PID Neural Network Control Of Soft Switching DC-DC Converter

Intelligent control method for controlling the development of a higher requirements, the introduction of intelligent control neural network to build a new platform, which greatly promoted the research and development of intelligent control field. In modern control applications, due to the complex control system using traditional methods can not meet the control requirements, make up the complex neural network control system defects, so that the control problem is difficult to solve using traditional control methods have been improved. Soft switching DC-DC converter with parameter uncertainty and nonlinear characteristics, the application of traditional methods to control highlights the lack of paper soft switching DC-DC converter for the study, its PID Neural Network Control.This paper analyzes the soft switching DC-DC converter of the composition, structure and properties of typical Buck Type ZCS QRC circuit, for example, for soft switching DC-DC converter theoretical analysis. The high-frequency network construction paper averaging method of ZCS QRC equivalent circuit, as the basis for the establishment of a converter state equation and small signal mathematical model.Firstly, for the defects and limitations of conventional PID control method, designed neural network tuning PID controllers indirect, through simulations show a small improvement in dynamic performance. Secondly, the controller of BP network layers, neuron nodes and connection weights Initial selection irregular shortcomings direct neural network tuning PID control(PIDNN), the system dynamic performance is greatly improved, but there is still insufficient. Finally, in PIDNN control is proposed based on RBF network based identification PIDNN controlled by RBF network online identification mathematical model soft switching DC-DC converter to obtain its Jacobian information, then online tuning PID parameters PIDNN module. By soft switching DC-DC converter control simulation results show that RBF network controlled object line identification system may well overcome the impact of the uncertainty, effectively suppressing interference, enhance system stability.