INTELLIGENT SYSTEM
NEUNGLUETHAI BOORASIT
5537857 EGBE/M
To get an understanding of the theoretical foundations of various types of intelligent systems technology at levels sufficient to achieve the objective
The Intelligent System learns how to perform in order to achieve the objectives or goals. It has main objectives of theirs own as well as senses and motivation. To achieve its objectives, select actions based on their experience. It can learn by generalizing the experience it has stored in its memory. The system should be able to support users in solving complex problems by modeling on the basis of information and by offering the ability to reason about the model.
It is well known that the intelligent system that can provide expertise as well as knowledge of human reason to unpredictable and adaptation to noisy environments and at different times, which is important in solving problems to use it in practical computer.
Define intelligent system techniques
Intelligent system techniques
Neural Networks (NN): Systems that simulate intelligence by attempting to reproduce the types of physical connections that occur in animal brains
Fuzzy Inference Systems (FIS): The system was used in the estimation between the input and output with accuracy and flexibility are often applied when faced with ambiguity.
Genetic Algorithms (GA): Inputs are derived from random, then the sort to try to find new ways to solve problems. Which will be produced crossing and mutations occur in the next version. To develop a new generation and better.
Evolutionary Strategies (ES): How nature solves optimization problem extremely successful. The mechanisms of natural evolution (reproduction, mutation, recombination, and selection) can also be replicated mathematically and are hereby available for any kind of engineering problem for improvement of functional design.
Support Vector Machine (SVM): The decision to identify the extent of the interest will be divided into different parts of the different group members.
Particle Swarm Optimization (PSO): The system is initialized with a population of random solutions and searches for optima by updating generations. However, unlike GA, PSO has no evolution operators such as crossover and mutation.
Memetic Algorithm (MA): A trying to resolve a complicated and continuous basis by the segregation of the population, the same as in groups, or in other words, is to develop a method used to search problems.
Ant Colony Optimization (ACO): An algorithm for finding optimal paths that is based on the behavior of ants searching for food. A similar approach can be used find near-optimal solution to the traveling salesman problem.
Intelligent System in Biomedical Engineering Applications
A Fastening Tool Tracking System Using an IMU and a Position Sensor with Kalman Filters and a Fuzzy Expert System
Seong-Hoon Peter Won, Farid Golnaraghi And Wael William Melek
From the Membership functions of Ang_vel and Acc_fluc , this system can identified the dynamic state of fastening tool in 3 state: Static, Quasi-static and Moving. To tracking a fastening tool by IMU and position sensor then use data of angular velocity and acceleration to be a set of fuzzy input.
Neural Networks Terrain Classification using Inertial Measurement Unit for an Autonomous Vehicle
Rubkwan Jitpakdee and Thavida Maneewarn
This research is an attempt to categorize the various surfaces of the 5 types of flat, rugged, grass, inclined and cannot be identified. Based on information from the gigantic sensitive, and then take the second IMU using fuzzy logic.
Systematic Improving the Performance of Semiconductor Gas Sensor using Fuzzy Artmap Neural Networks
Alexander Vergara and Eduard Llobet
They use IS techniques to investigated a single gases and multi-component mixture using semiconductor gas sensor arrays. The result shows that it has accuracy more than 90%.
A hybrid intelligent system for medical data classification
Manjeevan Seera and Chee Peng Lim
In this paper, the intelligent hybrid consisting of Fuzzy Min-Max neural networks, classification and regression trees. To be a decision support tool for classifying medical data to validate the results. Experiments have shown that the positive hybrid intelligent systems that are effective in the medical data classification.
Image Segmentation by Histogram Thresholding Using Fuzzy Sets
Orlando J. Tobias Member
Thresholding method in this paper is based on use image histogram to find similarity between grey levels and linguistic fuzzy set (B and W). Mathematical model of this paper is obtained from the fuzzy concept. However, this proposed method also has limitation, the applicability of this method is limited to images that satisfy an assumptions
What is fuzzy logic technique?
Fuzzy logic (FL) is a tool to assist in decision an approximate under uncertainty or the theory of fuzzy sets which relates to classified of objects the main reason to use this logic, because this technique has shown the ability to mimic the human mind. Fuzzy logic is different from Boolean logic. The idea is to extend the knowledge of the partial true by the fact that during the completely true to completely false. But in traditional logic is only true to false.
Fuzzy Logic system has a different approach to the problem of how to control or classification. This approach focuses on what the system should do more than try to simulate how it works.
Boolean Logic and Fuzzy Logic
What are the meaning of word?
Fuzzy: Designed to implement the principles of logic to be able to deal with imprecise information, or the extent of the uncertainty.
Uncertain: the idea of representing partial or incomplete information. By the theory of modal logic is possible with traditional probability.
Vagueness: the expansion of the natural language lacks clarity, the true nightmare for classical logic.
