i welcome you all to this presentation on: neural network applications systems engineering dept....

I welcome you all to this I welcome you all to this presentationpresentation

On:On:

Neural Network ApplicationsNeural Network Applications

Systems Engineering Dept. KFUPMSystems Engineering Dept. KFUPM

Imran Nadeem & Naveed R. ButtImran Nadeem & Naveed R. Butt

220504 & 230353220504 & 230353

Part II: LMS & RBFPart I: Introduction to Neural NetworksPart III: Control Applications

Part I: Introduction to Neural Networks

Part I: Introduction to NN’sPart I: Introduction to NN’s There is no restriction on the

unknown function to be linear. Thus, neural networks provide a logical extension to create nonlinear adaptive control schemes.

Universal Approximation Theorem: neural networks can reproduce any nonlinear function for a limited input set.

Neural networks are parameterized nonlinear functions whose parameters can be adjusted to achieve different shaped nonlinearities.

In essence, we try to adjust the neural network to serve as an approximator for an unknown function that we know only through its inputs and outputs

Human NeuronHuman Neuron


Artificial NeuronArtificial Neuron


Adaptation in NN’sAdaptation in NN’s


Single Layer Feedforward Single Layer Feedforward NN’sNN’s



Multi-Layer Feedforward Multi-Layer Feedforward NN’sNN’s

Recurrent (feedback) NN’sRecurrent (feedback) NN’s


A recurrent neural network distinguishes itself from the feed-forward network in that it has at least one feedback loop. For example, a recurrent network may consist of a single layer of neurons with each neuron feeding its output signal back to the input of all input neurons.

Recurrent (feedback) NN’sRecurrent (feedback) NN’s


The presence of feedback loops has a profound impact on the learning capability of the network and on its performance.

Applications of NN’sApplications of NN’s


Neural networks are applicable in virtually every situation in which a relationship between the predictor variables (independents, inputs) and predicted variables (dependents, outputs) exists, even when that relationship is very complex and not easy to articulate in the usual terms of "correlations" or "differences between groups”

Applications of NN’sApplications of NN’s


Detection of medical phenomena

Stock market prediction

Credit assignment

Condition Monitoring

Signature analysis

Process control

Nonlinear Identification & Adaptive Control

End of Part IEnd of Part I

Part II: LMS & RBF

Part II: LMS & RBFPart II: LMS & RBF

LMS: The Adaptation Algorithm

RBF: Radial Bases Function NN

Part II: LMS & RBF

LMS: The Adaptation Algo.LMS: The Adaptation Algo.

][~][][ nsndne

2{| [ ] | }J E e n

Jnwnw w2

][]1[

Estimation Error

Actual Response Estimated Response

Cost FunctionMean Square Error

Weight Updates

Adaptation Step Size

Part II: LMS & RBF

RBF-NN’sRBF-NN’s

Radial functions are a special class of functions. Their characteristic feature is that their response decreases (or increases) monotonically with distance from a central point and they are radially symmetric.

Part II: LMS & RBF


GaussianRBF

Part II: LMS & RBF


Neural Networks based on radial bases functions are known as RBF Neural Networks and are among the most commonly used Neural Networks

Part II: LMS & RBF

RBF-NN’sRBF-NN’s Two-layer feed-forward

networks.

Hidden nodes: radial basis functions.

Output nodes : linear summation.

Very fast learning

Good for interpolation, estimation & Classification

Part III: Control Applications

Part III: Control ApplicationsPart III: Control Applications

Nonlinear System Identification

Adaptive Tracking of Nonlinear Plants

Nonlinear System Nonlinear System IdentificationIdentification




2 3 4

2 3 2

2 2 2 3 3

( 1) 0.8606 ( ) 0.0401 ( ) 0.0017 ( ) 0.000125 ( ) 0.0464 ( )

0.045 ( ) ( ) 0.0034 ( ) ( ) 0.00025 ( ) ( ) 0.0012 ( )

0.0013 ( ) ( ) 0.0001458 ( ) ( ) 0.00002083 ( ) 0.00002083 ( ) ( )

y t y t y t y t y t u t

y t u t y t u t y t u t u t

y t u t y t u t u t y t u t

Continuously Stirred Tank Reactor



0 20 40 60 80 100 120 140 160 180 200-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

Time

Plant OutputNN Output

Simulation ResultsUsing SIMULINK

Adaptive Nonlinear TrackingAdaptive Nonlinear Tracking




)(2.0)()(1)(

)2(1.0)1()1(5.0)(32 tutututx

txtxtyty

Hammerstein Model



0 10 20 30 40 50 60 70 80 90 100-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

Time

ReferencePlant OutputModel Output





0 10 20 30 40 50 60 70 80 90 100

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Time

Co

ntr

ol

Sig

nal

Thank youThank you

i welcome you all to this presentation on: neural network applications systems engineering dept....

Documents

rbf neural networks

lms rbf lms

neural networkspart

recurrent neural network

applications of nns

recurrent feedback nns

adaptation algorithm

radial bases function