论文部分内容阅读
分布式并行计算机在进行一定程度的光谱学信号处理领域内容的研究基础上,根据实际样本在光谱曲线内部的数据测试效应,进行计算机不同配置平台系统的分布式运行计算处理,使得内部文本文件中涉及的必要原始数据实现分布式并行方法的读取,结合内部的遗传算法进行一定特征波数的提取,以满足神经网络的具体输入格式要求,保证后续网路结构的补充。这类并行计算运行结果的处理能够满足单机顺序计算的具体水准,同时实际的工作计算效率能够超过传统设计值的3成,有助于光谱学信号研究工作进行深度复杂科学效应的改革,促进内部计算效率工艺的不断改进。
Based on the research on the content of a certain degree of spectroscopy signal processing in distributed parallel computers, according to the data test effect of the actual samples in the spectral curve, the distributed operation calculation processing of different configuration platform systems in the computer makes the internal text files The necessary raw data involved to read the distributed parallel method, combined with the internal genetic algorithm for a certain characteristic wave number extraction to meet the specific input format of neural network requirements to ensure the follow-up network structure to add. The processing of such parallel computing results can meet the specific level of single-machine sequential calculation. At the same time, the actual work calculation efficiency can exceed 30% of the traditional design value, which is conducive to the reform of deep complex scientific effects in spectroscopy signal research, Calculating the efficiency of the process of continuous improvement.