Hydraulic AGC Roll Gap Control Based On PID Neural Network

Neural network is an active marginal interdisciplinary, a nonlinear dynamic system and an adaptive organizational system which can be used to describe the cognition, decision-making and control of intelligent behavior. Neural network is one of the most important methods in intelligent field. Some experts began to study and discuss the neural network control theory because of the limitations of traditional control method in modern control field. The development of neural networks makes the control theory has a tremendous development and solves many problems which can not be solved by traditional control method. In this paper, neural network will be studied based on hydraulic AGC roll gap system with uncertainty parameters.In this paper, the composition, construction and properties of hydraulic AGC system are analyzed, the main factors which impact the rolling mill gap are discussed, the spring equation is analyzed, and the rolling mill gap's effect to strip thickness are described. The mathematical model of hydraulic AGC roll gap system is established in the means of mechanism modeling.Firstly,hydraulic AGC system is controlled with conventional PID control method and PID control method based on BP neural network respectively. The simulation shows the limitations of the two control methods.Secondly , the parameters adjusting of conventional PID control are difficult, the selection for the BPNN layers and neurons of each layer has no rules, the connection weights between layers are random values. In order to solve the above defects, PID neural network (PIDNN) is used in the text. It improves the system dynamic and static performance, however, this method has still some shortcomings.Finally, an improved PID neural network control method (IPIDNN) is proposed, trapezoidal integral transformation is used in PIDNN hidden layer integral neuron, not fully differential transformation is used in PIDNN hidden layer differential neuron, the ratio threshold function is replaced by hyperbolic tangent function in each PIDNN neuron. Hydraulic AGC roll gap system is controlled by IPIDNN, the simulation results show that the IPIDNN controller has the advantages of higher anti-interference ability and adaptability to parameters' changing than conventional PID controller and BP-PID, and better dynamic property and stability.