A fast method based on deep learning for predicting the impact load of vehicle-mounted howitzer cab

In the topology optimization design and strength check of a vehicle-mounted howitzer cab, it is necessary to obtain plenty of impact load laws under different firing conditions. How to obtain the impact load on the cab quickly is one of the challenges in the design of a vehicle-mounted howitzer. In this paper, the deep learning (DL) method is introduced to solve the impact load on the cab. A fast prediction method for cab impact load based on a ConvLSTM multidimensional feature neural network is proposed. The calculation of the impact load on the cab under different operating conditions is achieved, and the solving speed is close to the real-time level. The numerical examples show that the accuracy of the DL model is comparable to that of traditional computational fluid dynamics (CFD) simulation, but the solving time is on the millisecond level. The computational efficiency has been greatly improved, with the potential for offline training and online computing. When there is a slight change in the morphology of the cab, the proposed model remains applicable. The results can quickly provide load conditions for cab strength checking and topology optimization, help to shorten the development period of a vehicle-mounted howitzer, and lay the foundation for the construction of a digital twin model of a vehicle-mounted howitzer system.