基于多元时间序列融合的飞行动作识别方法

发布时间：2018-05-21 03:28

本文选题：动作识别 + 动态时间规整　；参考：《空军工程大学学报(自然科学版)》2017年04期

【摘要】：针对飞行动作数据随机性强与长度不一致的问题,提出通过减小动态时间规整(DTW)算法的搜索空间,并定义不同特征参数贡献度的概念,实现对飞行数据的多元时间序列融合,从而完成对战术机动动作的识别。通过引入预分类和细分类结合的方式,对动作数据进行预处理,然后根据改进的动态时间规整(WDTW)算法对待测数据进行识别。仿真实验表明,相比传统DTW算法,WDTW算法通过降低算法复杂度,识别计算时间变化明显;对核密度与精准度系数的分析表明识别准确率亦有所提高。实验结果验证了所提方法的准确性。
[Abstract]:Aiming at the problem of strong randomness and inconsistent length of flight action data, this paper proposes a method to fuse flight data in multivariate time series by reducing the search space of dynamic time warping (DTW) algorithm and defining the concept of contribution degree of different characteristic parameters. In order to complete the tactical maneuvering action recognition. By introducing the combination of pre-classification and fine-classification, the action data is preprocessed, and then the measurement data is recognized according to the improved dynamic time warping (WDTW) algorithm. The simulation results show that compared with the traditional DTW algorithm, the complexity of the algorithm is reduced and the computing time is changed obviously, and the analysis of the kernel density and precision coefficient shows that the recognition accuracy is also improved. The experimental results verify the accuracy of the proposed method.
【作者单位】：空军工程大学航空航天工程学院;
【基金】：航空科学基金(20155196022)
【分类号】：E926.3;O211.61

【相似文献】