A Travel-time Engine for Seismic Petroleum Applications using Non-deterministic Methods.

摘要

The travel-time calculation of seismic wave propagation in synthetic models is one of the fundamental algorithms of seismic data processing. Using a layer-based geological model, a Genetic Algorithm (GA) and an Evolving Monte Carlo method (EMCM) are proposed to compute travel-times. The performance analysis and comparison of the proposed algorithms are described in this thesis. Using the EMCM algorithm, Kirchhoff Ray-Tracing Modeling and Kirchhoff Pre-stack Depth Migration are implemented. Parallel compute clusters are adopted to speed up these computations. Two-dimensional and three-dimensional synthetic models are used to generate seismograms and image data. Furthermore, based on the EMCM algorithm, an application specific hardware engine is proposed to speedup the travel-time computation. In this thesis, we present a detailed circuit design and its chip implementation of this hardware engine. The technology of Field Programmable Gate Array (FPGA) and Application Specific Integrated Circuit (ASIC) is used to simulate and implement the hardware engine design. The logic and physical synthesis have been accomplished on two different vendor FPGA chips. Experimental results prove that the EMCM hardware engine is suitable for the calculation of travel-times.