A schlieren detection algorithm is proposed for the ground-to-air background oriented schlieren(BOS) system to achieve high-speed airplane shock waves visualization. The proposed method consists of three steps. Firstl...A schlieren detection algorithm is proposed for the ground-to-air background oriented schlieren(BOS) system to achieve high-speed airplane shock waves visualization. The proposed method consists of three steps. Firstly, image registration is incorporated for reducing errors caused by the camera motion.Then, the background subtraction dual-model single Gaussian model(BS-DSGM) is proposed to build a precise background model. The BS-DSGM could prevent the background model from being contaminated by the shock waves. Finally, the twodimensional orthogonal discrete wavelet transformation is used to extract schlieren information and averaging schlieren data. Experimental results show our proposed algorithm is able to detect the aircraft in-flight and to extract the schlieren information. The precision of schlieren detection algorithm is 0.96. Three image quality evaluation indices are chosen for quantitative analysis of the shock waves visualization. The white Gaussian noise is added in the frames to validate the robustness of the proposed algorithm.Moreover, we adopt two times and four times down sampling to simulate different imaging distances for revealing how the imaging distance affects the schlieren information in the BOS system.展开更多
文摘A schlieren detection algorithm is proposed for the ground-to-air background oriented schlieren(BOS) system to achieve high-speed airplane shock waves visualization. The proposed method consists of three steps. Firstly, image registration is incorporated for reducing errors caused by the camera motion.Then, the background subtraction dual-model single Gaussian model(BS-DSGM) is proposed to build a precise background model. The BS-DSGM could prevent the background model from being contaminated by the shock waves. Finally, the twodimensional orthogonal discrete wavelet transformation is used to extract schlieren information and averaging schlieren data. Experimental results show our proposed algorithm is able to detect the aircraft in-flight and to extract the schlieren information. The precision of schlieren detection algorithm is 0.96. Three image quality evaluation indices are chosen for quantitative analysis of the shock waves visualization. The white Gaussian noise is added in the frames to validate the robustness of the proposed algorithm.Moreover, we adopt two times and four times down sampling to simulate different imaging distances for revealing how the imaging distance affects the schlieren information in the BOS system.
文摘由于不同井间工况差异显著,异常振动特征分布存在跨井不一致性,传统基于单井数据的监测方法难以适应跨井场景。为此,以黏滑振动为例,对不同工况下的黏滑振动数据特征进行了对比分析,提出了一种结合深度判别迁移学习网络(domain adaptive transfer learning network,DDTLN)与BO⁃Transformer⁃LSTM的跨井异常振动识别方法。将近钻头振动数据输入到DDTLN模型中,通过卷积层与改进的联合分布自适应(IJDA)机制减小域间特征差异,实现跨域特征提取;将提取的特征输入到BO⁃Transformer⁃LSTM模型中挖掘时序信息,实现跨井高效分类。试验结果表明:不同工况下井间振动信号差异显著,传统方法跨域分类效果较差;经过DDTLN处理后,不同域间的数据特征有了很好的对齐,跨域识别准确率高达91.5%;DDTLN⁃BO⁃Transformer⁃LSTM模型能够有效解决跨井识别问题,分类准确率最高达96.7%,显著优于传统单井识别方法,具有更好的泛化能力。该研究可为跨井场景下的井下异常振动识别提供新思路。
