2026-10-13T22:10:25Z https://www.scidb.cn/oai

10.57760/sciencedb.012pv 2026-09-23T19:49:38Z

2026-09-23 Dataset and Code for Load-Signal-Based Detection of Current-Carrying Stability in Current-Carrying Friction Pairs doi:10.57760/sciencedb.012pv en This dataset is derived from in-situ current carrying friction tests of 7075Al/CuCrZr current carrying friction pairs, used to study the correlation between mechanical load response and current carrying stability state, and to conduct research on current carrying stability classification based on load signals. The experiment was conducted using an in-situ current carrying friction testing device, which synchronously collected signals such as x-direction load, z-direction normal load, voltage, and current under different normal load, wear trajectory radius, and current conditions. Calculate the dynamic resistance based on voltage and current, and construct a current carrying stable state label based on the stability characterization parameters of the dynamic resistance. Label 0 represents the current carrying stable state, and label 1 represents the current carrying unstable state. The effective experimental data is segmented using a 50 second time window and a 5 second sliding step, and the load signal is subjected to short-time Fourier transform to generate the corresponding load time-frequency graph. The dataset contains a total of 9392 sets of samples, including 7647 sets of stable current carrying samples and 1745 sets of unstable current carrying samples. The data is organized in the original experimental sequence and can be used for load current state correlation analysis, load time-frequency characteristic analysis, and neural network classification research. Simultaneously provide code related to data processing and model analysis, including tag reading, time-frequency graph data organization, typical neural network classification, and non overlapping sample supplementation verification. To investigate the impact of overlapping adjacent sliding window original signal segments on classification results, non overlapping samples with time intervals greater than the length of the original analysis window were used for supplementary validation, and evaluation was conducted based on category weighted loss, balanced accuracy, recall, and F1 score. This dataset and code can provide a data and reproduction basis for the indirect characterization of electrical states through the correlation of multiple physical information related to current carrying friction and mechanical response.  current-carrying; friction pairs; load signal Li Chaofan PUBLIC https://mit-license.org dataset Science Data Bank