Abstract: Aiming at the problem that the fatigue life of wind power tower is difficult to predict effectively, a fatigue life prediction method based on acoustic emission characteristic parameter fusion (degradation curve) and PSO-LSTM (particle swarm optimization-long-short term memory) life prediction model was proposed. Firstly, acoustic emission signals of the whole process of fatigue of raw materials (Q355E) of wind power tower were collected based on acoustic emission technology under laboratory conditions, and characteristic parameters were extracted from the time-domain and frequency-domain characteristics of the original signals. Then PCA(principle component analysis) method was used to fuse the characteristic parameters, and the first principal component was used as the characteristic curve after fusion. Finally, based on LSTM model, PSO algorithm was used to optimize model parameters, and PSO-LSTM model was established to predict life. The results showed that the prediction accuracy of the optimized model was higher than that of the single model, which had a certain industrial prospects.