Simulation and tracking control based on neural-network strategy and sliding-mode control for underwater remotely operated vehicle
Introduction
Underwater vehicles are useful and considerable tools to accomplish various operations and have increasing applications in exploring underwater environments. These applications are involved in good maneuvering capabilities and a high-precision track-keeping along the specified path. Of course, considering a suitable control form for these underwater vehicles are very difficult due to the strong nonlinearity of the dynamic behavior and uncertainties. Therefore, research in the area of controlling the underwater vehicles is an important challenge for control engineers and has been considered by many authors.
In the last decade, for steering an underwater vehicle controllers based on the linearization theory were presented. It should be noted that many of them cannot reduce the tracking error and accomplish the operation with a high-accuracy process. Since the dynamics of remotely operated vehicles (ROVs) are highly nonlinear and usually contain uncertain elements, many efforts have been made in developing control schemes to get an appropriate solution to achieve the precise tracking control. However, ROVs have to face various uncertainties in practical application, such as internal and external frictions, damping and external disturbances [7]. All the uncertainties or time-varying factors could affect the system's control performance seriously. Many control techniques of the conventional PID-type controllers [3], [12], [16], [20] are used to control robot manipulators or other dynamic systems. University of New-Hampshire uses PID-control law for the EAVE-EAST vehicle. Tagegaki and Arimoto (1981) applied PID set-point regulation in terms of the Lyapunov stability to control robot manipulators. However, the nonlinear control of underwater vehicles in terms of Euler's angles feedback was first discussed by Fossen and Sagatun [9]. Later, this work was extended to quaternion feedback regulation in terms of vector quaternion, Euler rotation and Rodrigues parameter feedback by Fjellstad and Fossen [7]. Armito and Miyazaki (1984) applied square nonlinear PID-control for the system. Healey and Marco (1992) suggested a decoupled control design and Liceage et al. (1994) employed a robust control. Fossen and Sagatun (1991) had shown a controller with an adaptive method for marine vehicles. Many other methods to control the dynamic system could be found too. These are namely fuzzy control [10], [1] (Fairbrother and Stacy, 1991), neural network (NN) [24], and sliding-mode control [22] (Yoerger and Slotine, 1991; Cristy et al., 1996). Also, Healey and Lienard (1993) have applied the theory of sliding-mode regimes to control the NPS AUV II.
On the other hand, Kim and Lewis [11] have shown the application of quadratic optimization for the motion control of robotic systems using Cerebellar model arithmetic computer NNs. Lewis et al. [15] have presented a multilayer NN controller for a general serial-link rigid robot to guarantee the tracking performance.
Both system-tracking stability and error convergence can be guaranteed in these neural-based-control systems [11], [14]. However, the high-order term in the Taylor series, the functional reconstructed error, and the neural tuning weights are assumed to be known and bounded functions. Also, some inherent properties of ROV are required in the design process (bounded system parameters, disturbances, and skew-symmetry property). The purpose of this study is to design an intelligent control scheme for the tracking control of this ROV to achieve the precision position control.
This study is organised as follows. The next section is ROV description, which introduces the propulsion system. In this manner, the procedure of design and configuration of the vehicle are presented. Also, after description of equations of motion, the dynamics model of ROV is evaluated in controllable degrees of freedom (DOFs). Section 3 shows the design process of a traditional sliding-mode control system. In Section 4, considering the effect of uncertainties, an SMNNS control system is designed regarding the predefined motion of ROV. All adaptive-learning algorithms in the SMNNS control system are derived from the sense of the Lyapunov stability analysis, so that system-tracking stability can be guaranteed in the closed-loop system. In Section 5, simulation results of ROV under possible occurrence of uncertainties are provided to demonstrate the scalar and robust control performance of the proposed SMNNS control system. In Section 6, conclusions are drawn.
Section snippets
ROV description
This ROV is able to move 200 m down the surface of the water. With the aim of designing a low-cost vehicle, the underwater ROV has been built using low-cost materials. However, this ROV should be able to move on a predefined path with high accuracy, which is our purpose of designing the high-level controller.
To solve the problem of resistance to underwater pressure, the vehicle is equipped with a compressed air cylinder like those used by submersible vehicles in diving missions.
Air consumption
Controller design
The equation of motion (1) and (9) can be written in the earth-fixed reference frame in terms of position and attitude through the kinematics transformations:where
In this section, the purpose of controller design is a tracking problem one. Therefore, a control law will generate so that the state x(t) can track the desired command xd(t). To achieve this
Sliding-mode neural-network control system
Wai [23], for first time, designed an sliding-mode neural-network (SMNN) control for a robot manipulator. In this section, for the fasting convergence, a scalar controller τ0 will add to rules of the SMNN control system and an SMNNS control system with high accuracy and fast converging will generate. At the end, this SMNNS control system for the mentioned ROV will be used and the results of the tracking control problem for two fuzzy logic sliding-mode control systems (FLSMC) and SMNNS
Simulation results
The results of simulation for an underwater vehicle model of the ROV are presented in this section. The equation of motion for four DOFs of the underwater vehicle was considered, and it was controlled in all four controllable DOFs by four dc-motor-driven thrusters. The fifth motor is used for proper landing and does not have any effect on tracking performance of the vehicle.
