| With the development of marine seismic exploration to high density acquisition direction, seismic data processing of the growing volume of data, efficient seismic data processing has become the core problem of seismic data processing. However, the distributed and parallel computing has become the key technology to solve this problem. Owing to improving hardware level and establishing the large computer clusters, the distributed computing space to play more and more parallel. However, with the development of the distributed parallel computation, multi-computation unit of the complex control process becomes cumber some in the design and development process, which greatly reduces the efficiency of software development stage. The large number of inter process message passing also leads to the improvement of the complexity and error rate of communication.The thesis focus on the problems that is developing the distributed parallel computing process in the seismic data processing system(SPS).Through the study of SPS distributed parallel computing process, it establishes a set of suitable for SPS distributed parallel computing control process modeling language and control flow code automatic generation method, so as to reduce the distributed parallel the development of c-ode complexity, and improve the efficiency of code development. The main research contents of this thesis are as follows:1.Seismic Data Processing System Control Process ModelThe object of the SPS distributed parallel computing control is analyzed.And the SPS distributed parallel computing control flow model(SPCM) is designed by means of the object’s symbol definition and function description. The combination of elements in SPCM can be used to describe the complete SPS distributed and computational control processes.2.Seismic Data Processing System Control Process Modeling LanguageOn the basis of SPCM, the visual definition and semantic expansion of components are defined, and the modeling language of SPS distributed SPCML is designed. SPCML can be for SPS distributed parallel computing and control visual modeling and process design stage.According to SPCML, it designs the corresponding metadata markup language SPCMML to ensure information in the visual modeling of analyticity.3.Dynamic Code Generation Model for Seismic Data Processing SystemOn the basis of the SPCM, combined with the SPS code interface design requirements, code storage requirements and the corresponding to the SPCM element code to achieve demand, it designs the corresponding rule set, the formation of dynamic code generation model(DCGM). The code can be generated by the combination of the DCGM elements and the SPCM elements.4. Dynamic code generation conversion algorithmsBased on SPCMML, the analysis of the semantic tree of SPCML modeling logic relationship based on XML markup language is designed. Through analysis of the semantic parse tree, according to the elements to obtain the SPCM, it combines the DCGM rules. And it makes the conversion algorithm from SPCML to DCGM. Finally, it produces and designs the SPS control process to match the dynamic code.In summary, through the design of dynamic code generation framework and specific implementation process, it solves the problem in seismic data processing system of distributed parallel computing in the development process which software developers encountered. And for the automatic generation of dynamic code, it designs the specific application to verify, meanwhile, it proves this thesis is feasible and effective. |