Optimization of Dynamic Time Warping Algorithm for Abnormal Signal Detection

This paper proposes an algorithm based on a combination of DTW lower bound functions and Sakoe–Chiba constraints to improve the time efficiency of DTW distance measurement, which suffers from a high computational complexity and low efficiency while ensuring measurement accuracy. First, the Sakoe–Chiba-DTW algorithm is used to optimize the template and threshold in the original sequence. Then, different combinations of lower bound functions are introduced to filter out sequences that do not meet the similarity requirements compared to the optimized threshold, reducing the number of DTW calculations to improve efficiency. The proposed algorithm is evaluated on 5 sets of self-built data samples for the detection and removal of interference signals caused by redundant objects. The results show that the algorithm achieves the same level of accuracy as traditional DTW algorithm, but saves up to 73880 s in detection time, greatly improving efficiency and having significant implications for data mining tasks.