Synthesis of a neural regulator and experimental stability study for an adaptive quadcopter control system.
DOI:
https://doi.org/10.15276/opu.2.70.2024.12Keywords:
UAV, dynamics equation, Euler-Lagrange equation, high-frequency jitter effect, neural network approximation, sigmoidal continuous activation function, Lyapunov function, MIMO (Multi Input Multi Output), RBF (Radial basis function), SMC (Sliding mode control)Abstract
When solving a number of complex manipulation tasks, it is advisable to take into account the nonlinear dynamics of the control object. Such tasks include, in particular, the control of large space-based manipulators, as well as ground-based manipulation systems used in construction and in the aftermath of accidents and disasters. For such manipulation systems, the control task is complicated by the fact that the dynamics of the controlled structure is very complex and, in most cases, cannot be mathematically described. In this regard, methods based on solving the inverse dynamics problem cannot always be applied. The use of PID controllers, which are widely used in most industrial applications, also allows us to take into account the peculiarities of the motion dynamics of such systems. There are also problems with ensuring stability, including under the influence of external factors that are not known in advance. A new direction in this area is related to the use of neural networks, which can estimate the dynamics of the system in real time. On the other hand, the use of sliding modes in control systems ensures the independence of the control process from both external influences and parametric disturbances. The combination of these methods allows to create a system that can eliminate some of the disadvantages of each method. In this article, we develop a control method based on an adaptive neural network tuning algorithm. The proposed method allows controlling the system without a priori information about the structure and parameters of the dynamic model of the controlled object. Adaptive algorithms are used to determine the coefficients of the neural network controller, which allow its adjustment as a normal functioning of the system. The stability conditions of such a control system are obtained using the Lyapunov method. The effectiveness of the proposed control method is confirmed by the results of modeling the control system in MATLAB, as well as by experimental studies of robotic systems.
Downloads
References
Markazi, A.H.D., Maadani, M., Zabihifar, S., & Doost-Mohammadi, N. (2018). Adaptive fuzzy sliding mode control of under-actuated nonlinear systems. Int J Autom Comput., 15, 364–76.
Gruber, P., & Balemi, S. (2010). Overview of non-linear control methods. Swiss Society for Automatic Control, 1–33.
He, W., Ge, S.S., Li, Y., Chew, E., & Ng, Y.S. (2015). Neural Network Control of a Rehabilitation Robot by State and Output Feedback. J Intell Robot Syst., 80(1), 15–31. Retrieved from: http://link.springer.com/10.1007/s10846-014-0150-6.
Polydoros, A.S., Nalpantidis, L., & Kruger, V. (2015). Real-time deep learning of robotic manipulator inverse dynamics. 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 3442–3448. Retrieved from: http://ieeexplore.ieee.org/document/7353857/.
Muller, A. (2017). Recursive second-order inverse dynamics for serial manipulators. 2017 IEEE International Conference on Robotics and Automation (ICRA), 2483–2489. Retrieved from: http://ieeexplore.ieee.org/document/7989289/.
Duc, M.N., & Trong, T.N. (2014). Neural network structures for identification of nonlinear dynamic robotic manipulator. 2014 IEEE International Conference on Mechatronics and Automation, 1575–1580. Retrieved from: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6885935.
Rahmani, B., & Belkheiri, M. (2016). Robust Adaptive Control of Robotic Manipulators Using Neural Networks : Application to a Two Link Planar Robot. ICMIC-2016, 839–844.
He, W., Chen, Y., Yin, Z., & Member, S. (2015). Adaptive Neural Network Control of an Uncertain Robot With Full-State Constraints. IEEE Transactions on Cybernetics, 46(3), 620–629. DOI: 10.1109/TCYB.2015.2411285.
Yan, L., & Li, C.J. (1997). Robot Learning Control Based on Recurrent Neural Network Inverse Model, 14(3), 1–22.
Li, Y., Li, S., Caballero, D., Miyasaka, M., & Lewis, A. (2017). Hannaford B. Improving control precision and motion adaptiveness for surgical robot with recurrent neural network. IEEE Int Conf Intell Robot Syst, 3538–3543.
