利用類神經網路與遺傳演算法於槽孔消音器之最佳化設計

Optimum Design of Perforated Mufflers by Using Neural Network and Genetic Algorithms

鄭牧民(Mu-min Cheng)

指導教授 ： 張英俊

大同大學/工程學院/機械工程學系所/碩士(2007年)

摘要

本研究結合類神經網路系統（Neural network system）與遺傳演算法（Genetic algorithm, GA）之最佳化方法，於槽孔消音器之尺寸最佳化設計。在類神經網路之網路模式建構中，針對槽孔消音器單一頻率分為500Hz、1000Hz以及2000Hz三種不同目標頻率之案例，其輸入組與輸出組數據利用類神經網路建立其網路模式，再結合遺傳演算法搜尋消音器之最佳化尺寸及聲音傳輸損失（Sound of transmission loss, STL）。最後將類神經網路與遺傳演算法所設計之最佳化尺寸，代入消音器之數學理論模式求出相對的STL值。
研究結果顯示，使用類神經網路建立槽孔消音器網路模式，不但槽孔將消音器針對單一頻率之數學模式簡單化，而之後遺傳演算法所得的最佳化尺寸及其STL值與數學理論正解之STL值比較之下誤差極小，且各類消音器之消音性能均有增加10到25dB，並以共鳴型消音器針對1000Hz最佳化尺寸為最，其消音性能增加了26dB以上，表示類神經網路搭配遺傳演算法在槽孔消音器之尺寸最佳化設計上也有極佳的表現。對於工業界之應用，可有效的減少開發設計時間，加速產品開發與降低錯誤率，更可降低其開發設計時所需的費用。

關鍵字

類神經網路系統 ； 消音器尺寸最佳化設計 ； 遺傳演算法

並列摘要

In this research, the combination of neural network and genetic algorithm (GA) is applied to the optimal design of perforated mufflers. The target frequencies in the optimization process of perforated muffler are 500Hz, 1000Hz, and 2000Hz, respectively. The mathematical model of the muffler is built by means of input data and output data by neural network. Then GA is used to search the dimensions of optimum muffler and sound transmission loss (STL). Finally, substituting the optimal design into the transfer matrix, and deriving STL. The results show that the network model of perforated muffler that is built by neural network simplifies the theoretical analysis, the difference between the exact solutions and STL of optimal muffler is quite small, and can enhance STL efficiently. It is believed that the optimum algorithm proposed in this study can save the cost in developing silencers in industry.

並列關鍵字

Genetic algorithm ； Optimal design of perforated muffler ； Neural network

參考文獻

[5] M. L. Munjal, “Velocity Ratio-cum-transfer matrix method for the evaluation of a muffler with mean flow”, J. Acoust. Soc. Am., 39, pp.105-119, 1957.

[6] Joseph W. Sullivan, Malcolm J. Crocker, “Analysis of concentric-tube resonators having unpartitioned of cavities a)”, J. Acoust. Soc. Am., 64(1), pp.207-215, July, 1978

[7] Joseph W. Sullivan, “A method for modeling perforated tube muffler components. I. Theory a)”, J. Acoust. Soc. Am., 66(3), pp.772-778, Sept., 1979.

[8] Krishnamurthy Jayaraman, Kit Yam, “Decoupling approach to modeling perforated tube muffler components”, J. Acoust. Soc. Am., 69(2), pp.390-396, Feb., 1981.

[9] M. L. Munjal, K. N. Roat, A. D. Sahasrabudhe, “Aeroacostic analysis of perforated muffler components”, Journal of Sound and Vibration, 114(2), pp.173-188, 1987.