论文部分内容阅读
主要研究了基于模式识别的分布式光纤光学传感器的抗干扰控制。首先,通过理论分析分别建立了光信号模型及扰动(包括人为入侵与风扰动)模型,并将二者结合得到在实际应用环境下的人为入侵信号与风扰动信号,且通过仿真方法将人为入侵与风扰动加以区分。其次,根据上述仿真结果研究了几种特征向量提取方法。然后应用提取的特征向量,通过支持向量机的方法基本完成了模式识别,精度高达80%。最后提出了一种混合的支持向量机算法,从而将模式识别的精度提高到98.7%,并通过在线测试验证了该支持向量机算法的有效性。
Mainly based on pattern recognition distributed optical fiber optical sensor anti-jamming control. Firstly, the model of optical signal and the models of disturbance (including man-made invasion and wind disturbance) are established through theoretical analysis respectively, and the two are combined to obtain the man-made invasion signal and wind disturbance signal under the actual application environment. The artificial intrusion Distinction from wind disturbance. Secondly, several eigenvector extraction methods are studied based on the above simulation results. Then, the extracted feature vectors are used to achieve the pattern recognition basically by the method of support vector machine with the precision up to 80%. Finally, a hybrid support vector machine (SVM) algorithm is proposed to improve the accuracy of pattern recognition to 98.7%. The validity of the SVM algorithm is verified by online testing.