Optimization Of Analog Circuit Design Using Artificial Intelligence : Genetic Algorithms

Wong , Yan Chiew (2010) Optimization Of Analog Circuit Design Using Artificial Intelligence : Genetic Algorithms. Project Report. UTeM, Melaka , Malaysia. (Submitted)

PDF (Full Text) Optimization_of_analog_circuit_design_using_artificial_intelligence_FULL_TEXT.pdf - Submitted Version Restricted to Registered users only Download (13MB)

Abstract

Analog circuits, while being replaced by digital circuits in many cases, remain play an important part in the integrated circuit (IC) terminalogy. Analog design is very challenging and has traditionally been performed by specialists with wealth of experience and intuition. Analog circuit requires tedious design work and much of the design time occurred to fix the selected candidates to meet the specification. Recently, much progression has been made in automating the analog circuit design synthesis. Genetic Algorithms (GA) which offers various significiant advantages over other traditional search tools is a better alternative tool for highly non-linear circuit optimization. In this research, optimization software is developed based on Improved Non-Dominate Sorting Genetic Algorithms (NSGA_2) which performs better in multi objectives optimization as well as in exporation and exploitation which are not able performed in Simple Genetic Algorithms (SGA). The developed software, Genetic Circuit Optimizer (GCO) optimizes the power consumption, gain and size of the transistors in circuit under test (CUT). The GCO is developed based on auto data handling between circuit simulator and GA in the environment of Matrix Laboratory (MATLAB). The GCO is validated by using Inverter Circuit, four stage Amplifier Circuit and Operational Transconductance Amplifier circuit as the CUT. The users can select high speed, accuracy or window simple mode industrial standard circuit simulators to run with GCO according to their design requirement. The verification of the GCO is proven with statistical analysis.

Actions (login required)

View Item