Example of fuzzy based application and its fuzzy if-then rules
In everyday life we can found navigating FL applied in various fields. Here is an example of an electric heater inside the house. By the electric heater will have thermometer to measure the temperature inside the room, if you set the temperature within the range of 0 to 40 degrees Celsius, which is the input of the FL. The system will decide how much temperature room that should has to be reduced function of the electric heater or not. The output of the system is considered to be a function of the heater from 0 to 100 (requiring level 50 as a medium).
Fuzzy Inputs: Cold, Normal, Warm Fuzzy Outputs: Reduce, Maintain, Heat
When fuzzy system has fuzzy inputs and fuzzy outputs so it should to write if-then rules. The antecedent part or premise is if-part of the rule while the consequent or conclusion is then-part of the rule. Rules of an example can be like this:
If Room Temperature is cold Then Heater is heat
If Room Temperature is normal Then Heater is maintain
If Room Temperature is warm Then Heater is reduce
5 main steps of fuzzy logic approach
1. Fuzzify input
Resolve all issues in a statement before the appearance of membership between 0 and 1. There is only one antecedent part, this is the level of support for the rule.
2. Apply a fuzzy operator
If there are many parts that will come before the use of logic and corrected before a single number between 0 and 1, which is the level of support for the rule.
3. Apply an implication operation
The level of support for all rules to shape the output fuzzy sets consequent fuzzy set fuzzy rules as for export. If it is only partly true, then the output fuzzy sets is truncated to the method definition.
4. Aggregate the outputs
All of fuzzy sets that represent the output of each rule and combines them into a single occasion. The output of the integration process is one of the fuzzy sets for each output variable.
5. Defuzzify output
Has been exporting a comprehensive range of fuzzy sets, we need to export it to move from one slide to set the output resolution.
Meaning and example of fuzzy input
First, let's understand the word crisp set and fuzzy set before. Crisp set means to specify the value of what is logic 0 or 1, such as tall man and short man. The difference is that fuzzy set would identify as probability in a decimal between 0 and 1. Such a man is 175 cm tall, which some might say he's tall but some may say that moderate so that identifies the full height in 1, but he may be at 0.8.
As already said in Fuzzy System is designed Input and Output in the form of a priority from 0 to 1, which is called the degree of membership and the relationship between each of us is called the membership function.
Fuzzy Inputs: Cold, Normal, Warm
Another example that is easy to understand on fuzzy input set is the temperature of the room. In crisp set, it specifies the exact temperature to 12 degrees Celsius that the temperature is cold (assigned a value of 0). In fuzzy input set, it is noted that the temperature of 12 degrees Celsius is likely to be a cold 50%, had a chance to be normal of 30% and a chance to be a warm 0%, which can be identified as membership function that
T = cold/0.5 + normal/0.3 + warm/0.0
Membership function, It’s not just triangular membership function but it also trapezoidal, Gaussian and sigmoidal.
Meaning and example of fuzzy output
Fuzzy output is to take the output of each rule together as one and then do the defuzzification to get the output on a crisp value. The fuzzy output will come from the aggregation fuzzy input by implement fuzzy operators and apply implication operator.
An example of Fuzzy Logic System is an electric heater. If the room temperature is 12 degrees Celsius, the membership function of the fuzzy input should be
T = cold/0.5 + normal/0.3 + warm/0.0
which can be converted into the fuzzy output of each rule as follows
Fuzzy input has a degree of membership of the cold set of 0.5, which is make the fuzzy output of heat set with the degree of membership of 0.5. Similarly, the degree of membership of the fuzzy input a normal set is 0.3, which is maintain set of fuzzy output with the degree of membership is 0.3. and a degree of membership of warm set is 0.0, which is maintain to reduce set of fuzzy output with the degree of membership is 0.0. If the fuzzy output of the 3 rule combined together, it would have come out the membership function of the fuzzy output like this
Action = heat/0.5 + maintain/0.3 + reduce/0.0
Fuzzy Inputs: Cold, Normal, Warm Fuzzy Outputs: Reduce, Maintain, Heat Centroid computation
And to get the power current of heater the fuzzy output must be defuzzification to get crisp value.
Defuzzification method: Centroid, Bisector, Middle of maximum, Largest of maximum and smallest of maximum.
Fuzzy Inference System (FIS)
A Fuzzy Inference System (FIS) is the actual mapping of the input to the output, using logic; FIS is a collection of functions fuzzy membership and rules rather than logic. Lean to reason about FIS rules in a fuzzy production rules of the form if - then rules. FIS is divided into two methods.
Mandani Method: The method most commonly used and It is easy to understand. After the aggregation and fuzzy output of each rule will put them together using max / min. They will be used as a single spike output membership function instead of a distribution. Sometimes we call this the singleton.
Sugeno method: not very popular, it is usually used with a linear and constant output use by multiple spike makes may seem complicated than Mandani.