After testing the new approach in combination with the 1D-CNN machine learning technique under different conditions, they found an accuracy of more than 99%.