Font Size: a A A

A Study Of 10KW UPS Based On Neural Network

Posted on:2016-11-18Degree:MasterType:Thesis
Country:ChinaCandidate:G Q GuoFull Text:PDF
GTID:2272330479484126Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Uninterruptible Power System to the load to provide stable and uninterrupted AC, widely used in industrial, civil and military fields, with the load becomes more and more complex, performance index of waveform steady-state precision and dynamic response speed of the inverter output requirements are also getting higher and higher. According to the problem of the three-phase inverter output slow dynamic response speed, high THD when load mutation and with nonlinear dynamic with the ordinary PID control algorithm. This paper proposes a scheme that the neural network intelligent algorithm for tuning the outer ring (voltage loop) of PI parameter based on the double closed loop control algorithm.This paper first introduces the Boost converter topology structure and working principle; Three phase inverter topological structure、mathematical model、SVPWM modulation and the design of LC filter; The inductor design of the Boost converter and the Three phase inverter which includes the inductor core materials, core size selection and calculation of the optimal air gap.Secondly, design the controller of the Boost converter and UPS inverter three-phase inverter system. Build the simulation platform of MATLAB,verification of the inner ring and the outer ring PI parameters that the boost converter of double closed loop control method. Aiming at the problem of the three-phase inverter output slow dynamic response and THD high when the sudden load change and nonlinear load with the ordinary PID control algorithm, this paper proposes a scheme that using radial basis function neural network (RBF) algorithm for tuning the outer loop (voltage loop) PI parameters based on the double closed loop control algorithm, because the learning rate and momentum factor influence of neural network training time, the speed of convergence by network, in this paper use the genetic algorithm (GA) optimization the learning rate and momentum factor of the RBF to improve the convergence speed, shorten the time of the network tuning PI parameters, to improve the performance of three phase inverter, and use MATLAB simulation tools simulation verification. The simulation results show that, the control scheme can achieve high dynamic characteristics of three-phase inverter and get steady high quality waveform.Finally, build 10KWUPS uninterruptible power supply hardware platform. Including the production design of TMS320F2812 main control board, auxiliary power supply, IGBT driving and voltage and current sampling circuit; and select IGBT and current voltage sensor; the software design of the control system and the results of the analysis.
Keywords/Search Tags:Uninterruptible Power Supply, The Boost converter, Three-phase Inverter, Design of Inductance, Radial Basis Function Neural Network, Genetic Algorithm
PDF Full Text Request
Related items