论文部分内容阅读
单模型在处理不等长序列数据关联时不能兼顾计算精度、复杂度和抗扰性,为此提出了基于多模型(MM)的不等长序列数据关联算法。将基于滑动窗口和动态时间弯曲(DTW)的不等长序列相似度度量模型作为MM的输入模型,以2种模型计算得到的时似变化比作为模型判断指标进行模型转换,实现了2种模型的优势互补,并得到模型的应用条件,最后输出MM作用后的不等长序列相似度,以此作为关联指标进行关联判定。仿真实验验证了MM关联算法在处理不等长序列数据关联的有效性,并对序列长度和突变率变化对关联效果的影响进行了分析。
The single model can not balance the accuracy, complexity and robustness of the data when dealing with unequal length data association. Therefore, a multi-model (MM) -based data association algorithm with unequal length is proposed. Taking the unequal length sequence similarity measure model based on sliding window and dynamic time warp (DTW) as the input model of MM, the temporal change ratio calculated by the two models was used as the model judgment index to perform the model transformation, and two models , And get the application condition of the model. Finally, the similarity of unequal length sequence after the output of MM is output, which can be used as the correlative index to judge the association. The simulation experiments verify the effectiveness of the MM correlation algorithm in dealing with unequal length data association, and analyze the influence of the variation of sequence length and mutation rate on the correlation effect.