Search

As the increasing popularity of integrating hands-free telephony on mobile portable devices and the rapid development of voice over internet protocol, identification of acoustic systems has become desirable for compensating distortions introduced to speech signals during transmission, and hence enhancing the speech quality. The objective of this research is to develop system identification algorithms for speech enhancement applications including network echo cancellation and speech dereverberation. A supervised adaptive algorithm for sparse system identification is developed for network echo cancellation. Based on the framework of selective-tap updating scheme on the normalized least mean squares algorithm, the MMax and sparse partial update tap-selection strategies are exploited in the frequency domain to achieve fast convergence performance with low computational complexity. Through demonstrating how the sparseness of the network impulse response varies in the transformed domain, the multidelay filtering structure is incorporated to reduce the algorithmic delay. Blind identification of SIMO acoustic systems for speech dereverberation in the presence of common zeros is then investigated. First, the problem of common zeros is defined and extended to include the presence of near-common zeros. Two clustering algorithms are developed to quantify the number of these zeros so as to facilitate the study of their effect on blind system identification and speech dereverberation. To mitigate such effect, two algorithms are developed where the two-stage algorithm based on channel decomposition identifies common and non-common zeros sequentially; and the forced spectral diversity approach combines spectral shaping filters and channel undermodelling for deriving a modified system that leads to an improved dereverberation performance. Additionally, a solution to the scale factor ambiguity problem in subband-based blind system identification is developed, which motivates further research on subbandbased dereverberation techniques. Comprehensive simulations and discussions demonstrate the effectiveness of the aforementioned algorithms. A discussion on possible directions of prospective research on system identification techniques concludes this thesis.

The electric power system is undergoing a paradigm change on how electric energy is generated, transmitted, and delivered. Power electronics systems which can provide medium-voltage (MV) to high-voltage (HV) output (>13.8 kV ac, > 20 kV dc) with much faster dynamic response (> 10 kHz bandwidth) or high switching-frequency will enable new electronic energy network architectures, like MVDC power delivery, underground solid-state power substation (SSPS), and high-density power electronics building block (PEBB); help drive the levelized cost of electricity (LCOE) of renewable energy on par with conventional power generation; deliver precise and clean power to loads like high-speed electric motors; push the future power system toward 100% renewable energy and energy storage supplied. In the MV to HV area, the power conversion solution is dominated by silicon devices, like SCR, IGCT, and IGBT, which are slow in nature, posing significant switching losses and bulky auxiliary components like turn-on snubbers. Devices in series are required to reach higher voltage. High-frequency HV converter in two-level or three-level bridges running 20 kHz or higher in many emerging applications, like MVDC networks with high-frequency transformers and energy storage integration is hard to be built by silicon solutions. The emerging HV wide-bandgap (WBG) power semiconductors, e.g., 10 kV SiC MOSFETs offer higher blocking capability, faster and more efficient switching performances. This makes the high-frequency power conversion technology feasible for the MV area. To build a MV high-frequency power converter with high-power density, 10 kV SiC MOSFETs in series are required to reach >10 kV operation dc voltage as the single device rating is still limited by the semiconductor process and packaging capability. However, the knowledge of dynamic voltage sharing of high-speed HV SiC devices under high dv/dt rate and effective balancing methods are not fully explored. Both the voltage imbalance and the robust device voltage balancing control are not studied clearly in the existing literature. This dissertation evaluates the voltage imbalance of series-connected 10 kV SiC MOSFETs thoroughly. The parasitic capacitors connected with device terminals are found to be a unique factor for the voltage imbalance of series-connected SiC MOSFETs, which have a significant impact on the dv/dt of different devices based on the detailed analysis. The unbalanced dv/dt and the gate signal mismatch together result in the voltage imbalance of series-connected SiC MOSFETs and a set of new voltage balancing control methods are proposed. Passive capacitor compensation and closed-loop short pulse gate signal control are proposed to solve the voltage imbalance caused by the unbalanced dv/dt. Closed-loop gate delay time control is proposed to solve the voltage imbalance caused by the gate signal mismatch. Two gate driver prototypes are designed and verified for the proposed voltage balancing control methods. As the number of devices increases, the voltage balancing methods under the device-level will be complex and risky to coordinate. Therefore, the converter-level device voltage balancing methods are desired when over three devices are in stack. Therefore, this dissertation proposes to use the 3-level (3L) neutral-point-clamped (NPC) converter structure as a converter-level approach to simplify the voltage balancing control of series-connected SiC MOSFETs. A new modulation strategy is proposed to control the loss of clamping diodes, so compact MV SiC Schottky diodes can be selected to reduce the impact of extra components on the power density. Compared to the phase-leg with direct series-connected SiC MOSFETs, the phase-leg designed with the converter-level approach achieves similar power density, easier voltage balancing control, and better efficiency, which is attractive for both two and four devices in series connection. Finally, this dissertation studies the impact of series-connected 10 kV SiC MOSFETs on MV phase-leg volume reduction with the example of multi-level flying capacitor (FC) converters. The relation between the capacitances of FCs and the device voltage is studied and a new design procedure for FCs is developed to achieve minimum FC energy and regulate the maximum device voltage. With the design procedure, the total FC volumes of a 22 kV 5-level FC converter and a 22 kV 3-level FC converter with series-connected 10 kV SiC MOSFETs are calculated and compared. Series-connected 10 kV SiC MOSFETs are found to help significantly reduce the total FC volume (> 85 %). In summary, this dissertation demonstrates that the direct series connection of 10 kV SiC MOSFETs is a reliable solution for the MV converter design, and the converter-level approach is a better voltage balancing control method. This dissertation also presents a quantitative analysis of the volume reduction enabled by the series-connected 10 kV SiC MOSFETs in MV converter phase-leg design.