基于混合核SPSO-SVM飞机复合材料健康监测研究

发布时间：2018-01-19 02:07

本文关键词： 电阻抗谱自适应粒子群算法混合核函数支持向量机复合材料健康监测　出处：《现代防御技术》2016年04期 　论文类型：期刊论文

【摘要】：针对复合材料异常检测或健康监测的问题,提出了一种改进的粒子群算法训练混合核函数支持向量机,并用其进行复合材料健康状态辨别。通过有限元分析软件ANSYS14.5模拟碳纤维复合材料样板,获取了复合材料样板不同损伤类型的阻抗谱,提取了不同类型的阻抗特征参数,构建训练样本对混合核函数支持向量机进行训练,再将没有进行训练的阻抗特征数据送入到训练好的混合核函数支持向量机进行复合材料构件的健康状态辨别。结果表明:相比于Cole-Cole曲线分段各段电阻抗实部或虚部幅值的平均值,选取复合材料Cole-Cole曲线分段线性拟合斜率作为特征参数时,混合核函数支持向量机具有更高的健康状态辨识准确率。复合材料样板健康状态辨识仿真实验表明,该方法具有较高的健康辨识精度。
[Abstract]:An improved particle swarm optimization (PSO) algorithm for training hybrid kernel function support vector machines (HSVM) is proposed to solve the problem of composite anomaly detection or health monitoring. By using the finite element analysis software ANSYS14.5 to simulate the carbon fiber composite sample, the impedance spectrum of different damage types of composite material sample was obtained. Different types of impedance characteristic parameters are extracted and the training samples are constructed to train the hybrid kernel function support vector machine. Then the untrained impedance characteristic data are fed into the trained hybrid kernel function support vector machine to identify the health status of composite components. The results show that:. Compared with the average amplitude of the real or imaginary part of the impedance in the segmented sections of the Cole-Cole curve. When the slope of composite Cole-Cole curve is selected as the characteristic parameter. Hybrid kernel support vector machine has higher accuracy of health state identification, and the simulation results of composite sample health state identification show that this method has higher health identification accuracy.
【作者单位】：空军航空大学;
【分类号】：V267;TP18
【正文快照】： 0引言随着科学技术的不断发展,先进复合材料(如碳纤维复合材料)在先进飞机上的用量越来越大[1]。随飞机服役年限的不断增加,复合材料构件会产生裂纹、气泡、内部分层等损伤,严重影响飞行安全[2]。电阻抗谱(electrical impedance spectrosco-py,EIS)法作为一种非侵入式的检测手

【相似文献】