Abstract:  The fast dynamic time wrapping (fastDTW) algorithm  has the shortcomings of singularity in logging curves similarity measurement, and it's hard to find an appropriate search boundary. To overcome this issue, we combine the fastDTW algorithm with the SDTW (summation dynamic time warping) algorithm to obtain the fastSDTW algorithm,which solve the problem of outliers by reconstructing logging curves and integrating time series gradient information and numerical information. Next, an adaptive search boundary is proposed combined with  well-trajectory data. The fastSDTW algorithm is applied to measure the similarity of inclined and horizontal wells' logging curves based on the new boundary. The result shows that the fastSDTW algorithm based on adaptive search boundary has higher accuracy and time complexity of O(N), which ensures the operation speed of the algorithm. Furthermore, the algorithm is applied to the identification of water-flooded layers， and the  water-flooded layers are identified by comparing adjacent wells and  a good application effect is achieved.

Key words: dynamic time wrapping algorithm, logging curves similarity measurement, singularity problem, adaptive search boundary, time complexity