Improving the Architecture-Based Neural Network Model by Using Hierarchical Control Theory

The modeling method using architecture-based neural network was analyzed. The architecturebased neural network model was improved by using hierarchical control theory. A coordinator is added into the original model, each sub-system's output is modified before the next time's computing. The architecture-based neural network model of 52SFZ-140-207B type hydraulic bumper was established. The test result shows that the improved model can reduce the error of sub-systems and their interaction and the accuracy of the system model is raised.