论文部分内容阅读
为解决电磁逆散射问题,提出了一种实时逆散射方法,该方法利用支持向量机(SVM)将逆散射问题转化为一个回归估计问题.基于SVM的电磁逆散射方法成功地解决了逆散射问题中的非线性和不适定性.利用穿墙问题测试了该方法的可行性和有效性,测试结果表明,不论是无噪声还是有噪声的情况,该方法都能很好地对墙后目标进行探测与定位.此外,在穿墙环境下用SVM预测模型讨论了接收天线的采样位置数对预测结果的影响.最后对多源设置下的预测误差进行了分析和研究,研究表明,相比于单源情况多源设置有利于对墙后目标的识别.
In order to solve the electromagnetic backscattering problem, a real-time backscattering method is proposed, which uses the support vector machine (SVM) to transform the inverse scattering problem into a regression estimation problem.The electromagnetic backscattering method based on SVM has successfully solved the inverse scattering problem In the non-linear and ill-posed. The feasibility and effectiveness of this method are tested by using the wall penetration test. The test results show that this method can detect the target behind the wall well, no matter whether there is no noise or noises And positioning.In addition, the influence of the sampling position number of the receiving antenna on the prediction result is discussed by using the SVM prediction model in the through-wall environment.Finally, the prediction error under the multi-source setting is analyzed and studied. The research shows that compared with the single Source multi-source setting is conducive to the recognition of the target behind the wall.