Decoupling control of forging machine based on PID neural network

A PID neural network is proposed to solve the decoupling problem of multi-cylinder forging machine. Based on analyzing forging machine´s model the PID neural network is determined as 10-15-5 structure. 10 input- neurons connect the given value and the measure value of left front cylinder, left posterior cylinder, right front cylinder, right posterior cylinder and main cylinder.15 neurons of hidden layer respond 5 groups of P-neuron, I-neuron and D-neuron. And 5 output-neurons connect 5 circuit controllers. The learning process uses BP algorithm. The simulation shows this approach is effective.