装备预测运行核心算法库数据

针对高端装备数据样本不完备、故障样本不充分、诊断模型泛化难等特点，围绕机理模型与数据模型驱动的装备运行预测需求，形成一系列数据，用于多源混叠的深度特征提取与迁移、机理与数据联合驱动的异常检测、面向故障耦合和样本缺乏的故障诊断、基于知识图谱的故障溯源、多模态信息融合的装备状态评价与预测等方法的训练与测试。

Aiming at the challenges including incomplete data samples, insufficient fault samples, and poor generalization performance of diagnostic models for high-end equipment, a series of specialized datasets have been constructed to address the demand for equipment operational prediction driven by mechanism models and data models. These datasets are designed for the training and testing of various research methods, including multi-source mixed deep feature extraction and transfer learning, mechanism-data joint-driven anomaly detection, fault diagnosis for fault coupling and sample scarcity, fault traceability based on knowledge graphs, and equipment condition assessment and prediction based on multi-modal information fusion.

搜集汇总

背景与挑战

背景概述

该数据集是哈尔滨工业大学创建的国家重点研发计划项目成果，专注于高端装备的预测运行算法开发，数据量251.39MB，包含13个文件。它针对数据样本不完备和故障样本不充分等挑战，提供用于深度特征提取、异常检测、故障诊断、故障溯源和趋势预测的训练与测试数据，支持机理与数据联合驱动的装备状态分析。