More suitable for real-time control of dynamic recurrent neural network to replace the original proposal in recognition of its multi-layer feedforward network

Using a more suitable for real-time control of dynamic recurrent neural network to replace the original proposal in  Air circuit breaker of its multi-layer feedforward network. Simulation results show that the improved method than the original method and with the reader's neural network PID control method has better response performance. electric meter system, first studied the predictive performance index based on multi-function multi-variable system under the neural network PID control and simulation examples are given and conclusions; Then study multivariable PID neural  mcb circuit breaker  structure and calculation method: it is a sub-network of a plurality of parallel, when the control system has a controlled variable n, the n sub-network there. Each sub-network input layer of a given signal receiving system and the object output signal;