Unit 1: Introduction
1.1 Approaches to Intelligent Control
• Conventional Control vs. Intelligent Control:
o Conventional Control: Based on mathematical models, relies on precise
input-output relationships.
o Intelligent Control: Emulates human decision-making, handles uncertainty,
and is adaptive to changing environments.
• Types of Intelligent Control Approaches:
o Rule-Based Systems: Uses if-then rules for decision-making.
o Fuzzy Logic Systems: Deals with reasoning that is approximate rather than
precise.
o Neural Networks: Mimics the human brain's ability to learn from data.
o Genetic Algorithms: Utilizes evolutionary strategies for optimization.
1.2 Architecture of Intelligent Control
• Hierarchical Structure:
o Supervisory Level: Oversees the overall system, making high-level
decisions.
o Intermediate Level: Executes supervisory commands, interacts with lower
levels.
o Execution Level: Directly controls the physical processes.
• Components of Intelligent Control:
o Sensors: Collect data from the environment.
o Processors: Analyze data and make decisions.
o Actuators: Execute actions based on decisions.
1.3 Linguistic Reasoning
• Definition: The use of linguistic variables (words rather than numbers) to reason
about data.
• Example: "If the temperature is high, then reduce the heater power."
• Applications: Fuzzy logic control systems, where rules are expressed in natural
language.
1.4 Rulebase
• Definition: A set of rules that defines the behavior of an intelligent system.
• Structure:
o Condition: The situation or input scenario.
o Action: The response or output based on the condition.
• Example: "If traffic is heavy, then delay green light by 5 seconds."
1.5 Knowledge Representation
• Definition: The method used to represent information in an AI system.
• Types:
o Logical Representation: Uses logic to represent facts.
o Semantic Networks: Represents knowledge in graphs.
o Frames: Structured representations of stereotypical situations.
o Production Rules: If-then rules for decision-making.
Unit 2: Artificial Neural Networks
2.1 Biological Neuron
• Structure:
o Dendrites: Receive signals from other neurons.
o Cell Body: Processes the signals.
o Axon: Transmits signals to other neurons.
2.2 Artificial Neural Network (ANN)
• Definition: A computational model inspired by the biological neuron.
• Components:
o Nodes (Neurons): Process input data.
o Weights: Strength of the connections between neurons.
o Activation Function: Determines the output of a neuron.
2.3 Mathematical Models
• McCulloch-Pitts Model: A simplified model of a neuron that outputs binary values.
• Perceptron: A single-layer neural network for binary classification.
• Adaline and Madaline: Variants of perceptron with adaptive weights.
2.4 Learning & Training in ANN
• Supervised Learning: Training with labeled data.
• Unsupervised Learning: Finding patterns in unlabeled data.
• Backpropagation: A method for updating weights in a multi-layer network.
2.5 Hopfield Neural Network
• Definition: A recurrent neural network with binary threshold nodes.
• Applications: Associative memory, optimization problems.
2.6 Self Organizing Networks
• Kohonen Networks: A type of self-organizing map (SOM) used for clustering and
visualization.
2.7 Recurrent Networks
• Definition: Networks where connections form directed cycles.
• Applications: Time-series prediction, language modeling.
2.8 Associative Memories
• Definition: Memory systems that recall information based on input patterns.
• Types:
o Autoassociative Memory: Stores and retrieves patterns.
o Heteroassociative Memory: Maps input patterns to output patterns.
Unit 3: Fuzzy Logic System
3.1 Crisp Vs Fuzzy Set Theory
• Crisp Sets: Objects either belong to a set or not.
• Fuzzy Sets: Objects have degrees of membership in a set.
3.2 Membership Functions
• Definition: A function that defines how each element in the domain is mapped to a
membership value.
• Types:
o Triangular: Defined by a triangular shape.
o Trapezoidal: Defined by a trapezoidal shape.
o Gaussian: Defined by a bell curve.
3.3 Fuzzy Set Operations
• Union, Intersection, and Complement: Basic operations similar to classical set
theory but extended for fuzzy sets.
3.4 Fuzzy Rules
• Definition: Rules that use fuzzy logic to map inputs to outputs.
• Example: "If temperature is high and humidity is low, then fan speed is medium."
3.5 Mamdani and Sugeno Fuzzy Inference Systems
• Mamdani: Uses min-max operations and centroid method for defuzzification.
• Sugeno: Uses weighted average for defuzzification.
3.6 Defuzzification Methods
• Centroid Method: Calculates the center of area under the curve.
• Mean of Maximum: Averages the maximum membership values.
• Weighted Average: Weighs outputs based on membership values.
Unit 4: ANN Methods and Parameters
4.1 Introduction and Biological Background of GA
• Genetic Algorithm (GA): A search heuristic inspired by the process of natural
selection.
• Biological Basis: Involves processes like mutation, crossover, and selection.
4.2 String Encoding of Chromosomes
• Binary Encoding: Chromosomes represented as binary strings.
• Other Encodings: Real numbers, permutations, etc.
4.3 Selection Methods
• Roulette Wheel Selection: Probability of selection proportional to fitness.
• Tournament Selection: Selects the best individual out of a random sample.
4.4 Crossover Operation
• Single-Point Crossover: One crossover point.
• Multi-Point Crossover: Multiple crossover points.
4.5 Mutation
• Bit-Flip Mutation: Flipping bits in the chromosome.
• Swap Mutation: Swapping two elements in the chromosome.
4.6 Adjustment of Strategy Parameters
• Population Size: Number of individuals in the population.
• Mutation Probability: Likelihood of mutation.
• Crossover Probability: Likelihood of crossover.
Unit 5: Fuzzy Logic MATLAB Toolbox
5.1 Fuzzy Logic Toolbox
• Description: A MATLAB tool for designing and simulating fuzzy logic systems.
• Key Features: Fuzzy inference system editor, membership function editor, rule
editor.
5.2 Neural Network Toolbox
• Description: A MATLAB tool for designing and simulating neural networks.
• Key Features: Network creation, training, and visualization tools.
5.3 FLS for Antilock Braking System (ABS)
• Overview: Design of a fuzzy logic system to control ABS in vehicles.
• Process: Define membership functions, create rules, simulate behavior.
5.4 GA in Route Planning for Travelling Sales Person
• Problem: Find the shortest route that visits a set of cities.
• Solution: Use GA to optimize the route.
5.5 Time-Series Forecasting using ANN
• Overview: Use of ANN to predict future values in a time series.
• Process: Train the network on historical data, test predictions.
