Traditional PID control has been widely used in real applications. But for traditional PID control, mathematical model must exist when adjusting PID parameters, and the parameters are constant after adjusted. However, in real systems, when the system state and parameters change and become uncertain, system performance cannot keep in the best state. For the purpose of improving system performance of PID control and making use of the existing study fruit of PID control, BP neural network was used for adjusting PID parameters. After simulating, the results show that the algorithm based on BP neural network adjusting has better control characteristics than those of traditional PID and BP neural network.