核聚变实验中相似波形智能检索方法研究综述

摘要

There has been drastic data growth in many fusion plasma experiments.Some intelligent data retrieval methods to find similar waveforms have become indispensable for accelerating data recognition and analysis.Two very general approaches to search for similar waveforms are considered:searching for entire waveforms and searching subsequences within waveforms.LAB-COM group of Japan and JET-EFDA group of the EU have done a lot of research in this area, and got many achievements.This paper reviews the fundamental idea and key technologies of data retrieval methods for similar waveforms, compares and shows their advantages as well as limitations.Finally, some future research trends are given.%随着核聚变实验数据量的与日俱增，需要针对相似波形开发智能检索方法来加速数据识别和分析。波形的相似性查询包括全序列相似查询和子序列相似查询2个方面。日本的LABCOM小组和欧盟的JET-EFDA小组在这方面做了大量的研究工作，并取得一系列成果。通过详细评述波形相似性查询方法的基本思想和关键技术，对这些方法进行分析和比较，总结了这些方法的优点和不足，并对进一步的研究方向作出了展望。