PID neural network control of hydraulic roll gap control system

Based on BP neural network to control the complex hydraulic gap control (HGC) system and point out the boundness of selection uncertainty for the BP neural network layers and neurons and the randomness of connection weights between layers. In this paper, an improved PID neural network (PIDNN) is proposed to make trapezoidal integral transform for hidden integral neuron nodes and to make incomplete differential transformation for hidden differential neuron nodes. The output function of each network node is hyperbolic tangent function to replace proportion threshold function. To control the hydraulic gap system by improved PIDNN, the simulation results show that the improved control has better efficiency and tracking characteristics.