Considering the system parameters in body coordinate of the formwhich implies
Fuzzy-sliding-mode control system
The fuzzy-sliding-mode tracking control system [10], [19], [21], [4] for nine fuzzy rules is defined of the form:
IF is PB1 (positive big), THEN u is PB2 (positive big);
IF is PM1 (positive medium), THEN u is PM2 (positive medium);
IF is PS1 (positive small), THEN u is PS2 (positive small);
IF is P1 (positive), THEN u is P2 (positive);
IF is ZR1 (zero), THEN u is ZR2 (zero);
IF is N1 (negative), THEN u is N2 (negative);
IF is NS1 (negative small), THEN u is NS2 (negative
Conclusions
The ROV has been designed and developed by the Department of Mechanical Engineering of the University of Guilan for underwater exploration and with special application on studying and monitoring fish behavior. The main components and design considerations and configurations are described and presented in this article.
An SMNNS control system applied to this ROV considered the scalar control to achieve high-accuracy position control. The dynamics of the vehicle need not be known explicitly and no
Ahmad Bagheri was born in Tehran, Iran, on 1966 and completed his B.Sc., M.Sc., and Ph.D. studies in 1989, 1993, and 1998 from Iranian Ferdowsi Mashad University, Iranian Tarbiat Modarres University, and Czech Technical University at Prague, respectively, in Mechanical Engineering. He has published more than 70 papers in journals and conference proceedings. His main specialty focused on the control applications on Robotics and Mechatronics devices and nonlinear control theories. He worked for 4
References (24)
- et al.
Decoupled fuzzy controller design with single-input fuzzy logic
Fuzzy Sets and Systems
(2002) - et al.
A global output-feedback controller for stabilization and tracking of underactuated ODIN: a spherical underwater vehicle
Automatica
(2004) - et al.
Structural identification of fuzzy model
Fuzzy Sets and Systems
(1988) - et al.
Designing a fuzzy-like PD controller for an underwater robot
Control Eng. Pract.
(2003) - et al.
Adaptive Control
(1995) - I. Cha, C. Han, The auto-tuning PID controller using the parameter estimation, in: IEEE/RSJ International Conference on...
- G. Conte, A. Serrani, Modeling and simulation of underwater vehicle, in: Proceedings on IEEE International Symposium on...
Guidance and Control of Ocean Vehicles
(1994)Marin control system: guidance, navigation and control of ships, rigs, and underwater vehicles
Mar. Cybernet.
(2002)- T.I. Fossen, S. Sagatun, Adaptive control of nonlinear underwater robotic system, in: International Conference of...
Design of a sliding-mode fuzzy controller for the guidance and control of an autonomous underwater vehicle
Ocean Eng.
Optimal design of CMAC neural-network controller for robot manipulators
IEEE Trans. Systems Man Cybernet.
Cited by (69)
Optimized neuro observer-based sliding mode control for a nonlinear system using fuzzy static sliding surface
2022, Applied Soft ComputingSliding mode based fault tolerant control for autonomous underwater vehicle
2020, Ocean EngineeringAn efficient hybrid approach for trajectory tracking control of autonomous underwater vehicles
2020, Applied Ocean ResearchThruster fault reconstruction for autonomous underwater vehicle based on terminal sliding mode observer
2020, Fundamental Design and Automation Technologies in Offshore RoboticsAdaptive sliding mode control based on local recurrent neural networks for an underwater robot
2020, Fundamental Design and Automation Technologies in Offshore Robotics
Ahmad Bagheri was born in Tehran, Iran, on 1966 and completed his B.Sc., M.Sc., and Ph.D. studies in 1989, 1993, and 1998 from Iranian Ferdowsi Mashad University, Iranian Tarbiat Modarres University, and Czech Technical University at Prague, respectively, in Mechanical Engineering. He has published more than 70 papers in journals and conference proceedings. His main specialty focused on the control applications on Robotics and Mechatronics devices and nonlinear control theories. He worked for 4 years for the Iranian Ministry of Industries and also 15 years in academic areas. His research interests include the biped locomotion, robot modeling, and industrial mechatronics. He is currently serving as an Associate Professor of Robotics and Mechatronics and also as the Vice Dean of the Faculty of Engineering of the University of Guilan, Iran.
Jalal Javadi Moghaddam received his B.Sc. degree from Azad University of Tehran in 2004, and M.Sc degree from University of Guilan in 2007, all in Mechanical Engineering. His research interests include nonlinear dynamical systems, expert systems, intelligent control systems and soft computing.