複材層板微破壞的音洩訊號辨識研究

Abstract

熱塑性複材因為樹脂流動性緩慢，纖維含浸較差，容易產生纖維斷裂、纖維母材剝離及橫向母材裂縫等微破壞。過去有關複材內部微破壞音洩訊號辨識的相關研究，大都以共振式探頭擷取音洩訊號，並以這些訊號的時減特徵為音洩訊號辨識的相關參數。但是使用共振式探頭會犧牲材料破壞的頻率特性。複材層板具有方向性，音洩波傳與方向及頻率相關，本研究針對大尺寸之熱塑性玻璃纖維複材層板設計拉伸彎曲實驗，產生特定的微破壞機制，由寬頻壓電探頭分別在材料的0°及90°方向偵測試片內部所發生的音洩訊號，應用小波轉換將音洩訊號轉換至時頻域，然後由圖形識別的技巧將某些特徵參數輸入至倒傳遞類神經網路學習，並以學習後的網路辨識玻纖複材內部的微破壞音洩訊號，獲得不錯的辨識結果。

Tliermo-plastic composite materials commonly have weak fiber/matrix bonded strength due to their poor resin penetration. Several micro-defects, including fiber-breakage, fiber/matrix debonding, transverse matrix cracking, are frequently observed in thermo-plastic composites. Most previous researches characterized acoustic emission (AE) in composites often by temporal parameters. The frequency characteristics were also ignored from AE signals captured by resonant transducers. The propagation of AE waves in composite laminates is complex because of the anisotropy and dispersion. In this study several micro-fractures were induced in large-sized laminates made of glass fiber-reinforced, thermo-plastic composite during well-designed tensile and bending tests. The fractures were monitored by two broadband piezoelectric AE sensors mounted on surfaces of the specimens at 0° and 90° orientations to the fibers. The time-frequency patterns of those AE signals were determined by the continuous wavelet transform. Pattern recognition was carried out by back-propagation neural network through the characteristic parameters in frequency domain. A very good results of the AE signal recognition have been achieved.