并行计算在动态模式识别中的实现和应用

摘要

研究一种针对最近提出的动态环境下的机器学习理论--确定学习理论的算法实现,提出一种采用并行计算实现确定学习理论中的动态模式识别的方法.利用并行计算中的OpenMP多核编程环境,采用曙光16核服务器为硬件平台,实现对动态模式识别算法的快速性;同时,以压气机Mansoux模型为应用背景,把确定学习理论的动态模式识别方法应用到压气机旋转失速/喘振的快速检测中,利用多核并行计算实现了从包含多种旋转失速/喘振模式的模式库中快速识别当前模式的方法,为本方法提供了有效的验证.%The paper introduced a neural network-based learning theory, which was called the deterministic learning theory,and demonstrated the application of the theory in the dynamic pattern recognition. Based on the classical Mansoux compressor model, applied the deterministic learning theory to the model and achieved the parallel computing for the rapid recognition of compressor model under multi-dynamic patterns, by using the OpenMP multi-core programming environment and taking the Dawning server as the hardware platform. It provided an excellent application for the parallel computing in rapid pattern recognition.