Electrical and Electronic Engineering - Doctoral Theses
http://hdl.handle.net/10468/581
2018-05-21T14:16:00ZElectrical power optimisation of grid-connected wave energy converters using economic predictive control
http://hdl.handle.net/10468/5840
Electrical power optimisation of grid-connected wave energy converters using economic predictive control
O'Sullivan, Adrian C.M.
This thesis investigates the advanced control algorithms used for optimally extracting energy from a wave-to-wire wave energy converter system. The research focuses on the wave-to-wire system model as a whole, instead of its separate subsystems. This allows maximum exportation of average electrical power onto the grid from a wave energy array, with minimum mechanical and electrical constraint infringement and acceptable power quality. An economic model predictive control algorithm is first described for a wave-to-DClink system with a single wave energy converter connected to a simulated linear generator. This work investigates the importance of including the linear generator’s resistive losses in the cost function. Linear mechanical and non-linear electrical constraints are introduced into the model predictive control algorithm, where the effects on the average electrical power harvest are presented. A model predictive control algorithm with a field weakening enabled cost function is introduced, where the feasible region is extended for low DC-link voltages. By including a uni-directional power flow constraint into the algorithm, the power exported onto the DC-link bus is guaranteed to be positive. A detailed analysis of the effect of uncertainty on performance was carried out, where the controller’s internal model is mismatched from the simulation model. The results indicate that the high fidelity of the controller’s internal model is not required and that a sufficient amount of average electrical power is extractable. A non-linear model predictive control algorithm is described, where the non-linear viscosity forces are incorporated into the control algorithm - extracting maximum energy from a viscous system. It was shown that given the constraints on the system that the non-linear action of the control algorithm could be approximated, a linear model predictive control algorithm with an estimated viscous term. This produces a computationally inexpensive control algorithm, while maintaining good performance. A move-blocking was also introduced to further reduce the computation expense. Finally the thesis considers multiple point absorbers in an array and analyses the potential benefits of using either decentralised or centralised model predictive control algorithms. This demonstrated that the performance of a decentralised controller becomes comparable to the centralised controller when linear mechanical constraints are introduced into the viscous hydrodynamic array. However, when an upper power limit is introduced into the control algorithm the advantages of the centralised controller become apparent.
2018-01-01T00:00:00ZThin film technology for optoelectronics and their thermal management
http://hdl.handle.net/10468/4053
Thin film technology for optoelectronics and their thermal management
Quan, Zhiheng
Thin-film semiconductor optoelectronics are important for applications from optical communication, solid-state lighting, and wearable electronics to biomedical sensors. It is now possible to separate the micrometer-thick device layers from their native substrates and transfer them onto new platforms to optimize system performance and integration. The understanding of thermal management for such devices is very important in order to control the junction temperature effectively. Here, the laser-lift-off (LLO) technique was theoretically and experimentally studied. The temperature distribution at the III-nitride/sapphire interface induced by absorption of 248-nm KrF excimer energetic laser pulses was simulated to verify the experimental results. A 1.5-m-thick n-type Al0.6Ga0.4N membrane was separated from a c-plane sapphire substrate and then bonded to a Si substrate. The electrical behaviour of Ti/Al/Ti/Au contacts on the N-polar n-Al0.6Ga0.4N membrane was characterized. Furthermore, free-standing semipolar InGaN/GaN light-emitting diodes (LEDs) emitting at 445 nm were first realized by separation from patterned r-plane sapphire substrate using LLO. The LEDs showed a turn-on voltage of 3.6 V and output power of 0.87 mW at 20 mA. Electroluminescence measurements showed stronger emission intensity along the inclined c-direction. The -3 dB bandwidth of the LEDs is in excess of 150 MHz at 20 mA and a back-to-back data transmission rate at 300 Mbps is demonstrated. This indicates that the LEDs can be used for high bandwidth visible light communications. For thermal management of thin-film optoelectronics, a GaAs based laser diode (LD) was investigated. The 2-dimensional temperature distribution of the transfer-bonded LD was simulated; where the power dissipation, the thermal resistance of different cavity lengths and configurations were investigated. This can be utilized to optimize the device design and the choice of carrier substrate for efficient thermal management of thin-film optoelectronics.