He, W., Ge, S.S., How, B.V.E., & Choo, Y.S. (2014). Dynamics and Control of Mechanical Systems in Offshore Engineering. London: Springer London. Retrieved from: http://link.springer.com/10.1007/978-1-4471-5337-5.
Hana Boudjedir, Fouad Yacef, & Omar Bouhali N.R. (2012). Dual Neural Network for Adaptive Sliding Mode Control of Quadrotor Helicopter Stabilization. Int J Inf Sci Tech, 2(4),101–15.
Moreno-Valenzuela, J., Aguilar-Avelar, C., Puga-Guzman, S.A., & Santibanez, V. (2016). Adaptive Neural Network Control for the Trajectory Tracking of the Furuta Pendulum. IEEE Trans Cybern. 1–14.
Nelson, J., & Kraft, L.G. (1994). Real-time control of an inverted pendulum system using complementary neural network and optimal techniques. In: Proceedings of 1994 American Control Conference - ACC ‘94, p. 2553–2554. Retrieved from: http://ieeexplore.ieee.org/document/735019/.
Hsu, C-F. (2014). Adaptive backstepping Elman-based neural control for unknown nonlinear systems. Neurocomputing, 136, 170–179. Retrieved from: https://www.sciencedirect.com/science/article/pii/S0925231214001684.
Boudjedir, H. (2012). Dual Neural Network for Adaptive Sliding Mode Control of Quadrotor Helicopter Stabilization. Int J Inf Sci Tech, 2(4), 1–14. Retrieved from: http://www.airccse.org/journal/IS/papers/2412ijist01.pdf.
Khazaee, M., Markazi, A.H.D., Rizi, S.T., & Seyfi, B. (2017). Adaptive fuzzy sliding mode control of input-delayed uncertain nonlinear systems through output-feedback. Nonlinear Dyn, 87(3), 1943–1956. Retrieved from: http://link.springer.com/10.1007/s11071-016-3164-8.
Li, M., Bi, D., & Xiao, Z. (2015). Mechanism Simulation and Experiment of 3-DOF Parallel Robot Based on MATLAB. 2015 Ipemec, 489–494.
Huang, Y. (2015). Neural Network Based Dynamic Trajectory Tracking of Delta Parallel Robot. Int Conf Mechatronics Autom., 1938–1943.
Du, J., & Lou, Y. (2016). Simplified Dynamic Model for Real-time Control of the Delta Parallel Robot. 1647–1652.
Naidoo, Y., Stopforth, R., & Bright, G. (2011). Helicopter Modelling and Control. Int J Adv Robot Sy., 8(4), 139–149.
Van Cuong, P., & Nan, W.Y. (2016). Adaptive trajectory tracking neural network control with robust compensator for robot manipulators. Neural Comput Appl, 27(2), 525–536. Retrieved from: http://dx.doi.org/10.1007/s00521-015-1873-4.
Yildirim, S. (2005). Design of Adaptive Robot Control System Using Recurrent Neural Network. J Intell Robot Syst, 44(3), 247–261. Retrieved from: http://link.springer.com/10.1007/s10846-005-9012-6.
Alashqar, H.A.H., & Hussein, M.T. (2013). Modeling and High Precision Motion Control of 3 DOF Parallel Delta Robot Manipulator.
Tanaka, Y., & Tsuji, T. (2004). On-line Learning of Robot Arm Impedance Using Neural Networks. 2004 IEEE International Conference on Robotics and Biomimetics, 941–946. Retrieved from: http://ieeexplore.ieee.org/document/1521911/.
Brandt, R.D., & Lin, F. (1999). Adaptive interaction and its application to neural networks. Inf Sci (Ny), 121 (3–4), 201–215. Retrieved from: https://www.sciencedirect.com/science/article/pii/S0020025599000900.
Olfati-Saber, R. (1999). Fixed point controllers and stabilization of the cart-pole system and the rotating pendulum. Proceedings of the 38th IEEE Conference on Decision and Control (Cat No99CH36304), 1174–1181. Retrieved from: http://ieeexplore.ieee.org/document/830086/.
Turker, T., Gorgun, H., & Cansever, G. (2012). Lyapunov’s direct method for stabilization of the Furuta pendulum. Turkish J Electr Eng Comput Sci., 20(1), 99–110.
