Dissertation > Excellent graduate degree dissertation topics show

A Study of GPU-based FDTD Method and Its Application in LED Performance Optimization

Author: HuZuo
Tutor: LiKang
School: Shandong University
Course: Radio Physics
Keywords: Finite difference time domain method GPGPU Compute Unified Device Architecture Light-emitting diodes
CLC: TN312.8
Type: Master's thesis
Year: 2011
Downloads: 77
Quote: 0
Read: Download Dissertation

Abstract


LED because of its small size, high efficiency, long life and other advantages gained widely used, the luminous efficiency has greatly improved as technology development. In the blue, red LED internal quantum efficiency can be as high as 80% under the status quo, the lower extraction efficiency limit the LED luminous efficiency bottleneck for further enhancement. Improve light extraction efficiency LED improve performance, the key to expanding the application fields, and become a research focus. LED light extraction efficiency is proposed to improve the extraction efficiency of the photonic crystal surface roughening, FIG type substrate, the flip chip method, but most studies have focused on the preparation and experiment on how these methods of extraction Analysis and Research of the efficiency impact is relatively small. Relative wavelength LED size larger, more complex model, the LED to the traditional electromagnetic simulation software to calculate and analyze the presence of defects in the lengthy research makes it difficult, requiring a high-speed platform for LED electromagnetic calculated auxiliary LED research and analysis. Unified Device Architecture (CUDA)-based general-purpose graphics processor (GPGPU) has a powerful data processing capability and high-speed computing performance, the simple parallel computing power of the graphics processor used in general-purpose computing, this article LED research required an ideal platform for high-speed solenoid. The goal of this study is CUDA-based high-speed platform to achieve parallel finite difference time domain (FDTD) algorithm used to calculate the electromagnetic field LED to provide high-speed computing platform for the analysis of the light extraction efficiency, and study the photonic crystal structure LED performance optimization, analysis of the impact of various parameters to light extraction efficiency. This paper briefly introduces the LED light-emitting principle and optoelectronic properties of the reasons for low light extraction efficiency, the present study concerns the optimization method and results are described. And then focuses on the FDTD algorithm used to simulate the open domain calculation of time-domain convolution absorbing boundary conditions (CPML) GPGPU and CUDA model. On the basis of the analysis of LED basic theory, the FDTD algorithm, and CUDA hardware and software architecture, the work done in this article, as well as the results obtained are as follows: 1. CUDA platform FDTD parallel algorithm and optimization studies. In a heterogeneous system, combined with the FDTD algorithm, analyzes planned for the GPU and CPU allocation of tasks. CUDA thread structure, establish a mapping between the computational tasks of the different dimensions of space and threading model, one-dimensional, two-dimensional, three-dimensional Cartesian coordinate system parallel FDTD algorithm. CUDA-based hardware and software features of the CUDA program to improve the degree of parallelism and utilization of computing resources, compared to more than 25 times to enhance the calculation speed and CPU computing, LED extraction efficiency of high-speed computing platforms. Analog open-domain electromagnetic computing the CPML conditions, simplify and implementation of the two-dimensional and three-dimensional CPML absorbing boundary conditions, time-harmonic point source and the incident plane wave can better absorb the outgoing line parallel FDTD algorithm wave, may aid the FDTD method in actual electromagnetic problems such as LED. (2) in the algorithm implemented on the platform of the LED simulation and analysis of photonic crystal structure extraction efficiency of GaN-based blue LED. Calculation of LED light extraction efficiency, this paper presents a calculation method based on the extraction efficiency of a Poynting theorem. Accumulate points in the FDTD method, the energy flux density in space and time domain, the total energy of the excitation source and monitoring surface energy of the light, and the calculation of the average time-harmonic electric dipole radiation power as an example, the analytical solutions and numerical solutions verify the correctness of the energy calculation method. Calculated with different parameters photonic crystal LED light extraction efficiency, the impact of the various parameters in the surface of the photonic crystal triangular lattice circular hole and embedded photonic crystal LED extraction efficiency were analyzed, including the filling medium, crystal the grid cycle filling ratio, the crystal thickness, and the position of the photonic crystal. The value of the parameter optimization, two photonic crystals with respect to the ordinary LED, optimization of the extraction efficiency increased 184% and 179%. Above the work and conclusions of LED electromagnetic analysis provides a high-speed computing platforms FDTD algorithm is applied to the electromagnetic computation of practical significance, LED modeling and simulation as well as the calculation of the light extraction efficiency, improve the performance of GaN-based blue LED optimization a reference and guidance value.

Related Dissertations

  1. Research on Video Compression Algorithm Based on Multi-core Computing Platform,TN919.81
  2. Study of Binocular Stereo Matching Algorithm and Algorithm Implementation Based on Multi-core.,TP391.41
  3. Implement and Research of High-Quality Global Illumination on GPU,TP391.41
  4. Study on Real-time Screen Space Fluid Rendering,TP391.41
  5. GPU-based video transcoding H.264 to AVS parallel design,TN919.81
  6. Design of H.264 Parallel Encoding Algorithms and Implementation on GPU,TP391.41
  7. Simulation Research on Multi-Core Stream Architecture and Power Consumption of GPGPU,TP332
  8. The Filter Properties of Photonic Crystal and Optimization Design of a Photonic Crystal Beam Splitter,O734
  9. GPU-based targets for background scattering of infrared radiation,TP391.41
  10. Time Domain Analysis of Basic Eletromagnetic Characteristics for Left-handed Materials,O441.4
  11. A Study on Diagonal Computing Model for GPGPU Platform,TP391.41
  12. Real-Time Image Mosaic Based on CUDA,TP391.41
  13. White Light Interferometry for Fast Areal Surface Measurement Based on GPGPU,O439
  14. The Research on Characteristics of Reconfigurable Optical Add/Drop Multiplexer Based on Photonic Crystals,TN929.1
  15. Research on Parallel Computation of H.264 Based on CUDA,TN919.81
  16. Based on Machine Vision Instrument Design and Implementation of Automatic Detection System,TP274.4
  17. Microelectronics & Solid-state Electronics,TN305.7
  18. Study of Growth of ZnO Nanorod Structures and the Optoelectronic Devices,TN304
  19. GPU-based parallel processing of data streams,TP332
  20. Web-based multi-view panoramic camera video technology,TP391.41

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Semiconductor technology > Semiconductor diode > Diodes: structure and performance > Light-emitting diodes
© 2012 www.DissertationTopic.Net  Mobile