Please use this identifier to cite or link to this item:
|Title:||Parametric estimation of non-minimum phase switch mode power converters|
|Abstract:||Nowadays, switching mode power converters (SMPCs) are widely used in many applications. The advanced control technique for converters, such as adaptive control is also spread-used in many converter control scheme designs. System identification as a tool for estimating the converter operating conditions, and providing the information to the controller is a key technique for these applications; and parametric estimation, which is part of the system identification technique, is an advanced identification technique which can allow on-line system identification and adaptive control design. However, most of the research over the past decades has only covered parametric estimation of buck converters and there is barely anything about boost converters or other non-minimum phase converters. The reason behind this is that the parametric estimation results of non-minimum phase converters are not fitted to the calculated model weights, especially for the numerator weights of the model transfer function. Thus, the controller gains cannot be determined correctly by the wrong estimated model weights. It has been a big problem in the application of parametric estimation for decades. In this research, a modelling method which is based on trailing-edge PWM off-time sampling (TEOS) is introduced in order to address this problem. The objective of this research is to develop an approach to resolve the existing accuracy problems of non-minimum phase SMPC parametric estimation. The problem, which has existed for decades, is that commonly used state-space averaged model numerator weights are not fitted to the non-minimum phase SMPC parametric estimation results. There are several possible ways to address this problem, including modification of converter modelling, modification of parametric estimation mechanism, or with the help of compensators. In this research, the TEOS modelling method has been verified by both simulation and practical experiment to provide the best-fit model weights for the parametric estimation of buck converters and boost converters; and it has also been verified, by simulation, to be used for buck-boost converter parametric estimation, which has opened up great possibilities for its use on other non-minimum phase converters. The experimental results have shown that the proposed modelling approach has improved the accuracy of parametric estimation for boost converters by more than 20% compared with the commonly used state-space averaged modelling approach. IV In addition, the TEOS model will also present a thorough inspection of the relationships between system parameters (load resistance, capacitance and inductance) and the model transfer function parameters, which can then realise the sensor-less on-line system parameters estimation or monitoring. This function is also a novel approach to the area of system component monitoring. In this thesis, the reason behind the problem of non-minimum phase converter parametric estimation is analysed for the first time. The system parametric estimation of three converters (buck, boost and buck-boost) were tested with on-line simulation and off-line experimental tests for both the averaged model and the proposed model. Then, system parameters estimation was also tested for the buck converter and boost converter in the simulation and practical experiment. In addition, the platform setup, the interface build between the Matlab Simulink and the Code Composer Studio (CCS), the settings of the Digital Signal Processor (DSP) TMS320F28335, the parameters design of boost SMPC, and the design of the Printed Circuit Board (PCB) schematics and layout are also presented in this thesis. The outcome of the research should be able to further benefit many applications of advanced control systems, fault detection, and system component monitoring.|
|Appears in Collections:||School of Electrical and Electronic Engineering|
Files in This Item:
|Li R 2020.pdf||9.39 MB||Adobe PDF||View/Open|
|dspacelicence.pdf||43.82 kB||Adobe PDF||View/Open|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.