Run-to-failure vibration dataset of self-aligning double-row ball bearings - PART 1

Data sourced from actual operational systems constitutes a pivotal asset for scholarly investigations in the domain of machine diagnostics and prognostics. The exigency for such data has witnessed a notable surge in recent times, primarily propelled by the burgeoning interest in prognostic methodologies and the advancement of artificial intelligence (AI) technologies tailored for predictive maintenance applications. When harnessed for fault detection and prognostic endeavors, data must inherently possess the capacity to furnish insights into the degradation phenomena inherent to machinery. Moreover, a fundamental aim of prognostics entails the anticipation of the remaining useful life (RUL), a task necessitating substantial datasets for the application of data-driven techniques or the validation of physics-based models. Bearings, being subjected to a diverse spectrum of loads and fatigue stresses, carry the potential for catastrophic failure, thereby impacting the operational integrity of entire machinery systems or industrial plants. The Department of Engineering at the University of Ferrara has undertaken an extensive experimental campaign aimed at documenting the temporal evolution of vibration signals over the lifecycle of self-aligning double row rolling element bearings. Six accelerated run-to-failure trials were conducted, during which acceleration signals were continuously captured utilizing a uniaxial accelerometer. Concurrently, a radial load was imposed on the bearing housing and regulated via a load cell. Ensuring consistency, the shaft speed was maintained at a constant level, facilitated by an electric motor actuated by an inverter. The resultant dataset encapsulates acceleration signals along the radial axis spanning the entirety of the experimental tests, thereby offering a valuable resource for both scholarly inquiry and industrial applications alike. This dataset provides the Part 1 of data collected during the experimental campaign. Data is provided in .mat format and each file contains two variables: the acceleration signal in radial direction and the sampling frequency.

源自实际运行系统的数据，是机械诊断与故障预测（Prognostics）领域学术研究的核心资产。近年来，此类数据的迫切需求显著攀升，主要得益于故障预测方法的日益受关注，以及面向预测性维护应用定制开发的人工智能（AI）技术的进步。若将数据用于故障检测与故障预测相关研究，则其本应具备揭示机械设备固有退化现象的能力。此外，故障预测的核心目标在于预估剩余使用寿命（Remaining Useful Life, RUL），开展此类任务需要大规模数据集，以支撑数据驱动技术的应用或物理机理模型的验证。轴承承受多样载荷与疲劳应力，存在发生灾难性失效的风险，进而影响整个机械系统或工业厂房的运行可靠性。费拉拉大学（University of Ferrara）工程系开展了一项大规模实验研究，旨在记录调心双列滚动轴承全生命周期内振动信号的时域演化特性。实验共开展6次加速失效试验，试验期间采用单轴加速度计持续采集加速度信号；同时，通过载荷传感器向轴承座施加径向载荷并进行调节。为保证实验一致性，采用变频器驱动的电动机将轴转速维持在恒定水平。最终获取的数据集涵盖全部实验测试中沿径向轴采集的加速度信号，可为学术研究与工业应用提供宝贵资源。本数据集为本次实验研究采集的第1部分数据，以.mat格式存储，每个文件包含两个变量：径向加速度信号与采样频率。

搜集汇总

背景与挑战

背景概述

该数据集包含自对准双列球轴承在恒定速度和负载条件下运行至失效的振动信号，适用于机器诊断和预测性维护研究。数据以.mat格式提供，包含径向加速度信号和采样频率，由费拉拉大学工程学院通过实验收集。