To enhance rotating machinery stability and overcome the limitations of passive bearings, this paper investigates an active hybrid oil lubricated bearing (AHB) controlled by a PD controller and electro-hydraulic servo valve. To address high computational costs in stability analysis, an efficient method using a multi-layer perceptron (MLP) surrogate model is proposed. A comprehensive numerical model is established, coupling servo valve dynamics, orifice throttling, and a modified Reynolds equation. The MLP model is trained to predict nonlinear film forces, enabling the orbit method to analyze control parameters' effects on dynamic characteristics. The AHB model is coupled with a rotor finite element model for unbalance response analysis, validated via a test rig. Results show that active control significantly enhances performance: static minimum film thickness increases by 780%, while dynamic primary stiffness, damping, and critical mass increase by 613%, 105%, and 380%, respectively. Experimental and simulation analyses confirm superior vibration suppression: appropriately increasing the derivative gain k d reduces the maximum vibration amplitude by 13.4 μm, corresponding to a significant decrease of 40.1% relative to the traditional passive bearing. Conversely, increasing proportional gain k p reduces system stiffness and critical speed. Under a switched control strategy, a pseudo first-order critical speed is induced, achieving a 22% reduction compared to traditional bearings. These results highlight the AHB’s potential to actively reshape system frequency responses and overcome the inherent limitations of traditional passive bearings.