Neural Optimal Control Research Articles

A New-Generation Motion-Control System for Twin-Hull Vessels Using a Neural Optimal Controller

A new ride control system using a neural optimal controller (NOC) is developed and applied to improve the heave and pitch motion responses of two twin-hull vessels operating in regular head seas. A time domain model for the vessel dynamics in the presence of active fin control is used to simulate the vessel and fin motion responses. An online switching procedure is introduced to select among a number of linear quadratic regulator optimal controllers, designed for different operating conditions of the vessel, to improve the system robustness. Although the online switching offered better robustness and performance characteristics, in between switching operating points, it still remained suboptimal. Therefore, an artificial neural network (ANN) controller was developed as an alternative and initially trained to emulate the same level of control at a number of design operating points, as a NOC. The advantage of this novel application is that practical difficulties in applying an online switching procedure are no longer present and, more importantly, the ANN has been capable of nonlinear generalization to give a near optimal solution away from the trained operating conditions.

Optimization of a building's cooling plant for operating cost and energy use

An optimal neural network-based controller for an ice thermal storage system has been developed and tested. The controller consists of four neural networks, three of which map equipment behavior and one that acts as a global controller. The controller self-learns equipment responses to the environment and then determines the control settings required to minimize operating cost. It has the advantage over other controllers in that it always remains calibrated. Since it does not rely upon rules or assumptions, it is able to provide optimal control under any utility pricing and operating condition. Although originally designed to minimize operating costs, simulation and optimization techniques often determine minimum energy use as well.

Optimal Control of a Turbocharged Diesel Engine Using Neural Network

This paper describes an optimal neural controller, applied to a turbocharger diesel engine. It presents the training of a neural controller, based on the theory of backpropagation, for trajectory tracking of the engine speed. The Gauss-Newton algorithm will be explained in a recursive manner, including a forgetting factor for time varying systems. Some simulation results are presented.

Twin-Hull Vessel Motion Control Using a Neural Optimal Controller Design

A Neural Optimal Controller is designed for the reduction of vertical plane motions of a twin-hull vessel over a wide range of encounter wave frequencies in head seas. Improvement in system performance is achieved by dynamic switching between optimal controller designs for different sea conditions. To overcome control problems associated with the switching procedure, a dynamic Artificial Neural Network is used and is initially trained to emulate the same level of control at a number design operating points. The ANN is capable of generalising to give a near optimal solution away from the operating conditions.

Feedback optimal neural network controller for dynamic systems — A ship maneuvering example

The present paper deals with the application of the feedforward neural network in feedback control for dynamic systems. First, the problem of non-linear control is solved using the optimal control approach. Then assuming that prior knowledge is available in a certain form (input/output data-base and the specification between the data points), a neural network and a corresponding parameter design method are proposed. The feedforward network, trained to map the non-linear optimal control strategy, is used in feedback loop within the system to validate the proposed approach. Finally, the example of ship maneuvering and control is presented.

Disturbances of learning processes in the basal ganglia in the pathogenesis of Parkinson's disease: a novel theory.

A hypothesis is proposed that the cortico-basal ganglia-thalamocortical circuit is a neural optimal control system containing a model of the controlled object. To predict the behaviour of the object, the model uses the language of afferent signals that enter the system. Based on these ideas, it is suggested that the skeletomotor basal ganglia-thalamocortical circuit serves to model the motor behaviour of the body of an individual and its environment during motor planning and performance. Association basal ganglia-thalamocortical circuits are capable of modeling the behaviour of abstract objects such as thoughts. Within the limits of the proposed theory, the dopaminergic system serves to distribute an error signal within the striatum. These error signals contain information about mismatch between model (predicted) afferent flow and real afferent flow coming from the controlled object. An error signal is necessary to tune the model on the object, and is delivered to a structure within the circuit that is responsible for the production of the error signal. An error signal is minimal when the model properly describes the object behaviour. The process of learning is initiated when an error signal increases and is complete when the error signal is minimized. Parkinson's disease is thus considered to be the consequence of modeling disfunction because of progressive functional and structural degeneration of the error distribution system in the substantia nigra pars compacta. Other clinical applications of the proposed theory are also discussed.

Intelligent Control Systems: Are They for Real?

The fundamental concept of feedback to control dynamic systems has played a major role in many areas of engineering. Increases in complexity and more stringent requirements have introduced new challenges for control systems. This paper presents an introduction to and appreciation for intelligent control systems, their application areas, and justifies their need. Specific problem related to automated human comfort control is discussed. Some analytical derivations related to neural networks and fuzzy optimal control as elements of proposed intelligent control systems, along with experimental results, are presented. A brief glossary of common terminology used in this area is included.