Research on stability of reverse unloading diaphragm pressure reducing regulator using BP neural network model

Back propagation (BP) neural network model was used to study the dependency between the structural parameters and the stability of reverse unloading diaphragm pressure reducing regulator (PRR). The stability of PRR obtained by adjusting structural parameters, in particular, multi-structure parameters, proves that the stability of PRR is very sensitive to the diameter of damping orifice and the stiffness of diaphragm, and also relatively sensitive to the damping coefficient of springy elements material and the effective length of the low-pressure chamber. Thus, various measures for improving the dynamic stability were proposed, including increasing the diameter of damping orifice, the stiffness of diaphragm, the damping coefficient of springy elements material within a certain range (6.5 times of standard value), the effective length of the low-pressure chamber and reducing the mass of valve spool. The BP neural network model tested by the error analysis of the numerical experiments does not show any phenomenon of over-fitting and local optimum. The predictions of the model are reliable, which can be used to support the decision for the design of the PRR and system analysis. In addition, the model is applicable for different data sets, and can be used to study the dependency between the structural parameters and the performance of other components.