利用FPGA電路實現演化設計

Abstract

電子電路的設計通常是一個複雜的工作，而且需要大量相關背景知識。隨著演化式演算法和人工智慧的進步，我們能夠利用生物在自然界演化的法則，使得在設計的過程中，不必太仰賴人們的經驗，而僅需事先想定一些適當的條件，讓電路本身自然的演化，達成優良的創新設計。本文主要是利用演化式演算法來設計數位電路；包括算術、和組合電路，然後檢討演化所得到的電路結果和傳統設計及利用VHDL設計的不同。本文所指的演化式演算法是指基因演算法。當電路被演化完成後，再下載至FPGA的電路板，驗證所得到的電路是否運作正確。此外，也利用基因演算法從事汽車懸吊機構的設計，在要求能量流值最小的原則下，我們成功地演化出數種新穎的機構。

The design of electronic circuits is generally a complex task requiring knowledge of large collections of domain-specific rules. With the advent of evolutionary algorithm and artificial intelligence, it is feasible to employ evolution principles in biology to assist in the design process, not so rely on the experience of human beings, and prescribe appropriate conditions that allow solution to evolve. This study aims to carry out evolutionary algorithms for digital circuit creative design, including arithmetic, and combinational circuits, and compare the evolution circuits with results from conventional design and VHDL design. The evolutionary algorithm used in this study is Genetic Algorithms (GA). Once the circuits are evolved, they are downloaded to Field Programmable Gate Array demonstration boards for verification. In addition, this study also uses GAs to design vehicle suspensions. Based on minimum flow values, this study devises novel suspension mechanisms.