基于BEMD-IPSO-SVM的扣件完损状态检测

BEMD-IPSO-SVM-based fasteners status detection

扣件的完损状态关乎铁路系统的安危,而传统检测算法运算复杂且精度不足,为进一步提升检测性能,提出基于BEMD-IPSO-SVM的扣件完损状态检测算法。该算法首先对初始化的扣件图像进行二维经验模态分解,提取固有模态函数的频谱特征,通过改进粒子群算法优化支持向量机来实现检测分类,达到了简化运算,增强泛化性,提升识别准确度的目的。通过实验仿真得出平均检测准确率可达95.15%,证明该算法在扣件检测方面切实可行。

The damage status of the fastener is related to the safety of the railway system.However,the traditional detection algorithm is complex and inaccurate in calculation.To further improve the detection performance,a BEMD-IPSO-SVM-based fastener damage detection algorithm is proposed.The algorithm firstly performs BEMD on the initial fastener images,then extracts the frequency spectrum characteristics of the IMF,finally implements detection classification through the IPSO-SVM,which reaches the goal of simplifying calculations,enhancing generalization,and improving recognition accuracy.Through experimental simulation,the average detection accuracy rate can reach 95.15%,which proves that the algorithm is feasible in fastener detection.