融合Informer和深度哈希的时序数据相似性搜索研究

Research on Similarity Search of Time Series Data Based on Informer and Deep Hash

针对工业大数据相似性搜索的效率和准确率不高的问题,提出了一种融合Informer和深度哈希算法的时序数据相似性搜索方法。首先,基于Informer搭建深度哈希数据特征提取模型;然后,通过贪婪哈希函数和层归一化构建深度哈希函数,通过对损失函数进行优化提高深度哈希算法的性能;最后,对M树(M-tree)进行改进,提高时序数据相似性搜索的效率。基于不同数据集的实验结果表明,该方法在保证较高准确性的前提下,可以有效提高时序数据相似性搜索的速度。

To solve the problem of low efficiency and accuracy for similarity search in industrial big data,a similarity search method for time series data based on Informer and deep hash algorithm is proposed.Firstly,a deep hash data feature extraction model based on Informer is built.Then,a deep hash function is established by hash greed function and layer normalization,and the performance of deep hash algorithm is improved by optimizing the loss function.Finally,M-tree is improved to improve the efficiency of similarity search of time series data.The experimental results based on different data sets show that the method can effectively improve the speed of similarity search of time series data on the premise of ensuring high accuracy.