The inverted pendulum is one of the most studied systems in automatic control, serving as a testbed for new control strategies. One of its variants is the planar inverted pendulum, which consists of a pendulum articulated to a mobile base that moves in a horizontal plane. Although this system has been widely studied, the models employed are simplified and a formulation that captures its full dynamics is still lacking. Moreover, the impact of these simplifications has not yet been assessed. In this research, using the bond graph methodology, a complete and energetically coherent dynamic model of the planar inverted pendulum is developed. No previous work provides a dynamic model with this level of detail. This model was validated by comparing it with a Simulink reference model. Eight simplified models were built and compared using open‐loop and closed‐loop simulations with a hybrid nonlinear control strategy. The results show that most simplified models exhibit closed‐loop responses that closely match those of the full dynamics model, whereas under open‐loop conditions significant discrepancies arise in some cases. The study also demonstrates that controllers designed using simplified models can achieve performance comparable to that obtained with a controller designed from the full dynamics model. These findings provide strong support for the widespread use of simplified models for controller design, analysis, and simulation of the planar inverted pendulum and similar systems.
This study analyzes the impact of modeling simplifications on the dynamics and control performance of the Furuta pendulum. A complete multibond graph model is developed, incorporating full system dynamics, and providing an energetically consistent and modular framework. The model is validated against a Simulink-Simscape reference, achieving a maximum normalized root mean squared error (NRMSE) of 0.2035 × 10–3. The commonly used simplified model, which neglects secondary inertias, is compared with full-dynamics models under open- and closed-loop conditions. Open-loop results show appreciable discrepancies, with NRMSE values increasing notably for both θ1 and θ2 as the inertias are progressively increased. A nonlinear control scheme based on energy shaping, collocated partial feedback linearization, and linear quadratic regulation (LQR) is designed using the simplified model. In closed loop, the controller achieves swing-up and stabilization in all cases within 10 seconds. For nominal and moderately increased inertias, performance degradation is minimal, with settling times around 7.4 s and control effort between 0.453 and 0.475 N2m2s. However, for large inertias, the settling time increases to 8.91 s, control effort rises to 0.655 N2m2s, and an additional oscillation is required. These results show that simplified models are suitable for control design under typical conditions, while full-dynamics models are essential for validation and robustness assessment.