Loading...

Discrete Hardware Neural Networks for Civil Engineering Application

Avestakh, Saber | 2010

697 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: English
  3. Document No: 41048 (53)
  4. University: Sharif University of Technology, International Campus, Kish Island
  5. Department: Science and Engineering
  6. Advisor(s): Joghataie, Abdolreza
  7. Abstract:
  8. In recent years considerable effort has been made to advance the hardware Artificial Neural Networks where in most cases, many neurons are placed on a VLSI chip. This research attempts to build individual neurons and connect them to build a hybrid of analog and digital neural network. In fact, every neuron is an AVR microcontroller which has a number of inputs and outputs. The data transfer between neurons are done by both analog (PWM- ADC) and digital (UART). In the first part, the necessary voltage source and programmer and how to build are discussed. Next A\D convertor, PWM technique, UART and their usage in this project are demonstrated. After that, the neurons are calibrated to improve the accuracy of the network and decrease the errors and then some applications of neural networks in civil engineering are mentioned. At end of this dissertation a methodology for predicting compressive strength of concrete with suitable workability in hardware neural network is presented. For this aim, a computer program was developed in Matlab. Using this program, a neural network model with two hidden layers was constructed, Trained and tested using the available test data of 208 different concrete mix-designs of concrete gathered from the web page of UCI (Machine Learning Repository). After this step weights and construction of network from Matlab are extracted and are implemented them to hardware neurons and then test the accuracy of the hardware in this issue
  9. Keywords:
  10. Artificial Neural Network ; AVR Microcontroller ; Pulse Width Modulation (PWM) ; Hardware Neuron ; Concrete Compressive Strenght

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC