ARTIFICIAL NEURAL NETWORK WITH APPLICATIONS TO POWER SYSTEMS (18EE745)

ARTIFICIAL NEURAL NETWORK WITH APPLICATIONS TO POWER SYSTEMS

Subject Code 18EE745 
CIE Marks 40
Number of Lecture Hours/Week (L:T:P) 3:0:0 
SEE Marks 60
Credits 4 
Exam Hours 03

Course Learning Objectives:

· To understand the fundamental concepts and models of Artificial Neural Systems.
· To understand neural processing, learning and adaptation, Neural Network learning rules.
· Ability to analyze multilayer feed forward networks.
· Ability to develop various ancillary techniques applied to power system and control of power
systems.

Module-1

Fundamental Concepts and Models of Artificial Neural Systems Biological Neurons and their artificial models – Biological Neuron, McCulloch-Pitts Neuron Model, Neuron modeling for Artificial neural systems. Models for Artificial Neural Networks – Feedforward Network, Feedback network. 

Module-2

Neural Processing, Learning and Adaptation, Neural Network Learning Rules Neural Processing. Learning and Adaptation – Learning as Approximation or Equilibria Encoding, Supervised and Unsupervised Learning. Neural Network Learning Rules – Hebbian Learning Rule, Perceptron Learning Rule, Delta Learning Rule, Widrow-Hoff Learning Rule, Correlation Learning Rule, Winner-Take-All Learning Rule, Outstar Learning Rule, Summary of Learning Rules.

Module-3

Multilayer Feedforward Networks Feedforward Recall and Error Back-Propagation Training – Feedforward Recall, Error Back- Propagation Training, Training Errors and Multilayer Feedforward Networks as Universal Approximators (Excluding Examples). Learning Factors – Initial Weights, Cumulative Weight Adjustment versus Incremental Updating, Steepness of the Activation Function, Learning Constant, Momentum Method, Network Architectures Versus Data Representation, Necessary Number of Hidden Neurons.

Module-4

Neural Network and its Ancillary Techniques as Applied to Power Systems Introduction, Learning versus Memorization, Determining the Best Net Size, Network Saturation, Feature Extraction, Inversion of Neural Networks, Alternative Training Method: Genetic Based Neural Network, Fuzzified Neural Network.

Module – 5

Control of Power Systems Introduction, Background, Neural Network Architectures for modeling and control, Supervised Neural Network Structures, Diagonal Recurrent Neural Network based Control System, Convergence and Stability.

Course Outcomes: At the end of the course the student will be able to:

· Develop Neural Network and apply elementary information processing tasks that neural network
can solve.
· Develop Neural Network and apply powerful, useful learning techniques.
· Develop and Analyze multilayer feed forward network for mapping provided through the first
network layer and error back propagation algorithm.
· Analyze and apply algorithmic type problems to tackle problems for which algorithms are not
available.
· Develop and Analyze supervised/unsupervised, learning modes of Neural Network for different
applications.

Question paper pattern:

· The question paper will have ten questions.
· Each full question is for 20 marks.
· There will be 2 full questions (with a maximum of three sub questions in one full question) from
each module.
· Each full question with sub questions will cover the contents under a module.
· Students will have to answer 5 full questions, selecting one full question from each module.

Text Books

1 Introduction to Artificial Neural Systems. Jacek M. Zurada JAICO Publishing House 2006
2 Artificial Neural Networks 200 with Applications to Power Systems Edited by – Mohamed El – Sharkawi and Dagmar Niebur