Abstract: As a new type of braking, the frictional-magnetic compound brake could combine the advantages of both frictional braking and magnetic braking, which is considered as an excellent braking method. However, for the frictional-magnetic compound brake's braking performance and controlling method, few researches are studied in the past. Firstly, based on a frictional-magnetic compound brake tester, the braking performance under and without magnetic field are carried out respectively. The changing rules of braking performance are analysed by experiments. Secondly, on the basis of the changing rule of braking performance, a control target of the constant force moment brake of the braking process is proposed. The intelligent control method for braking performance of frictional-magnetic compound brake is designed and established by using BP neural network and fuzzy PID and other intelligent techniques. At last, conventional friction braking tests and the constant force moment brake tests are carried out respectively. The results show that the final prediction model of friction braking torque based on BP neural network can effectively predict the final value of friction braking torque according to working parameters:initial braking speed and braking pressure. The feedback control method based on fuzzy PID can effectively reduce the fluctuating amplitude of braking torque. The braking system is implemented to braking with an approximate constant brake torque. In the braking process of frictional-magnetic compound brake, the braking speed is greater, the braking time is shorter, and the braking performance is further improved.

Key words: braking performance, constant brake torque, frictional-magnetic compound, fuzzy PID, neural network