作者：李大中, 赵杰

作者单位：华北电力大学自动化系, 河北 保定 071003

关键词：端点检测;果蝇算法;支持向量机;小波变换

摘要：

在超声缺陷识别系统中,端点检测是确保缺陷准确识别的重要环节。为提高在实际探伤过程中端点检测的准确率,提出一种以果蝇算法优化支持向量机的端点检测方法。针对超声检测信号的特点,采用小波包变换提取反映该信号性质的特征向量。鉴于传统方法检出率不高及支持向量机(SVM)参数难确定的问题,利用果蝇算法(FOA)优化SVM的惩罚子和核参数,提高支持向量机建模准确度。试验结果表明:FOA-SVM模型的平均检出率达到97.5%,端点检测效果明显优于传统的双门限法、普通SVM模型和GA-SVM模型。

Ultrasonic signal endpoint detection based on FOA optimized SVM

LI Dazhong, ZHAO Jie

Dept of Automation, North China Electric Power University, Baoding 071003, China

Abstract: Endpoint detection is an important step to ensure accurate identification in the ultrasonic defect recognition system. In order to improve the accuracy of endpoint detection during actual flaw detection, an endpoint detection method using fruit fly optimization Algorithm (FOA)-support vector machine (SVM) has been proposed. Based on the characteristics of ultrasonic detection signals, wavelet transform was applied to extract the feature vector that reflects the nature of these signals. As the common double-threshold method is low in detection rate and the parameters of the SVM are difficult to determine, the FOA was used to optimize the penalty factor and kernel parameter of the SVM to improve the precision of the SVM. The experimental results show that the average detection rate of FOA-SVM is 97.5%. The endpoint detection effect significantly outperforms that of the traditional double threshold method and common SVM and GA-SVM models.

Keywords: endpoint detection;FOA;SVM;wavelet transform

2016, 42(5): 103-106,123  收稿日期: 2015-10-15;收到修改稿日期: 2015-12-28

基金项目: 

作者简介: 李大中(1961-),男,内蒙古包头市人,教授,博士,研究方向为新能源发电系统控制、智能优化理论及应用、分布式新能源发电及冷电联产控制系统。

参考文献

[1] 刘琨,何畅,刘铁根,等. 一种用于光纤周界安防系统的端点检测方法[J]. 光电子·激光,2014,25(11):2136-2140.
[2] 卢志茂,金辉,张春祥,等. 基于HHT和OSF的复杂环境语音端点检测[J]. 电子与信息学报,2012,34(1):213-217.
[3] SHIN J, CHANG J, KIM N. Voice activity detection based on statistical models and machine learning approaches[J]. Computer Speech & Language,2010,24(3):515-530.
[4] 张铁威. 基于小波分析与神经网络的语音端点检测研究[J]. 电子测量与仪器学报,2013,27(6):528-534.
[5] 张晓雷,吴及,吕萍. 基于支持向量机与多观测复合特征矢量的语音端点检测[J]. 清华大学学报(自然科学版),2011, 51(9):1209-1214.
[6] WANG L, ZHENG X L, WANG S Y. A novel binary fruit fly optimization algorithm for solving the multidimensional knapsack problem[J]. Knowledge Based Systems,2013,48(2):17-23.
[7] 马强,荣命哲,贾申利. 基于振动信号小波包提取和短时能量分析的高压断路器合闸同期性的研究[J]. 中国电机工程学报,2015,25(13):149-154.
[8] WEI Y, ZHANG J. Application of finite-state vector quantization in speech endpoint detection[J]. Computer Engineering & Applications,2012,48(17):161-163.
[9] 马杰,刘璐璐,姚勇. 小波变换超高压输电线路雷电干扰识别判据研究[J]. 中国测试,2013,39(5):34-38.
[10] 康守强,王玉静,姜义成,等. 基于超球球心间距多类支持向量机的滚动轴承故障分类[J]. 中国电机工程学报,2014, 34(14):2319-2325.
[11] 王长喜,王海云,王维庆,等. 基于FOA和LS-SVM的风电场风速预测研究[J]. 可再生能源,2014,32(9):1312-1318.
[12] 郭海帆,陈建华,陈鑫友,等. 基于矢量网络分析仪功率测量的噪声系数测量方法[J]. 中国测试,2014,40(6):35-38.