show that the OCGFOA algorithm completes the task of locating the source gases faster and is closer to the location of the leaking source gases, which shows that its positioning effect can better meet the requirements for the positioning of dangerous source gases. Finally, we performed a verification experiment in a actual scenarios, and it is proved that the proposed algorithm is feasible in practical scenarios.