Neural network modeling and dynamic behavior prediction of nonlinear dynamic systems

In practical engineering, it is difficult to establish complex nonlinear dynamic equations based on theories of mechanics. Data-driven models are built using neural networks in this paper meet the needs high dimension, multi-scale and precision. We construct a two-coefficient loss function for whole data-driven modeling substructure according linear multi-step method. The forward Euler method combined with trained predict five-degree-of-freedom duffing oscillator system. Comparative results show that prediction accuracy higher than modeling, generalization robustness model verified. Meanwhile, selection training data number hidden layers have great impact ability. Adopting an adjustable learning rate, adding control parameters network input shows better performance not input.