基于BP神经网络的聚酯玻璃钢加速老化寿命预测  

Accelerated Aging Life Prediction for GFRP Based on BP Neural Network

作　　者：张小锋[1] 郑冉[1] 孟江燕[1] 李超[2] 

机构地区：[1]无损检测技术教育部重点实验室(南昌航空大学),南昌330063 [2]成都飞机工业集团公司复合材料厂,成都610092

出　　处：《失效分析与预防》2011年第2期75-79,共5页Failure Analysis and Prevention

基　　金：江西省自然科学基金(2009GZS0084);航空检测与评价航空科技重点实验室资助项目(ZK200929002)

摘　　要：采用BP神经网络对聚酯玻璃钢氙弧灯加速老化的弯曲寿命进行了预测。通过对聚酯及其玻璃钢的人工氙弧灯加速老化,测试其不同老化时间的弯曲强度,对弯曲强度与老化时间进行BP神经网络的建模分析,借助MATLAB软件对聚酯玻璃钢的使用寿命分别进行分析与预测,并采用最小二乘法对所预测的结果进行了对比。结果表明:在以弯曲强度达到初始强度值的一半作为失效条件下,聚酯的氙灯老化寿命为813 d,含填料玻璃钢老化寿命为1 031 d,无填料玻璃钢老化寿命为1 065 d,说明BP神经网络可以预测玻璃钢的老化寿命,预测结果与最小二乘法预测结果误差不大于8%,而且预测结果与该材料性能的实际情况相符。The bending life of polyester glass fiber reinforced polyester（GFRP） under xenon arc lamp accelerated aging was predicted by BP neural network.The relationship between aging time and bending strength was tested.And models on the relationship between aging time and bending strength were built by BP neural network.Based on MATLAB,the aging life of polyester FRP was analyzed and forecast.The simulation results were compared with those by least square method.The results show that if the decrease in the bending strength by 50% is considered as failure,then the aging life of polyester is 813 d,that of GFRP with filler is 1 031 d and the GFRP without filler is 1 065 d.It is shown that the BP neural network can predict the aging life of FRP.The error between the predicted results by the BP neural network and those by least square method is less than 8%,and the predicted results is in accordance with the practical change of the material property.

关 键 词：BP神经网络 聚酯玻璃钢 氙弧灯加速老化 寿命预测 

分 类 号：TQ327.1[化学工程—合成树脂塑料工业]