The availability of off-highway machines is gaining in importance, particularly with respect to seasonal use, such as agriculture. For example, machine failures can lead to considerable income losses during harvesting. At KIT, a method for the development of real-time capable models for the determination of the component loads of off-highway machines was developed. The strength calculation of the piston of an axial piston pump serves as an example of use. The stresses in the component are determined by a multi-body simulation coupled with a structure simulation, and are then used to create an artificial neural network. The quality of the generated models is shown on the basis of the criteria accuracy, reproducibility and computation time.