The main focus is nonlinear model-based dynamic positioning (DP) control system design. A nonlinear uniform global exponential stability (UGES) observer produces noise-free estimates of the position, the slowly varyin...The main focus is nonlinear model-based dynamic positioning (DP) control system design. A nonlinear uniform global exponential stability (UGES) observer produces noise-free estimates of the position, the slowly varying environmental disturbances and the velocity, which are used in a proportional-derivative (PD) + feedforward control law. The stability of this observer-controller system is proved by introducing a specific nonlinear cascaded system. The simulation results have successfully demonstrated the performance of designed DP control system.展开更多
现有的烟火检测方法主要依赖员工现场巡视,效率低且实时性差,因此,提出一种基于YOLOv5s的复杂场景下的高效烟火检测算法YOLOv5s-MRD(YOLOv5s-MPDIoU-RevCol-Dyhead)。首先,采用MPDIoU(Maximized Position-Dependent Intersection over U...现有的烟火检测方法主要依赖员工现场巡视,效率低且实时性差,因此,提出一种基于YOLOv5s的复杂场景下的高效烟火检测算法YOLOv5s-MRD(YOLOv5s-MPDIoU-RevCol-Dyhead)。首先,采用MPDIoU(Maximized Position-Dependent Intersection over Union)方法改进边框损失函数,以适应重叠或非重叠的边界框回归(BBR),从而提高BBR的准确性和效率;其次,利用可逆柱状结构RevCol(Reversible Column)网络模型思想重构YOLOv5s模型的主干网络,使它具有多柱状网络架构,并在模型的不同层之间加入可逆链接,从而最大限度地保持特征信息以提高网络的特征提取能力;最后,引入Dynamic head检测头,以统一尺度感知、空间感知和任务感知,从而在不额外增加计算开销的条件下显著提高目标检测头的准确性和有效性。实验结果表明:在DFS(Data of Fire and Smoke)数据集上,与原始YOLOv5s算法相比,所提算法的平均精度均值(mAP@0.5)提升了9.3%,预测准确率提升了6.6%,召回率提升了13.8%。可见,所提算法能满足当前烟火检测应用场景的要求。展开更多
文摘The main focus is nonlinear model-based dynamic positioning (DP) control system design. A nonlinear uniform global exponential stability (UGES) observer produces noise-free estimates of the position, the slowly varying environmental disturbances and the velocity, which are used in a proportional-derivative (PD) + feedforward control law. The stability of this observer-controller system is proved by introducing a specific nonlinear cascaded system. The simulation results have successfully demonstrated the performance of designed DP control system.
文摘现有的烟火检测方法主要依赖员工现场巡视,效率低且实时性差,因此,提出一种基于YOLOv5s的复杂场景下的高效烟火检测算法YOLOv5s-MRD(YOLOv5s-MPDIoU-RevCol-Dyhead)。首先,采用MPDIoU(Maximized Position-Dependent Intersection over Union)方法改进边框损失函数,以适应重叠或非重叠的边界框回归(BBR),从而提高BBR的准确性和效率;其次,利用可逆柱状结构RevCol(Reversible Column)网络模型思想重构YOLOv5s模型的主干网络,使它具有多柱状网络架构,并在模型的不同层之间加入可逆链接,从而最大限度地保持特征信息以提高网络的特征提取能力;最后,引入Dynamic head检测头,以统一尺度感知、空间感知和任务感知,从而在不额外增加计算开销的条件下显著提高目标检测头的准确性和有效性。实验结果表明:在DFS(Data of Fire and Smoke)数据集上,与原始YOLOv5s算法相比,所提算法的平均精度均值(mAP@0.5)提升了9.3%,预测准确率提升了6.6%,召回率提升了13.8%。可见,所提算法能满足当前烟火检测应用场景的要求。
