CAJ | 학술논문

为解决蕴含有复杂变形、船摇周期影响的测量船异常数据检测难题，提出了基于小波分析的异常数据的动态识别与修正方法。基于小波分析的多尺度、局部分析等特性，采用具有紧支撑特性的小波去噪方法，根据数据自适应选取阈值，对噪声进行抑制凸显异常，在实现噪声抑制的同时，较好地保留了局部异常特征，显著提高了异常数据的检出率。测试结果表明，该方法切实可行，效果良好。
위해결온함유복잡변형、선요주기영향적측량선이상수거검측난제，제출료기우소파분석적이상수거적동태식별여수정방법。기우소파분석적다척도、국부분석등특성，채용구유긴지탱특성적소파거조방법，근거수거자괄응선취역치，대조성진행억제철현이상，재실현조성억제적동시，교호지보류료국부이상특정，현저제고료이상수거적검출솔。측시결과표명，해방법절실가행，효과량호。
It is difficult to detect the outlier data influenced by space tracking ship's complicated move-ment such as deformation and swaying. To effectively solve the problem, dynamic outlier data identification and elimination method based on wavelet analysis is put forward. According to the features of wavelet anal-ysis such as multi-scale representation and local analysis, compactly supported wavelet denoising method is applied, in which threshold value is selected based on data adaptability, by suppressing the noise anom-aly can be easily checked out, at the same time local anomaly characteristics are better maintained, there-fore anomaly detection rate is obviously improved. The test results show the proposed detection and elimi-nation algorithm is both practical and effective.