为了有效地判别和定位大坝安全监控数据中的粗差,将数学形态滤波应用于粗差检测,针对大坝监控数据的变化特点研究了其算法和结构元素的选取。首先根据实测数据的特点选择合适的结构元素,然后根据所选的结构元素对数据进行数学形态滤波处理,最后根据实测值与滤波结果之间差值的大小应用四分点法拟定门限来判别粗差。对某混凝土坝裂缝开度数据加入粗差以模拟实际的情况,并采用该方法来检测加入的粗差,滤波得到的结果有效抵挡了粗差的影响,加入的粗差全部被检测出。
To detect gross errors in dam safety monitoring data, mathematical morphology filter is used to detect gross errors. The filtering algorithm and the selection of structuring element are studied considering the features of monitoring data. First, we select an appropriate structuring element according to the features of dam monitoring data; then, we perform mathematical morphology filtering to monitoring data according to the selected structuring element; finally, we compute the residuals between the measured value and the filtering result, determine the threshold for gross error detection using quartiles and detect gross errors using the threshold. The filtering result shows that the adverse influence of gross errors is effective, and all the added gross errors are detected.