In order to enhance the optical fiber sensor network′s ability to recognize similar disturbance signals and improve the accuracy of target classification,an improved particle swarm optimization support vector machine (PSO SVM) algorithm is proposed.Based on the analysis of the spectral characteristics of the echo signal,three judgment indexes are designed to describe the signal characteristics.The main wave signals energy,main wave pulse width and waveform change rate are used as the characteristic parameters of data preprocessing,which improve the traditional data classification model.Three typical interference forms in practical applications are simulated experimentally for impact testing on the ground in the testing area with mechanical,artificial and falling objects,and the response effects at different distances and different strengths are compared.The results show that there are 3 spectral characteristics corresponding to 6 different situations,and the similarity of the spectral shapes of the same mechanism of action is very high.The response value of the characteristic parameter decreases with the increase of the test distance,and increases with the increase of the impact strength.The same test data are analyzed for disturbance signals,and the average recognition probability of the traditional SVM algorithm is 69.3%,while the average recognition probability of the algorithm is 90.1%.It is showed that this algorithm has certain advantages in improving the ability of fiber sensor network disturbance signal classification.