2026-10-14T17:31:50Z https://www.scidb.cn/oai

10.57760/sciencedb.43517 2026-07-24T17:40:45Z

2026-07-24 An industrial signal generation method based on a lightwei article linked dataght conditional diffusion model article linked data doi:10.57760/sciencedb.43517 en This dataset contains the associated data of the article entitled “An industrial signal generation method based on a lightwei article linked dataght conditional diffusion model.” It is provided to support the main experimental results and research conclusions regarding the performance of LC-DiffAE on the XJTU-SY dataset and the three-axis bearing vibration dataset. The dataset consists of eight files. File 1 contains the overall model framework, the loss function optimization process, and related figures. File 2 contains the experimental data acquisition platform, time- and frequency-domain analysis results of real signals, experimental settings, parameter comparison tables, and descriptions of the evaluation metrics. File 3 contains time- and frequency-domain analysis results of the generated signals, distribution similarity evaluation results, and comparison figures and data tables for the distribution overlap between real samples and samples generated by different models. File 4 contains t-SNE and PCA dimensionality reduction visualizations, as well as quantitative evaluations of generated sample diversity. File 5 contains fault diagnosis confusion matrices, diagnostic performance comparison results, and comparisons of the computational costs and generation performance of different models. Files 6 and 7 contain the ablation study results and parameter sensitivity analysis results, respectively. File 8 contains quality evaluation results for generated industrial signals of different types, including vibration, acoustic, temperature, and current signals. generative models; neural network; diffusion model; autoencoder; contrastive learning Rong Fei Xiangxin Li Anqi Shangguan Xinhong Hei Xiaogang Song Yajun Liu EMBARGO https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank