Hybrid Deep Learning for Modeling Electro-Hydrostatic Actuators: A CNN-GRU Approach

Electro-hydrostatic actuators are difficult to model accurately due to their nonlinear and time-varying behavior. To address this issue, this paper proposes a hybrid deep learning-based modeling approach. First, the dynamic characteristics of the actuator are systematically analyzed, and a simulation model is developed. Experiments are conducted under various operating conditions to construct a dataset comprising multidimensional time-series data, including motor voltage, current, and rotational speed. Subsequently, a hybrid deep learning network integrating a convolutional neural network, gated recurrent unit, and attention mechanism is designed to enable high-precision predictions of both displacement and rotational speed. The proposed method is validated through comparative experiments under multiple operating conditions. Experimental results demonstrate that the proposed approach effectively models the characteristics of the actuator with high accuracy.