國立高雄大學圖資館 |

Language: English

Back

Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation /Shaoze Fan.
Author:	Fan, Shaoze,
Description:	1 electronic resource (92 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 87-02, Section: B.
Notes:	Advisors: Li, Jing Committee members: Oria, Vincent; Musialski, Przemyslaw; Zhang, Shuai; Zhang, Xin.
Contained By:	Dissertations Abstracts International87-02B.
Subject:	Computer science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31841217
ISBN:	9798290935478

Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation /
Fan, Shaoze,

Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation /Shaoze Fan. - 1 electronic resource (92 pages)

Source: Dissertations Abstracts International, Volume: 87-02, Section: B.

This dissertation presents a comprehensive automated framework for power converter design, leveraging reinforcement learning (RL) and graph-transformer networks (GTN) to address critical inefficiencies in traditional manual topology optimization. Motivated by the combinatorial increase of circuit design spaces and the computational cost of iterative simulations, this work develops a robust framework for generating energy-efficient topologies requiring rapid and reliable circuit design. The framework integrates three key components: (1) an upper-confidence-bound-tree-based (UCT-based) RL model for circuit topology space exploration, (2) parallelized UCT algorithms to accelerate exploration processes, (3) a Graph-Transformer-based Network enabling fast circuit performance evaluation. Experimental validation demonstrates the whole framework achieves great effectiveness. The UCT-based topology generation achieves 67% faster convergence than baseline methods. The parallelization scheme achieves a 9.7x speedup with little performance loss. Moreover, the GTN model reduces total optimization time by 98.2% when replacing simulations. This dissertation advances circuit design automation through three innovations: (1) RL-based topology generation with offline-data-informed default policies, (2) parallel UCT for distributed circuit topology exploration, and (3) a GTN surrogate model resolving data discontinuity in topology spaces. This research represents a significant advancement in the field of power converter design automation. By combining reinforcement learning, parallel exploration techniques, and a graph-based surrogate model, the highly efficient framework can generate near-optimal circuit topologies with reduced computational costs. The experiment results demonstrate that this approach can accelerate the design process while ensuring high-quality results. Future work includes extending the framework to broader analog circuit classes and integrating active learning for adaptive GTN refinement.

English

ISBN: 9798290935478Subjects--Topical Terms:

199325
Computer science.
Subjects--Index Terms:

Circuit design automation

Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation /
LDR:03521nmm a22004213i 4500 001 693668
005 20260511063008.5
006 m o d
007 cr|nu||||||||
008 260819s2025 miu||||||m |||||||eng d
020 $a 9798290935478
035 $a (MiAaPQD)AAI31841217
035 $a AAI31841217
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Fan, Shaoze, $e author. $0 (orcid)0000-0003-3496-6076 $3 1010263
245 1 0 $a Model-Based Reinforcement Learning and Deep Learning for Power Converter Circuit Design Automation / $c Shaoze Fan.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (92 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 87-02, Section: B.
500 $a Advisors: Li, Jing Committee members: Oria, Vincent; Musialski, Przemyslaw; Zhang, Shuai; Zhang, Xin.
502 $b Ph.D. $c New Jersey Institute of Technology $d 2025.
520 $a This dissertation presents a comprehensive automated framework for power converter design, leveraging reinforcement learning (RL) and graph-transformer networks (GTN) to address critical inefficiencies in traditional manual topology optimization. Motivated by the combinatorial increase of circuit design spaces and the computational cost of iterative simulations, this work develops a robust framework for generating energy-efficient topologies requiring rapid and reliable circuit design. The framework integrates three key components: (1) an upper-confidence-bound-tree-based (UCT-based) RL model for circuit topology space exploration, (2) parallelized UCT algorithms to accelerate exploration processes, (3) a Graph-Transformer-based Network enabling fast circuit performance evaluation. Experimental validation demonstrates the whole framework achieves great effectiveness. The UCT-based topology generation achieves 67% faster convergence than baseline methods. The parallelization scheme achieves a 9.7x speedup with little performance loss. Moreover, the GTN model reduces total optimization time by 98.2% when replacing simulations. This dissertation advances circuit design automation through three innovations: (1) RL-based topology generation with offline-data-informed default policies, (2) parallel UCT for distributed circuit topology exploration, and (3) a GTN surrogate model resolving data discontinuity in topology spaces. This research represents a significant advancement in the field of power converter design automation. By combining reinforcement learning, parallel exploration techniques, and a graph-based surrogate model, the highly efficient framework can generate near-optimal circuit topologies with reduced computational costs. The experiment results demonstrate that this approach can accelerate the design process while ensuring high-quality results. Future work includes extending the framework to broader analog circuit classes and integrating active learning for adaptive GTN refinement.
546 $a English
590 $a School code: 0152
650 4 $a Computer science. $3 199325
650 4 $a Electrical engineering. $3 454503
653 $a Circuit design automation
653 $a Graph neural network
653 $a Parallel computing
653 $a Reinforcement learning
690 $a 0984
690 $a 0800
690 $a 0544
710 2 $a New Jersey Institute of Technology. $b Department of Computer Science. $3 886529
720 1 $a Li, Jing $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 87-02B.
790 $a 0152
791 $a Ph.D.
792 $a 2025
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31841217