By building a three-dimensional finite element model of a pickup truck, the dynamic processes of a truck colliding against a rigid wall and the collision between two trucks are numerically simulated, and the corresponding collision force-displacement curves of the truck with different initial speeds are obtained. Taking the initial speed of the truck as the input and the collision force-displacement curve as the output, the BP neural network machine learning algorithm is used to establish the vehicle collision surrogate model in the full speed domain. The numerical simulation samples are divided into a training set and a test set. The surrogate model is trained with the training set and its accuracy is verified by the test set samples. The model can be used to predict the force-displacement curve of a vehicle at an arbitrary collision speed quickly, providing an instruction for its safety design.