以遺傳演算法鑑別結構接點參數

Identification of Structural Joint Parameters Using Genetic Algorithms

10.29948/JAE.200804.0002

楊大中(Ta-Chung Yang)；何信宗(Shin-Tzung Ho)；林崇賢(Chorng-Shyan Lin)

接點鑑別 ； 自然頻率 ； 頻率響應函數 ； 遺傳演算法 ； genetic algorithm ； parameter identification ； joint ； frequency response function

先進工程學刊

3卷2期（2008 / 04 / 01）

105 - 111

繁體中文

複雜結構是由許多結構透過接點來連接所組合而成，這些接點的動態特性影響結構機械動態性能。接點不易準確建模分析，接點參數鑑別程序在分析上相當複雜，易形成一病態數值系統，及受雜訊干擾。遺傳演算法來自於生物自然演化，對於搜索全域空間之最佳解有很好的成效，具有相當好的強健性。本文利用結構的頻率響應函數資訊與遺傳演算法，針對一單邊支撐樑之螺栓接點參數進行鑑別。首先，本文利用結構的自然頻率鑑別接點勁度參數，並且將鑑別結果與子結構合成法比較，再加入結構的振幅與相位資訊來鑑別接點阻尼參數。當使用的資訊越多時，鑑別的準確性有明顯提升。本文採取兩階段設定群體搜索範圍之方式以增進參數鑑別之精確度。

Dynamic characteristics of complicated structures are greatly affected by various joints where the components are connected with each other. However, the joints are very difficult to model adequately and accurately. The parameter identification procedures of the joints are numerically complicated, ill-conditioned, and very sensitive to noise. Genetic Algorithm (GA), which imitates the evolution of nature, seems a very promising means in dealing with these identification problems. In this paper, the parameter identification procedure is carried out by implementing the Genetic Algorithm (GA), and using the Frequency Response Function (FRF) of the structure in the objective functions. First, the joint stiffness parameters of translation and rotation are identified using the natural frequency information and compared with those identified by substructure synthesis method. Damping parameters of the joint, then, are identified by using amplitudes and phase angles of FRFs. The identification of the joint parameters of a cantilever beam bolted at one end is conducted in this paper. The results show that the accuracy of identification increases if more FRFs information is used in the objective functions of the identification procedure. A two-stage strategy of defining search regions is used in this paper to promote the accuracy of the identified parameters.