为实现大坝长久稳定的安全服役, 避免监测资料中的粗差对大坝安全监测结果产生影响, 需要对监测数据中的粗差进行剔除。由于目前的粗差识别方法依旧会造成粗差漏判、误判情况的发生, 通过模仿人工识别数据粗差的过程, 运用程序设计语言, 提出一种基于图像处理技术的自动化粗差识别方法。首先对依据监测数据绘制出的散点图进行高斯模糊和二值化处理, 再提取主要趋势线, 最后识别出监测数据中的粗差点并进行剔除。选取某实际工程大坝监测资料, 运用该方法对其进行粗差识别, 并与传统 3R 识别准则的粗差识别效果进行对比。算例结果表明: 该方法对数据粗差的识别效果更加显著, 避免了粗差漏判情况的发生, 对粗差的剔除更彻底; 利用该方法识别后得到的统计模型复相关系数为 0.999, 标准差为 0.192, 模型精度更高, 也更符合工程实际情况。因此, 该方法具有一定的工程应用前景和实用价值。

In order to realize the longterm and stable service of the dam and avoid the influence of gross errors in the monitoring data on the dam safety monitoring results, it is necessary to eliminate the gross errors in the monitoring data. Because the current gross error recognition method can still cause the gross error to be missed or misjudged, an automatic gross error recognition method based on image processing technology by imitating the process of manual data gross error recognition and using a programming language. T he scatter map drawn according to the monitoring data is processed by Gaussian blur and binarization, the main trend line is extracted, and the gross error in the monitoring data is identified and eliminated. The monitoring data of a real dam are selected, the gross error is identified, and the results are compared with those of the traditional 3 identification criteria. The results show that: the recognition effect of the method is more significant, the applied met hod avoids the false negatives of gross errors and eliminates gross errors more tho roughly, in addition, the complex correlation coefficient of the statistical model obtained by the method is 0. 999, while the standard dev iation is 0. 192, which shows that the accuracy of the model is higher and the model is more in line with the actual situation of the project. Therefore, the method has a certain engineering application prospect and practical value.

国家自然科学基金( 51739003)