Testing technology is an elementary component of industrial production. The testing phase ensures that statutory standards are met, and the product quality corresponds to the agreement with the producer. For the production of electric vehicles, new innovative production and test processes must be introduced outside existing structures. The high-voltage battery (HVB) of an electric vehicle is examined for possible leaks as part of the leak test and, if found, forwarded for leak location and sealing. To ensure a high level of safety, the volume flow of the leakage must not exceed a specified limit value. To this end, a method will be developed in which carbon dioxide gas leaks from a pressurized HVB recorded with a highly specialized camera are quantified by a neural network (NN) measuring the volumetric flow rate based on the image or video recordings. The experimental setup was optimized to produce a high-quality data set, and subsequently, a convolutional neural network and a recurrent neural network were trained for a regression task. Both NNs’ resulting accuracy of leakage quantification was satisfactory but not yet sufficient to replace the currently used method of volume flow measurement in combination with leak location. However, further research in this area offers high potential to achieve the intended time savings in leakage testing.     «

Testing technology is an elementary component of industrial production. The testing phase ensures that statutory standards are met, and the product quality corresponds to the agreement with the producer. For the production of electric vehicles, new innovative production and test processes must be introduced outside existing structures. The high-voltage battery (HVB) of an electric vehicle is examined for possible leaks as part of the leak test and, if found, forwarded for leak location and seal...     »