A Study on Welding Penetration Depth Detection Based on Battery Pack Tab Samples

This article mainly studies the research on the fusion depth detection of the electrode tabs of the battery module. Using the COMSOL simulation tool, a simplified model of laser deep fusion welding samples was established. Through simulation, the experimental data can be better collected by rotating the sensor once above the weld seam to obtain the collected data. For samples with virtual welds, the collected signals can be used to determine the virtual weld position through position calibration. For the measurement of fusion depth, it can be regarded as a time series classification (TS) problem. We constructed a DTW+CNN model to distinguish and classify the fusion depths, and trained the model using the data collected from labeled samples to judge the fusion depths of unlabeled samples. The accuracy of the fusion depth judgment can be measured through the DTW algorithm, so that the model can be further improved and have higher accuracy.