提升硬碟傳輸效率減振最佳化設計

由於硬碟的容量越做越大，硬碟也越趨精密，故容易受到振動影響傳輸效率，尤其是在硬碟傳輸中的狀態下，甚至會產生損壞，而本研究是將硬碟安裝在振動試驗機上模擬外在環境下，磁碟系統受到不同角度、不同振動頻率及不同振動加速度之外力振動，並透過將硬碟連接電腦以IOmeter軟體測試硬碟振動狀態下之傳輸效率，以傳輸效率高低來判斷振動對硬碟影響的嚴重程度。並結合田口法(Taguchi methods)、類神經網路系統(Neural network system)與遺傳演算法(Genetic algorithm, GA)之最佳化方法，搭配實驗針對減振墊片硬度作最佳化設計。本文分成二部分：一、墊片減振性能測試：針對不同減振墊片硬度、厚度與不同載具角度測試讀取效率，再列出圖表分別對(a)只改變墊片硬度，(b)只改變載具角度，(c)只改變墊片厚度三種條件做比較；二、以實驗搭配田口法、類神經網路系統與GA演算法針對由實驗得到讀取效率較低之頻率為目標頻率，做單一振動頻率的硬碟最佳讀取效率演算，其輸入組與輸出組數據利用類神經網路建立數學模式，再結合遺傳演算法搜尋減振墊片最佳化硬度與傳輸效率，並將此方法所得結果以實驗與最佳演算前結果做比對。結果顯示使用類神經網路建立網路模式並結合遺傳演算法所得的最佳化墊片硬度以實驗測試後，皆較原始讀取效率提升15 IOps以上，由此結果可證明類神經網路搭配遺傳演算法在防振設計實務上的可行性。

關鍵字

遺傳演算法；最佳化設計；硬碟；振動；減振墊片； IOmeter

並列摘要

As hard disks continue to grow in capacity, they are becoming more and more delicate. Thus, data transfer rates are easily affected by vibration; hard disks are especially vulnerable when they are transferring data. This study places a hard disk on a vibration tester to simulate the external environment; the hard disk is put through different vibration frequencies, acceleration and angles. The hard disk is linked to a computer and how much it is affected by vibration is determined based on the data transfer rate by use IOmeter software. Taguchi methods, neural network system and genetic algorithm (GA) optimization methods are applied along with an experiment to find the optimum design for anti-shock pad hardness. This study is divided into two parts: 1. Anti-shock pad performance test: Data transfer rate is tested for anti-shock pads of different hardness, thickness and at different angles; data transfer rates are charted for when (a) only the hardness is changed, (b) only the angle is changed and (c) only the thickness is changed. 2. The frequency for when the lowest data transfer rate was measured is used as the target frequency, and then Taguchi methods, neural network system and GA is applied to find the optimum data transfer rate for a single vibration frequency. A mathematical model is constructed using neural network system based on input and output figures, and then GA is applied to find the optimum hardness of anti-shock pads, as well as the optimum data transfer rate. The resulting optimal hardness is compared with the optimal hardness calculated before conducting the experiment. According to results, for the optimum anti-shock pad hardness found using the network model created from neural network system and GA, reading speed was at least 15 IOps higher than the original reading speed. This result proves the feasibility of applying neural network system and GA in the practice of anti-shock design.

並列關鍵字

Optimum Design ； anti-shock pad ； IOmeter ； Genetic Algorithm (GA) ； vibration ； hard disk

參考文獻

[3] 陳浩然，2002，元智大學機械工程研究所，元智大學碩士論文，磁碟振列機之振動分析

[7] 康智豪，2007，大同大學機械工程研究所，大同大學碩士論文，硬碟載具之減振設計之研究

[9] 游永豐，1999，元智大學機械工程研究所，元智大學碩士論文，電子裝置衝擊減振之研究

[11] Tsai, M. S., W. H. Yuan and J. M. Chang, 2007, The 13th International Congress on Sound and Vibration, July, Cairns, Australia, Active Vibration Control of Hard-disk Drives using PZT Actuated Suspension Systems.

[15] 徐秀枝，2008年11月，第三十二屆全國力學會議，H015，磁碟系統在不同傾斜角度下承受基底振動之傳輸效率研究

國際替代計量

提升硬碟傳輸效率減振最佳化設計

未授權

主題瀏覽