This repository contains datasets used for analyzing low-level argon and air impurity in helium. These datasets were utilized in the research paper titled "Impurity Gas Detection for SNF Canisters Using Probabilistic Deep Learning and Acoustic Sensing" available at (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4675917). Files 1. Input_set.npy: This file contains the input dataset, consisting of excitation and gas concentration pairs.2. Response_set.npy: This file contains the response dataset.3. Concen_set.npy: This file contains the concentration pairs of Argon and Air used in the experiments.4. Image_train.npy: This file contains the STFT spectrum data for the training set.5. Image_test.npy:This file contains the STFT spectrum data for the testing set. Data Structure Image_train.npy- Shape:(65,011, 38, 44, 2)- Description: This array contains the STFT spectra for the training set. - The first dimension (65,011) represents the number of training samples. - The second and third dimensions (38, 44) represent the size of the STFT spectrum. - The fourth dimension (2) represents the two channels: - One channel contains the STFT of the excitation signal. - The other channel contains the STFT of the response signal. Image_test.npy- Shape: (27,862, 38, 44, 2)- Description: This array contains the STFT spectra for the testing set. - The first dimension (27,862) represents the number of testing samples. - The second and third dimensions (38, 44) represent the size of the STFT spectrum. - The fourth dimension (2) represents the two channels: - One channel contains the STFT of the excitation signal. - The other channel contains the STFT of the response signal. Input_set.npy- Shape: (101,919, 253)- Description: This array contains the input parameters for the ultrasonic experiments. - The first 251 columns contain the excitation parameters. - The last 2 columns represent the concentration pairs of Argon and Air. Response_set.npy- Shape: (101,919, 426)- Description: This array contains the ultrasonic response data corresponding to the input parameters in `Input_set.npy`. - The 426 columns represent the time-series data points of the ultrasonic response. Concen_set.npy- Shape: Corresponds to the concentration pairs of Argon and Air.- Description: This array contains the concentration pairs used in the experiments. X_pure_new.npyShape: (4,718, 427)Description: This array contains data representing pure helium samples.The first 426 columns represent the response features of each sample.The 427th column is the label for binary classification, and all values in this column are zeros, indicating that these are pure samples. X_with_impurity_new.npyShape: (101,700, 429)Description: This array contains data representing samples with impurities.The first 426 columns represent the response features of each sample.The 427th column is the label for binary classification, and all values in this column are ones, indicating that these are impure samples.The last two columns represent the concentrations of impurities in the samples. CitationIf you use this dataset in your research, please cite the above paper. ContactFor any questions or issues related to this dataset, please contact the repository owner via bozhouzh@usc.edu.