A. Tiano

On the identification of non-linear models of unmanned underwater vehicles

Abstract This documentation presents a comparison between two identification methods for the off-line identification of non-linear models of unmanned underwater vehicles (UUVs), one based on the minimization of the acceleration prediction error (direct method) and another one based on the minimization of the velocity one step prediction error (integral method). The direct method has already been used in UUVs identification (IFAC Conference CAMs’ 2001, Control Application in Marine Systems, Glasgow, Scotland, UK, 2001). Our new proposal, the integral method, can be applied to a quite general class of non-linear multivariable models and is characterized by an excellent numerical performance. Both methods are compared through their application to the identification of the dynamic model of URIS UUV. Results suggest that better models can be obtained using the proposed method (integral method).

The design of a navigation guidance and control system for an unmanned surface vehicle for environmental monitoring

Maintaining the ecosystem is one of the main concerns in this modern age. With the fear of ever-increasing global warming, the UK is one of the key players to participate actively in taking measures to slow down at least its phenomenal rate. As an ingredient to this process, the Springer vehicle has been designed and is being developed for environmental monitoring and pollutant tracking. This paper highlights the Springer hardware and software architecture including various navigational sensors, a speed controller, and an environmental monitoring unit. In addition, details regarding the modelling of the vessel are outlined which are based mainly on recent trials data. The formulation of a fault tolerant multi-sensor data fusion technique is also presented. Moreover, control strategy based on a linear quadratic Gaussian controller is developed and simulated on the Springer model.

On the Identification of Non Linear Models of Unmanned Underwater Vehicles

Abstract This documentation presents a comparison between two identification methods for the off-line identification of non-linear models of unmanned underwater vehicles (UUVs), one based on the minimization of the acceleration prediction error (direct method) and another one based on the minimization of the velocity one step prediction error (integral method). The direct method has already been used in UUVs identification (IFAC Conference CAMs’ 2001, Control Application in Marine Systems, Glasgow, Scotland, UK, 2001). Our new proposal, the integral method, can be applied to a quite general class of non-linear multivariable models and is characterized by an excellent numerical performance. Both methods are compared through their application to the identification of the dynamic model of URIS UUV. Results suggest that better models can be obtained using the proposed method (integral method).

Experimental Comparison of Identification Methods for an Open-Frame ROV

Abstract This paper deals with the identification of hydrodynamic derivatives in a Remotely Operated Vehicle (ROV). In-water tests have been performed to identify yaw inertial and drag coefficients on the prototype ROV Roby2, developed at IAN. Vehicle heading has been measured losing a commercial digital fluxgate compass. A least square off line polynomial data fitting technique to compute signal derivatives is applied to experimental compass data. Subsequently, least squares and random-search algorithms are applied to experimental data to identify the vehicle models coefficients. The experimental set-up and modalities are discussed and results of the least squares and random-research algorithms are compared.

Comparison of non linear identification methods for underwater vehicles

This paper deals with the identification of non linear multivariable models of underwater vehicles. For this purpose two different, recently proposed methods, are presented and compared from a methodological point of view. The first method is based on a Lyapunov approach, while the second one is based on the least squares approach. The two considered methods operate in the continuous time domain and can be applied to non linear models that are linear with respect to the unknown parameter vector. After an introduction, where the role of identification methods in the area of guidance and control of underwater vehicles is outlined, the mathematical models that are generally used for describing the dynamics of underwater vehicles are concisely presented. The main features of the two identification methods are then discussed and compared.

A Fuzzy Controller for Integrated Ship Motion Control

Abstract This paper describes The development of fuzzy controller for integrated control of the coupled yaw-roll motion of a container ship. The basic fuzzy rules used for designing an integrated multivariable rudder roll damping and yaw autopilot are discussed. A random search learning mechanism for automatically tuning the fuzzy controller parameters is presented. The efficiency and robustness of the controller are evaluated by a number of simulation tests conducted at different sea states. The results are used to evaluate the performance of the fuzzy controller in comparison with a conventional LQ design.

Fuzzy course-keeping autopilot for ships

Abstract In this paper a course-keeping autopilot for a containership designed with fuzzy logic theory is presented. The autopilot control strategy is deduced heuristically by exploiting expert knowledge and is implemented by means of fuzzy logic. In order to facilitate analytical analysis of the closed loop system non-linear control theory is used to guide the choice of control structure. An interpretation in terms of Fourier analysis of the control strategy is given and used in order to improve the performance of the autopilot in different sailing conditions while preserving the grey nature of fuzzy systems. The final autopilot is proved to be locally stable in the Lyapunov sense while a set of simulation results on a non-linear model of a containership shows the viability of the proposed approach.

An Adaptive Fuzzy Autopilot for a Containership

Abstract During the last few years, fuzzy controllers have emerged as a viable alternative to others conventional control design techniques. The primary advantage of fuzzy systems, is the ability to manipulate uncertainties and expert knowledge. In this paper, the problem of designing an indirect adaptive fuzzy autopilot for a containership is addressed. The proposed adaptation mechanism is based on the interpretation of a performance function by a fuzzy system, which in turn will infer the change (adaptation) in the fuzzy controller parameters. The main advantages of using a fuzzy system to perform the adaptation with respect the use of a classical adaptation algorithms (i.e. back-propagation, least-square, MIT rule etc.), is related to the amount and form of a-priori information needed about the system to be controlled. The speed with which the adaptation is achieved makes this approach viable for on-line implementation. Simulations on a non-linear model of a containership are presented for both course-changing and course-keeping manoeuvres in the presence of environmental disturbances.

Ship Monitoring and Control by Multisensor Data Fusion

This paper discusses the applicability of ship monitoring and control by multisensor data fusion After introducing some critical issues in the area of ship motion control, an analysis is presented concerning the choice of different architecture of ship monitoring and prediction systems. A simulation study is presented concerning the design of a multisensor data fusion system for improving estimate of ship position based on low-cost onboard hardware.Copyright

SHIP MONITORING AND CONTROL

Abstract This paper discusses the applicability of monitoring systems for improving ship control. After introducing some critical issues in the area of ship motion control an analysis is presented concerning the choice of different architecture and algorithms with particular reference to the applicability of new methods capable to implement robustness and adaptivity features and to the software requirements for building up advanced ship monitoring and control systems. An examples of ship monitoring and control system actually under development at the Department of Information and Systems of the University of Pavia are presented