Neural Networks For Fluid Power Systems
Artificial Intelligence (AI) is gaining significant traction across various fields, including the modelling of complex systems. Traditional, highly detailed mathematical models of such systems often demand excessive computational resources, especially when implemented as submodels within larger simulations. This limitation commonly leads to model simplifications, potentially affecting accuracy. While most neural networks do not incorporate conservation laws, the use of a Lagrangian Neural Network (LNN) is proposed, which can parametrize arbitrary Lagrangians using neural networks. However, Lagrangian Neural Networks have predominantly been applied in mechanical systems, with no prior applications in hydraulics. This paper introduces and explores the use of an LNN in hydraulic system modelling. This approach is applied to a simple hydraulic cylinder as a proof of concept. The results obtained from the LNN model are compared with data from the experimental set-up to evaluate accuracy, showing great performance. Additionally, we analyze the computational efficiency by comparing the simulation times of the LNN model and the Simulink simulation.