Font Size: a A A

Research On Variable Switching Frequency PWM Based On Real-time Prediction Model

Posted on:2021-06-14Degree:DoctorType:Dissertation
Country:ChinaCandidate:Q LiFull Text:PDF
GTID:1482306518483934Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
The power electronic converter using pulse width modulation(PWM)is the main device for ac/dc power conversion.Among them,switching frequency is a key factor that can be used to improve the power conversion performance.Model Predictive Variable-SwitchingFrequency PWM(model predictive VSFPWM)means that the switching frequency can be changed in real time based on the switching ripple prediction model,so as to control the switching ripple and improve other performance of the converter.Model predictive VSFPWM,as a new branch of spread spectrum modulation technology,shows many excellent performances,such as: the current/voltage ripple of the converter system can be precisely controlled,switching losses can be optimized,electromagnetic interference peak can be suppressed and so on.These advantages make the model predictive VSFPWM has good research value and good industrial application prospect.In the preliminary work at home and abroad,the basic concept of model predictive VSFPWM and its implementation method have been studied.However,from the breadth and depth,the model predictive VSFPWM still has great research value and space.On this background,this paper first reviews the current ripple prediction method on the ac side of the typical three-phase two-level converter,and reveals that the inductance parameters have important influence on the current ripple prediction.On the basis of ideal circuit and parameter model,considering the fact that the permeability of magnetic element will decrease with the rise of winding current,this paper studies the influence of inductance saturation on the calculation of current ripple on ac side,proposes the concept of virtual inductance,and deduces the general single-phase prediction model of current ripple considering inductance saturation.Then the traditional VSFPWM based on current ripple prediction is modified,which can effectively cope with the problem of inductance parameter variation in the actual circuit.The proposed modification scheme considering the variation of inductance parameters has been verified in detail on the experimental prototype.An important problem that restricts the power density of converter systems is the need for expensive and bulky capacitors on the dc side to absorb the current harmonics and limit the voltage ripple to the allowable range.In the process of capacitor design and selection,the dc side current rms value and voltage ripple peak-to-peak value are two important reference indexes.This paper presents a time domain prediction method for the dc side current,deduces the analytical calculation results of the dc side capacitor current rms value,and analyzes the influence of the ac side current ripple.Furthermore,this paper establishes a real-time prediction model of voltage ripple on the dc side,and proposes a VSFPWM with the peak-to-peak value of voltage ripple as the control target.The above analytical calculation method and VSFPWM scheme have been verified in detail on the experimental prototype.Compared with a single three-phase two-level converter,the paralleled converter and the multilevel converter have better ac/dc side power quality,but more attention is paid to the circulation current ripple and voltage ripple of the divider capacitor.The circulation current ripple determines the equalization effect of shunt and the saturation of coupling inductor,while the voltage ripple of divider capacitor determines the multilevel voltage effect.This paper then studies the extended application of model predictive VSFPWM in complex topology.In order to make full use of the maximum design magnetic density of the coupling inductor of paralleled inverters,this paper proposes a VSFPWM based on the prediction of circulation current ripple,which takes the peak-to-peak value of circulation current ripple as the control target.In order to make full use of the standard of voltage ripple peak-to-peak value of flying capacitor,this paper takes the peak-to-peak value of voltage ripple as the control target,and proposes a VSFPWM based on the prediction of voltage ripple of flying capacitor.Relevant work has proved that,based on the establishment of the new prediction model,the concept of model predictive VSFPWM can be effectively applied in all kinds of complex topologies to realize the improvement of switching losses and EMI.In the end,the key common problems of model predictive VSFPWM are studied.Firstly,this paper uses the method of double Fourier series to calculate and analyze the output voltage spectrum of the model predictive VSFPWM.The theory proves that: under the condition of high carrier ratio,the influence of model predictive VSFPWM on the baseband harmonic of output voltage can be ignored,and it has the effect of spread spectrum and harmonic suppression for the switching harmonic.Secondly,considering the delay effect of digital controller,this paper studies the influence of model predictive VSFPWM on current closedloop control of three-phase converter.Furthermore,in order to suppress the low frequency pulsation of dq current caused by periodic switching frequency variation,this paper proposes a real-time phase compensation method for output voltage.All the above researches have been verified by simulation and experiment,solidifying the theoretical foundation of model predictive VSFPWM,and optimizing its performance.In this paper,the basic problems and key technologies of model predictive VSFPWM in three-phase ac/dc conversion are systematically studied.Related work can effectively cope with the actual impact of circuit inductance parameter change,and extend the model predictive VSFPWM to all kinds of topologies under different focuses,analyzes the output voltage harmonic spectrum of model predictive VSFPWM and its influence on the closed loop control,lay a theoretical and practical foundation for industrial application of model predictive VSFPWM.
Keywords/Search Tags:PWM, Variable switching frequency, Switching ripple, Prediction model, EMI, Switching losses, Inductance saturation, Spectral quantization, Phase compensation
PDF Full Text Request
Related items