冰裂隙检测雷达关键技术研究

Crevasses are vertical fractures in Antarctic ice sheets, which may be covered by snow and therefore present no surface expressions. Because the size of crevasses changes all the time and its location is also moving, which put great danger and difficulty to ground scientific exploration activities. In order to further enhance the security capability of polar scientific research in China, we will develop researches on frequency modulated continuous wave (FMCW) radar technology applied to crevasse detection. The gated FMCW technology is used to solve the isolation problem between the transmitter and receiver, ensuring the radar has a large dynamic range. A ramp signal is added to the forward pass of phase-locked loop (PLL), so that type ⅠPLL can be used to produce high linearity, ultra wideband FM continuous wave signal at the transmitter, which has higher loop gain and closed-loop bandwidth. Also, automatic gain control technology is used to ensure that the transmitted signal has high amplitude flatness. A compartmental feature detection and extraction method is proposed to realize crevasses detection automatically with high efficiency and high resolution. We also propose a crevasses feature recognition and learning method based on Sparse Principal Component Analysis (SPCA) method and elastic network to enhance the detection reliability and efficiency. Finally, a prototype of crevasses detection radar system will be developed.

冰裂隙作为潜藏在南极冰盖下的特殊冰缝，常常被浮雪掩盖，肉眼根本无法分辨，它的大小在时刻发生变化，位置也在不断地移动，给地面科学考察活动带来极大的危险和困难。为进一步增强我国极地科学考察的安全保障能力，课题组拟开展基于FMCW体制的超宽带冰裂隙检测雷达系统关键技术研究工作。采用门控调频连续波技术，解决系统收发之间的隔离问题，保证雷达具有大的动态范围；在锁相环前向通道中引入斜坡信号，得到更高的锁相环路增益和闭环带宽，产生超宽带、高线性度的调频连续波信号；在发射机实现自动增益控制，保证发射信号具有较高的幅度平坦度；提出分部式特征相关性检测和提取方法，实现对冰裂隙特征的自动、高效和高分辨率的检测和提取；提出基于稀疏主成份分析和弹性网络的冰裂隙特征识别和学习方法，提高冰裂隙自动检测识别的准确性和效率。在此基础上，完成一套冰裂隙检测雷达原理样机。

冰裂隙检测雷达凭借其穿透冰盖能力强、探测分辨率高和数据信息丰富等优点，在探测冰盖内部浅层结构、冰下空洞与裂缝方面具有突出的技术优势。本课题开展冰裂隙检测雷达关键技术和冰裂隙识别算法研究，为未来雷达实现南极科学考察常规化应用提供技术积累和支持。. 在雷达硬件技术方面，针对冰裂隙检测雷达对超宽带、低杂散、高线性度的调频连续波信号的需求，本课题提出了一种线性调频源的线性化技术，通过Ⅰ型锁相环以及在前向通道中附加的预失真补偿信号对压控振荡器进行实时补偿。针对调频连续波信号中残余的非线性相位，提出了一种基于差分滤波的非线性估计及校正方法。为保证发射信号幅度的带内平坦度，提出了一种增益自动控制的放大技术。在提高系统动态范围方面，研究了基于M序列的门控调频连续波技术，有效抑制了系统的直达波信号。在此基础上，研制了一套带宽2GHz、冰下分辨率优于5cm的冰裂隙检测雷达原理样机。. 在冰裂隙识别算法方面，针对冰裂隙特征检测和提取，提出了基于分部式高斯混合模型的冰下层位检测方法，实现了反映冰下介质层位属性差异的层位边界有效提取。针对冰裂隙识别和学习，提出了一种基于机器学习和深度学习的冰裂隙自动检测和分析算法，实现了对冰裂隙的自动、高效和准确识别。在此基础上，搭建了冰下层位检测和冰裂隙识别软件平台。. 课题组申请发明专利5项，其中已经获得授权3项，发表论文13篇，申请软件著作权1项。