hubtru/nonastreda
收藏资源简介:
Nonastreda是一个多模态数据集,用于铣削加工中高效的刀具磨损状态监测。它包含512个样本级记录,结合了视觉、时频和力信号表示的刀具磨损信息。该数据集集成了视觉检测数据、力信号衍生的图像表示和原始力信号,用于刀具磨损监测。每个样本通过一个标识符(如T10R10B1)进行索引,并且相同的标识符在可用的图像模态中使用。每个样本级行包括:从样本标识符解析出的tool_id和run_id、一个基于显微镜图像的刀具磨损分类标签、一个从力信号振幅衍生的附加力相位分类标签、三个回归目标和九个图像模态。原始力信号文件也包括在内,用于未来研究,但未用作样本级可视化表,因为序列是按刀具组织的,而不是为每个刀片/图像样本切割成子序列。
Nonastreda is a multimodal dataset for efficient tool wear state monitoring in milling. It contains 512 sample-level records combining visual, time-frequency, and force-signal-derived representations of tool wear. The dataset combines visual inspection data, force-signal-derived image representations, and raw force signals for tool wear monitoring. Each sample is indexed by an identifier such as T10R10B1, and the same identifier is used across the available image modalities. Each sample-level row includes: tool_id and run_id parsed from the sample identifier, one microscope-image-based tool-wear classification label, one additional force-phase classification label derived from force-signal amplitudes, three regression targets, and nine image modalities. The raw force-signal file is also included for future research, but it is not used as a sample-level visualization table because the sequences are organized per tool rather than cut into sub-sequences for each blade/image sample.




