Abstract:In this paper, we study the feasibility of support vector machine in small sample seismic signal recognition. The results show that with the increase of sample size, the recognition rate of this method increases first and then decreases. Using the seismic data of Shandong and Jiangsu in 2006-2017, only 30 training samples per class can achieve the correct recognition rate of about 85%. The improvement of recognition rate does not depend on the addition of a large number of samples. It is not only suitable for the study of seismic signal recognition in areas with few seismic data samples, but also provides a new idea for simplifying the sample library and reducing the operation cost.