Attitude tracking of 2-DOF aero system with hierarchical sliding mode control under affine and exogenous disturbance

Abstract

Dynamic coupling in a 2 DOF flight control system leads to nonlinearity, affine (matched), and exogenous (mismatched) disturbances, which cause instability during attitude control of the system. Although extended state disturbance observer-based Sliding Mode Controllers often provide approximate estimates of the disturbances, the finite time convergence is compromised with increased complexity. To address this issue, this paper designs and implements a unique hierarchical sliding mode control (HSMC) that decouples the nonlinear system into two-stage subsystems consisting of linear position and nonlinear velocity control systems. Furthermore, the hierarchical control structures allow us to generate the desired nonlinear input angular velocities for the linear position control system while minimizing the effect of exogenous disturbance using a proportional-integral (PI) sliding surface. The input control voltages are generated using a proportional-integral-integral (PII) sliding surface, which tracks the previously generated desired input angular velocity while mitigating the affine disturbance. Closed-loop stability is ensured via Lyapunov-based analysis, whereas the controller parameters are optimally tuned using a genetic algorithm, resulting in enhanced trajectory tracking and reduced error. Extensive evaluations demonstrate the superiority of HSMC over Fuzzy Logic Control (FLC), Backstepping Control (BSC), Sliding Mode Control (SMC), and Non-Singular Terminal Sliding Mode Control (NSTSMC). For pitch dynamics, HSMC achieves a settling time of $1.5s$ (vs $2.5-4s$ ), overshoot of $4\%$ (up to $33\%$ for competing controllers), and steady-state error of $0.{04}^{°}$ (vs $0.1-0.3^{°}$ ). For yaw, it attains $4.8s$ settling time, $13\%$ overshoot, and $0.{05}^{°}$ steady-state error. In trajectory tracking, HSMC reduces IAE, ISE, ITAE, and ITSE by $40-85\%$ , with the lowest ITAE $(200-1758)$ for pitch and $(129-1878)$ for yaw, compared to competing approaches $(497-5035)$ , $(346-7472)$ respectively demonstrating its robustness, efficiency, and real-time suitability.