Active Sensing Signal and Wavelet Packet Energy Dataset for In-situ Damage Detection of Metallic Shear Plate Dampers

This dataset contains active sensing signals and extracted wavelet packet energy features for in-situ damage detection of metallic shear plate dampers (MSPDs). The data were collected from laboratory experiments using a piezoelectric active sensing system. For each specimen, stress wave signals were acquired under different excitation signals (Signal0–Signal5), group configurations (ao0–ao2), and sensing channels (S1–S3). The raw signals were originally recorded in NI TDMS format and subsequently converted into MATLAB-compatible TDMS files for batch processing. Each signal underwent band-pass filtering (1 kHz – 100 kHz), followed by wavelet packet decomposition (Daubechies 5, level 8). The energy of each sub-band was computed to form a wavelet packet energy feature vector. These feature vectors were used to construct supervised learning datasets for damage classification. The dataset includes: Converted TDMS signal files Time-domain signals Frequency-domain representations Wavelet packet energy feature matrices Corresponding damage labels Damage states are categorized into four classes: 0 – Healthy1 – Damage progression2 – Web buckling3 – Failure The final dataset is organized as feature matrices (N × D) and label vectors (N × 1), suitable for machine learning applications such as KNN classification and cross-validation evaluation. This dataset supports the study: “In-situ Damage Detection Method for Metallic Shear Plate Dampers Based on the Active Sensing Method and Machine Learning Algorithms.” It is intended for research purposes in structural health monitoring, damage detection, and intelligent sensing systems.本数据集包含用于金属剪切板阻尼器（metallic shear plate dampers, MSPDs）原位损伤检测的主动传感信号与提取得到的小波包能量特征。 该数据源自压电主动传感系统的室内实验室试验。针对每个试件，采集了不同激励信号（Signal0至Signal5）、组配方案（ao0至ao2）以及传感通道（S1至S3）下的应力波信号。原始信号最初以NI TDMS格式存储，后续转换为兼容MATLAB的TDMS文件以进行批量处理。 每一条信号均经过1 kHz至100 kHz的带通滤波处理，随后进行小波包分解（采用多贝西5（Daubechies 5）小波，分解层数为8）。计算各子频带的能量以构建小波包能量特征向量，这些特征向量被用于构建用于损伤分类的监督学习数据集。 本数据集包含： 转换后的TDMS信号文件 时域信号 频域表征 小波包能量特征矩阵 对应损伤标签 损伤状态共分为四类： 0 – 健康状态 1 – 损伤演化状态 2 – 腹板屈曲状态 3 – 失效状态 最终数据集以特征矩阵（N × D）与标签向量（N × 1）的形式组织，适用于KNN分类、交叉验证评估等机器学习应用场景。 本数据集支撑的研究为：《基于主动传感方法与机器学习算法的金属剪切板阻尼器原位损伤检测方法》。 该数据集旨在服务于结构健康监测、损伤检测以及智能传感系统领域的相关研究工作。