Along with the technological advance of the industrial sector, the boom in process control systems has increased in recent years; however, the models employed in the industry are complex and non-linear, and conventional methods of modeling, identification and implementation of controllers are made from linear models in specific regions of operation. This linearization restricts the results and efficient data collection of the system, because it rejects useful information to optimize the processes. An inverted rotary pendulum is one of the most studied platforms in system identification and modeling theory, due to its known instability and complexity of its analytical model. This article shows the modeling and nonlinear identification of an inverted rotary pendulum of the manufacturer Quanser, along with its simulation, as part of the process of development of nonlinear controllers on complex plants, being the identified model the starting point for the design of these controllers, as a result of the process, an identified system is obtained for the simulated and real model with a low mean square error in its output signals process corresponding to pole and arm acceleration.