UQfault dataset: uncertainty of model based fault detection dataset based on a multilevel converter

This study is utilized for submodule open-circuit fault detection uncertainty analysis of modular multilevel converters. The dataset consists of 8 uncertainty factors and 15 system variables under four operation scenarios. The 1000 sets of uncertainty factor samples are generated randomly as initial configuration of the system. The 15 system variables are obtained by 1000 Monte Carlo simulations. We found that there are 153 residual samples exceeded the threshold of 0.8, which indicated a high false alarm rate.The dataset is related to the article: Y. Liao, Y. Zhang, "Rethinking Model-based Fault Detection: Uncertainties, Risks, and Optimization Based on a Multilevel Converter Case Study," IEEE Transactions on Power Electronics, Accepted in 2024.

本研究旨在对模块化多电平转换器子模块的开路故障检测不确定性进行分析。该数据集包含8个不确定性因素和15个系统变量，涵盖四种操作场景。通过对不确定性因素的1000组样本进行随机生成，作为系统的初始配置。通过1000次蒙特卡洛模拟，获取了15个系统变量的数据。研究发现，有153个残差样本超过了0.8的阈值，这表明存在较高的误报率。该数据集与2024年IEEE电力电子学报上接受发表的论文《基于多电平转换器案例研究的基于模型的故障检测的再思考：不确定性、风险与优化》相关。