Safety assessment plays a vital role in the operation of oil and gas production systems, and data processing is important for the analysis of defects to make correct diagnosis. Therefore, the research to find a reliable data processing method is carried out. The magnetic eddy current signal detected of the defect is generally an abnormal data of “one peak and double valley” type. When the distance between the two defects is close, the leakage magnetic fields of the two defects interfere with each other. In order to facilitate the extraction of the characteristics of such anomalous data, it is necessary to separate the overlapping abnormal data. In this article, the above methods for identifying and processing anomaly detection data of complex defects are studied. The data fitting method is used to find the most suitable fitting function, and the accuracy of overlapping peak separation is optimized by the improved peak separation method based on GA algorithm. The results show that the Gaussian function is most suitable for fitting prediction of the “defective cluster” detection data after separation. The overlapped peak separation result optimized by GA algorithm has less error with the actual data. Therefore, the relevant features of the separated data can accurately reflect the defect related information and effectively improve the pipeline safety assessment.

中文翻译：

---

基于多阶微分法和遗传算法的缺陷簇磁涡流信号重叠峰分离算法

安全评估在油气生产系统的运行中起着至关重要的作用，数据处理对于缺陷分析以做出正确诊断至关重要。因此，开展了寻找可靠数据处理方法的研究。检测到的缺陷磁涡流信号一般为“一峰双谷”型异常数据。当两个缺陷之间的距离很近时，两个缺陷的漏磁场会相互干扰。为了便于提取此类异常数据的特征，需要对重叠的异常数据进行分离。本文对上述复杂缺陷异常检测数据的识别和处理方法进行了研究。数据拟合方法用于寻找最合适的拟合函数，基于GA算法的改进峰分离方法优化了重叠峰分离的精度。结果表明，高斯函数最适合对分离后的“缺陷簇”检测数据进行拟合预测。GA算法优化的重叠峰分离结果与实际数据误差较小。因此，分离数据的相关特征可以准确反映缺陷相关信息，有效提高管道安全评估。GA算法优化的重叠峰分离结果与实际数据误差较小。因此，分离数据的相关特征可以准确反映缺陷相关信息，有效提高管道安全评估。GA算法优化的重叠峰分离结果与实际数据误差较小。因此，分离数据的相关特征可以准确反映缺陷相关信息，有效提高管道安全评估。