Abstract: BP Neural Network is an effective method for prediction so far，but there are some self-determinations in its practical applications，so a new BP Neural Network based on improved Particle Swarm Optimization（PSO） is proposed.The capacity of solving nonlinear problems of this algorithm is enhanced effectively through adjusting inertia factor ω dynamically.At the same time，the convergence speed of this algorithm and the capacity of searching global extremum is increased.The improved BP Network model is built up，and the rainfall in a certain city is analyzed by this model and gradually regression method.Evidence shows this model has high accuracy and stability.