2017-01-01T00:00:00ZHigh speed IC designs for low power short reach optical links
http://hdl.handle.net/10468/4056
High speed IC designs for low power short reach optical links
Zhou, Shiyu
In this thesis, I have briefly introduced the background of my PhD research, current state-of-the-art design, and my PhD research objectives. Then, I demonstrate how to optimize the performance of PAM-4 transmitters based on lumped Silicon Photonic Mach-Zehnder Modulators (MZMs) for short-reach optical links. Firstly, we analyze the trade-off that occurs between extinction ratio and modulation loss when driving an MZM with a voltage swing less than the MZM’s Vπ. This is important when driver circuits are realized in deep submicron CMOS process nodes. Next, a driving scheme based upon a switched capacitor approach is proposed to maximize the achievable bandwidth of the combined lumped MZM and CMOS driver chip. This scheme allows the use of lumped MZM for high speed optical links with reduced RF driver power consumption compared to the conventional approach of driving MZMs (with transmission line based electrodes) with a power amplifier. This is critical for upcoming short-reach link standards such as 400Gb/s 802.3 Ethernet. The driver chip was fabricated using a 65nm CMOS technology and flip-chipped on top of the Silicon Photonic chip (fabricated using IMEC’s ISIPP25G technology) that contains the MZM. Open eyes with 4dB extinction ratio for a 36Gb/s (18Gbaud) PAM- 4 signal are experimentally demonstrated. The electronic driver chip has a core area of only 0.11mm 2 and consumes 236mW from 1.2V and 2.4V supply voltages. This corresponds to an energy efficiency of 6.55pJ/bit including Gray encoder and retiming, or 5.37pJ/bit for the driver circuit only. In the future, system level analysis should be carried out to investigate the critical pattern issue of the PAM4 optical transmitter. The potential solutions toward 1pJ/bit are given (lumped EAM and micro-ring modulator). In addition, the advanced modulation formats (16 QAM, discrete multitone modulation, and FFE) are presented based on the switched capacitor approach.
2017-01-01T00:00:00ZDigital signal processing for fiber-optic communication systems
http://hdl.handle.net/10468/4700
Digital signal processing for fiber-optic communication systems
Ouyang, Xing
As the available bandwidth of optical fibers has been almost fully exploited, Digital Signal Processing (DSP) comes to rescue and is a critical technology underpinning the next generation advanced fiber-optic systems. Literally, it contributes two principal enforcements with respect to information communication. One is the implementation of spectrally-efficient modulation schemes, and the other is the guarantee of the recovery of information from the spectrally-efficient optical signals after channel transmission. The dissertation is dedicated to DSP techniques for the advanced fiber-optic systems. It consists of two main research topics. The first topic is about Fast-orthogonal frequency-division multiplexing (OFDM) — a variant OFDM scheme whose subcarrier spacing is half of that of conventional OFDM. The second one is about Fresnel transform with the derivation of an interesting discrete Fresnel transform (DFnT), and the proposal of orthogonal chirp-division multiplexing (OCDM), which is fundamentally underlain by the Fresnel transform. In the first part, equalization and signal recovery problems result from the halved subcarrier spacing in both double-sideband (DSB) and single-sideband (SSB) modulated Fast-OFDM systems are studied, respectively. By exploiting the relation between the multiplexing kernels of Fast-OFDM systems and Fourier transform, equalization algorithms are proposed for respective Fast-OFDM systems for information recovery. Detailed analysis is also provided. With the proposed algorithms, the DSB Fast-OFDM was experimentally implemented by intensity-modulation and direct detection in the conventional 1.55-μm and the emerging 2-μm fiber-optic systems, and the SSB Fast-OFDM was first implemented in coherent fiber-optic system with a spectral efficiency of 6 bit/s/Hz at 36 Gbps, for the first time. In the second part, Fresnel transform from optical Fresnel diffraction is studied. The discrete Fresnel transform (DFnT) is derived, as an interesting transformation that would be potentially useful for DSP. Its properties are proved. One of the attractive properties, the convolution-preservation property states that the DFnT of a circular convolution of two sequences is equal to the DFnT of either one convolving with the other. One application of DFnT is practically utilized in the proposal of OCDM. In the OCDM system, a large number of orthogonal chirped waveforms are multiplexed for high-speed communication, achieving the maximum spectral efficiency of chirp spread spectrum systems, in the same way as OFDM attains the maximum spectral efficiency of frequency-division multiplexing. Owing to the unique time-frequency properties of chirped waveforms, OCDM outperforms OFDM and single-carrier systems, and is more resilient against the noise effect, especially, when time-domain and frequency-domain distortions are severe. Experiments were carried out to validate the feasibility and advantages of the proposed OCDM systems.
2017-01-01T00:00:00Z