论文部分内容阅读
针对真空热试验过程中的数据自动化监视需求,利用数据挖掘手段开展数据异常监测方法研究,提出了一种改进DTW-形态距离相似性度量算法,通过调整形态符号的计算方法,避免了数据规范化带来的形态符号计算失真问题。对实际样本数据相似性聚类准确率进行统计分析,获得了相关参数的最佳取值范围,达到了较高的聚类精度。
Aiming at the requirement of data automatic monitoring during vacuum thermal test, the research on data anomaly monitoring method by means of data mining is proposed. An improved DTW-shape distance similarity measurement algorithm is proposed. By adjusting the morphological symbols, the data normalization band Form symbols come to calculate the distortion problem. The accuracy of similarity clustering of actual sample data is statistically analyzed, and the best range of the relevant parameters is obtained, which achieves a higher clustering accuracy.