米特交流故障电弧检测算法训练测试数据集

故障电弧数据体系整合常见家用电器在正常运行（如微波炉功率切换等稳态 / 暂态特征）与人工模拟故障场景下的全周期电流电压动态采样数据，所有故障电弧信号均通过可控电弧发生器手动生成。数据采集包含电热（电热水器）、充电设备（充电器）等多类负载在人工诱发故障时的电弧信号，同步记录正常运行基线（如设备无故障时的平滑电流曲线）与故障波形（电弧特有的高频谐波、电流过零畸变）。构建可复现的标准化电弧数据集。该方法通过人工控制电弧发生的位置、持续时间及能量等级，确保数据的一致性与场景可控性，为电弧检测算法提供纯净的故障样本（区分真实电弧与电器正常开关瞬态），支撑模型在不同故障模式下的泛化能力训练，同时保留设备固有电气指纹，助力复杂工况下的电弧精准识别与干扰抑制。

The arc fault dataset integrates full-cycle dynamic sampling data of current and voltage collected from common household electrical appliances under both normal operation (e.g., power switching of microwave ovens, which exhibit steady-state and transient characteristics) and artificially simulated fault scenarios. All fault arc signals are manually generated using a controllable arc generator. The dataset covers arc signals from multiple load types including electrothermal appliances (electric water heaters) and charging equipment (chargers) under artificially induced faults, while simultaneously recording normal operation baselines (e.g., smooth current curves when the equipment is fault-free) and fault waveforms such as arc-specific high-frequency harmonics and current zero-crossing distortions. This work constructs a reproducible and standardized arc fault dataset. By manually controlling the arc generation position, duration and energy level, the method ensures data consistency and scenario controllability. It provides purified fault samples for arc detection algorithms to distinguish real arcs from normal switching transients of electrical appliances, supports the training of model generalization capability across different fault modes, retains the inherent electrical fingerprints of equipment, and facilitates accurate arc identification and interference suppression under complex working conditions.

搜集汇总

背景与挑战

背景概述

米特交流故障电弧检测算法训练测试数据集是一个用于故障电弧检测的数据集，包含家用电器在正常和故障状态下的电流电压数据，适用于信号处理和特征提取等技术领域。