Font Size: a A A

Research On Application Of Artificial Neural Networks In CAD Of Microwave Circuits

Posted on:2010-05-07Degree:MasterType:Thesis
Country:ChinaCandidate:S Y YangFull Text:PDF
GTID:2178360278459265Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Complexity and precision in circuit modeling, time consumption, and capability in deviation-tolerance analysis and in optimization are all important factors to be considered in microwave-circuit design, apart from agreeing with electrical specifications of circuits. Recent years, Artificial Neural Network (ANN) technology has been introduced into microwave-circuit CAD and optimization. Practices have shown that ANN can help improvement of precision of modeling with high efficiency and flexibility as a better alternative to traditional design methods for microwave-circuit CAD and pave a new way for passive/active components and circuits design.MATLAB is the most popular simulation tool in microwave-circuit CAD. However, its low processing speed and low efficiency plus limitation in supporting microwave ANN make it unsatisfactory for microwave-circuit design, although it is good at matrix processing and digital analysis. Consequently, there is a need for a dedicated microwave-circuit CAD tool with features like ANN-based modeling capability, easy-to-use, high-efficiency and etc.Backed by a circuit-design project of key laboratory of Microfabrication Technology, at Institute of Optics and Electronics, Chinese Academy of Sciences, the author has devoted her efforts to research on applying ANN technology to microwave-circuit CAD. As a result, an ANN support-tool called the ANNMod, for microwave-circuit CAD was developed. Compared with NNTool in MATLAB, it is faster, more efficient, and easier to use with a friendly user interface. Results of applying ANNMod to current-voltage (I-V) modeling for FET (Field Effect Transistor) and simulation of transmission-characteristics of spur-lines have shown that the response time to circuit inputs for a trained ANN, only ranges from a few mini-seconds to a few seconds. This has greatly improved circuit design efficiency. Furthermore, test results of an experimental device called "Open-loop resonance band-pass filter", have shown conformity to simulation results. This dissertation also provides a comprehensive discussion on modeling and key technique in microwave-circuit design and provides improved ANN BP and RBF algorithms, which have been adopted in ANNMod.
Keywords/Search Tags:Microwave Circuits CAD, ANN, BP, RBF, Open-loop Resonance Band-pass Filter
PDF Full Text Request
Related items