The learning algorithms and structures which have become popular in the neural network community over the last few years have been successfully applied to various challenging modelling problems. In contrast to the lin...
详细信息
The learning algorithms and structures which have become popular in the neural network community over the last few years have been successfully applied to various challenging modelling problems. In contrast to the linear modelling methods, there is still no clearly defined engineering process from which a model can reliably be created from measurements taken from a physical system. Problems for the practical application of neuralnetworks involve the interpretation of the trained models, and the explicit introduction of a priori models into the learning system, as well as the use, in many cases, only basic ad hoc validation and experiment design techniques. This talk will discuss several methods which can improve the engineering aspects of model identification with neural nets.< >
The use of a neural network to learn the nonlinear current profiles required to minimise torque ripple in a switched reluctance motor (SRM), at low to medium speeds, has been demonstrated using a digital signal proces...
详细信息
The use of a neural network to learn the nonlinear current profiles required to minimise torque ripple in a switched reluctance motor (SRM), at low to medium speeds, has been demonstrated using a digital signal processor (DSP). However, the DSP (Texas Instruments TMS320C25) implementation of a neural network in this application is a limiting factor on motor speed (if maximum current profile integrity is to be maintained). Fortunately, the neural network architecture used (cerebellar model articulation controller (CMAC)) is amenable to hardware implementation a point noted by Albus in one of his original papers and evidenced by at least one previous field programmable gate array (FPGA) implementation. Guided by the requirements of the switched reluctance motor application, a prototype accelerator based on a Xilinx XC4000 FPGA implementation of the neural network has been constructed that operates an order of magnitude faster than the DSP implementation.
The ability to learn is the cornerstone of current neural technology. It was responsible for their decline in the late sixties when it was shown that the perceptron training algorithm could not be easily extended to m...
详细信息
The ability to learn is the cornerstone of current neural technology. It was responsible for their decline in the late sixties when it was shown that the perceptron training algorithm could not be easily extended to multilayered networks, and also the revival of interest in these techniques was initiated by the discovery of a multilayer network adaptation rule. Artificial neuralnetworks attempt to mimic a human's information processing capabilities by building neuronally inspired systems which learn to interact with their environment in a desirable fashion. There are many engineering problems which could benefit from the use such of such systems, although there are also problems with applying these adaptive networks. Most artificial neuralnetworks can be regarded as "black box" learning systems; they are difficult initialise as knowledge is stored in an opaque fashion and validation can only performed using input/output data; their internal structure provides little information to an engineer. In addition, the generalisation, modelling and learning abilities of these networks are generally poorly understood, although such results are necessary when these adaptive systems are applied online in safety critical situations.< >
In this paper we prove the stability of a certain class of nonlinear discrete MIMO systemscontrolled by a multilayer neural net with a simple weight adaptation strategy. The proof is based on the Lyapunov formalism. ...
详细信息
In this paper we prove the stability of a certain class of nonlinear discrete MIMO systemscontrolled by a multilayer neural net with a simple weight adaptation strategy. The proof is based on the Lyapunov formalism. The stability statement is, however, only valid if the initial weight values are not too far from their optimal values that allow perfect model matching. We therefore propose to initialize the weights with values that solve the linear problem. This extends our previous work (Renders, 1993; Saerens, Renders and Bersini, 1993), where SISO systems were considered.< >
作者:
Zhu, Q.M.Faculty of Engineering
University of the West of England Frenchay Campus Coldharbor Lane Bristol BS16 1QY United Kingdom
The author attended the IFAC Workshop on Digital control - past, present, and future of PID control, Teresa, Spain, 5-7 April, 2000 and presented a joint paper. This presentation reports back the development on intell...
详细信息
The author attended the IFAC Workshop on Digital control - past, present, and future of PID control, Teresa, Spain, 5-7 April, 2000 and presented a joint paper. This presentation reports back the development on intelligent PID controller design in the workshop. At present structure of PID controllers is quite different of the original analogue PID controllers, the PID controller design combining with neuralnetworks, fuzzy logic, auto-tuning, and inductive learning mechanisms is one of the key sessions during the workshop, which has attracted scientific specialists in control methodology and users with industrial control applications to work together for the exploitation of these new capabilities.
