The Difference of Simple Neural Networks in Testing Penetration Speed

Aiming at the difference of different networks in the penetration problem, the paper starts from the precision, time and so on. Comparison of deep neural networks (DNN), Decision Trees, Ran Dom Forests, XGBoost, and Support vector machine (SVM), shows that different networks are different in predicting the ballistic limit of ball penetration. The test results show that the predictive value of the decision tree is 0.01 higher than that of the depth neural network (DNN), but the use time is much longer than that of DNN. The predictions were 0.05 higher when compared with a deep neural network (DNN) using Random Forests, and the predictions were not evenly distributed but the usage time was much lower. When comparing SVM with DNN, the predicted value is 0.0.9 higher, but the Support vector machine time is much less, and the predicted value distribution is even and curvilinear. The predicted values were found to be 0.09 higher when compared with deep neural network (DNN) using XGBoost, but were much lower with the use of specimens (DNN) and the predicted values were unevenly distributed.