The application of the improved BP algorithm in the project investment prediction

The traditional BP learning algorithm converges slowly, and obviously depends on the step size. The paper has proposed a step-size rule using to improve BP algorithm. In the network training, it can search a relatively reasonable step length in each iteration, so it can reduce the impact of the choice of step length on learning speed greatly. The simulation results of project investment prediction show that the new algorithm is better than the traditional in the iteration time and times.