An online data-driven approach for performance prediction of electro-hydrostatic actuator with thermal-hydraulic modeling

The Electro-Hydrostatic Actuator (EHA) plays an essential part in power-by-wire (PBW) systems due to its compact volume and high power density ratio. However, it is fairly usual for the performance of a highly integrated EHA to be adversely affected by heat dissipation. In this paper, taking into account the effect of physical heat characteristics, thermal network model is created to depict the heat dissipation of an EHA system. A dynamic performance degradation model is enhanced to appropriately evaluate the performance of the EHA system. A novel real-time corrected thermal network model based on artificial neural network (RCTN-ANN) is developed, the key idea of the proposed model is to correct parameters by using trained RCTN-ANN model and online data, and simulate the performance deterioration of online EHA, which can then be used for prognostics and health management (PHM) of EHA under actual working conditions. Validated using actual EHA experiment, the results show that the proposed method provides an accurate performance prediction with dynamic data, which is significant for the real-time PHM of the EHA system.