Lightning activity is closely related to People's Daily life and property safety. Therefore, accurate lightning prediction can effectively prevent and reduce disasters and provide strong protection for people's lives and property. Therefore, this paper proposes lightning activity prediction research based on improved fuzzy c-means clustering and t-s fuzzy neural network model (IFCM-T-S). The traditional fuzzy clustering method (FCM) is analyzed, and the subtraction clustering algorithm is used to optimize the initial clustering center of the FCM algorithm, which is called the improved fuzzy c-means clustering algorithm (IFCM). T-S fuzzy neural network is improved by IFCM algorithm, which is called ifcm-t-s model. On the basis of lightning activity data, IFCM-T-S is used to establish the lightning activity prediction model. Simulation comparison experiment shows that IFCM-T-S algorithm is 1% lower than the traditional BP neural network and fuzzy neural network MAPE, and IFCM-T-S has the fastest convergence speed and the highest prediction accuracy, which verifies the accuracy and rapid-fire of the method proposed in this paper in lightning prediction.