Characteristic problems of data collection in the process of fracturing operations are due to high frequency of data collection and complex downhole operating environment. In order to compensate noise and fluctuations in the fracturing data, we propose to apply an online piecewise linear representation method for hydraulic fracturing time series to re-describe the original sequence. First, the unique engineering characteristics of start and stop sanding in fracturing scenarios are combined, integrating the fitting errors to segment the sliding window; second, a mechanism of online optimization of the fitting form is proposed for adjacent fitting line segments with similar trends. When optimization is performed, the degree of optimization is calculated using the trend factor and the vertical distance factor. Finally, the original series is replaced by the fitted line segment. Based on actual fracturing data, we experimentally compare a proposed method with a variety of benchmark methods. The results show that thу developed method can retain important information of the original sequence to the greatest extent while coarse-graining the original sequence. Applying the linear representation method of the fracturing data, the computing time of the upper-level task can be further reduced and the task performance improved.

压裂作业过程中数据采集的特点是数据采集频率高、井下作业环境复杂。为了补偿压裂数据中的噪声和波动，我们提出对水力压裂时间序列应用在线分段线性表示方法来重新描述原始序列。一是结合压裂场景下起停砂的独特工程特点，整合拟合误差对滑窗进行分段；其次，针对具有相似趋势的相邻拟合线段，提出了拟合形式的在线优化机制。执行优化时，使用趋势因子和垂直距离因子计算优化程度。最后，原始系列被拟合线段替换。基于实际压裂数据，我们将提出的方法与各种基准方法进行了实验比较。结果表明，所开发的方法可以最大程度地保留原始序列的重要信息，同时对原始序列进行粗粒度处理。应用压裂数据的线性表示方法，可以进一步减少上层任务的计算时间，提高任务性能。