This paper presents firstly the use of artificial neuralnetworks to classify power system faults. Examples will be used to demonstrate this approach such as faults occurring in high voltage transmission systems, or t...
详细信息
ISBN:
(纸本)0852969120
This paper presents firstly the use of artificial neuralnetworks to classify power system faults. Examples will be used to demonstrate this approach such as faults occurring in high voltage transmission systems, or those stored with a digital recorder. The paper proposes an adaptive scheme employing the neural network for developing digital distance relay. High impedance faults and variable source impedance will also be considered. An example based on a three-terminal line configuration will be used to illustrate the effectiveness of the method. Secondly, a discussion on the future use of NN in protection will be given. In conclusion, neuralnetworks should be integrated with different computational techniques to enhance its application to fault classification and protection.
Many authors have shown by simulated studies, that a great number of non-linear dynamical systems could be identified and controlled by using neural network models. The authors applied these results to a real process ...
详细信息
Many authors have shown by simulated studies, that a great number of non-linear dynamical systems could be identified and controlled by using neural network models. The authors applied these results to a real process : an oven with two inputs, one for heating and one for cooling. The output to be controlled is the temperature inside the oven. The choice of the control strategy on the one hand and the choice of the neural network architecture for the oven identification on the other hand had to satisfy two main objectives. First, the control strategy should be quite insensible to random disturbances (air leaks, door openings, ...) and the neural model should be able to fit any modification of the plant dynamics during its lifetime (modification of the internal load alteration of the heating system ...). The authors chose to use the internal model control as their control strategy. They also used a radial basis function network to identify the plant. The process identification is composed of two phases, an off-line one and an online one. The off-line part consists first in training the network while determining its internal structure using an initial training set. The on-line phase is the adaptive part of the control scheme.< >
Provision of reliability of the walking robots is complex problem. Application of soft computing allows to provide the reliability of robots. Soft-computing is new discipline that bring together all features of fuzzy-...
详细信息
Provision of reliability of the walking robots is complex problem. Application of soft computing allows to provide the reliability of robots. Soft-computing is new discipline that bring together all features of fuzzy-logic, genetic programming, and neuralnetworks. The main peculiarity of soft-computing is capability to treat with uncertain systems that cannot be easily modelled an controlled by using the classical approaches. The walking robots are typical example of systems affected by uncertainty: leg kinematic is often non-linear and known with low accuracy. Overview of soft-computing techniques developed that have been applied to several walking robots is given.
This paper describes a neural network and its simulation results for fault diagnosis in HVDC systems. Fault diagnosis is carried out by mapping input data patterns, which represent the behaviour of the system, to one ...
详细信息
This paper describes a neural network and its simulation results for fault diagnosis in HVDC systems. Fault diagnosis is carried out by mapping input data patterns, which represent the behaviour of the system, to one or more fault conditions. The behaviour of the converters is described in terms of the time varying patterns of conducting thyristors and ac & dc fault characteristics. A three-layer neural network consisting of 20 input nodes, 12 hidden nodes and 2 output nodes is used. This paper will describe the performance of the network for ac and dc faults due to changes in number of hidden layers, number of neurons in the layer, learning rate and momentum. Dynamic characteristics of networks for different configurations are studied too. The time performance of the network is also included. neuralnetworks provide an effective way for fault diagnosis and identification. Simulation data obtained from the EMTP will be used to test the performance of this network. The comparison will be given and the result is promising.
The developmental stages leading to the use of artificial neural network (ANN) in the automation of water clarification control are presented. Before ANN, the purchase of an equipment to automate dosing was considered...
详细信息
The developmental stages leading to the use of artificial neural network (ANN) in the automation of water clarification control are presented. Before ANN, the purchase of an equipment to automate dosing was considered a great success. However, the automated dosing system proved inaccurate at times. Analysis revealed some discrepancies between meter readings and laboratory readings on color at high turbidity. Chemists suggested that other variables such as temperature and conductivity may play a role in clarification control. It seemed that with more data available a more accurate law could be determined using ANN. Finally, ANN brought a clearer picture of the situation. Inferential estimation can therefore prove to be a cost-effective means of determining plant measurements for variables which are impractical to measure. By this means, it may also be possible to speed up the sampling time of a measured variable for more accurate control.
暂无评论