Surface pressure prediction of convolutional vehicle based on machine learning algorithm

Surface pressure is an important index in the evaluation of convolutional vehicle attitude control and motion characteristics. In order to determine the full-domain surface pressure distribution of the vehicle during navigation, a surface pressure reconstruction algorithm based on machine learning is proposed. The surface pressure distribution of convolutional vehicle may vary due to different sailing environment. By arranging pressure observation points on the surface of convolutional vehicle and obtain the distribution of these pressure, the corresponding navigational conditions can be characterized. In this paper, the pressure obtained from a finite number of observation points on the surface, as well as their coordinates, are used as input information of model. Then we can obtain a mapping model from discrete pressure distribution to full domain pressure distribution. To investigate the performance of the model, full-domain prediction experiments of surface pressure are conducted on several different test data. The results demonstrate that our machine learning-based model can achieve high-precision surface pressure reconstruction, and the relative error of the predicted value can be reduced to within 10%.