摘要
某型火炮火力系统PHM(prognostic and health management,故障预测与健康管理)进行信号预处理时,计算量大、实时性要求高,现有PHM使用的嵌入式通用CPU平台计算方案无法满足其对运算性能的要求。为此该文研究了一种基于ZYNQ软硬件协同计算的PHM信号快速处理方法,该方法使用FPGA流水线实现卷积计算,使单个时钟周期能够完成16次MAC(multiply accumulate,乘累加)计算,从而提升运算效率。以邻域均值降噪算法为例,在180 MHz的工作频率下,该方法可使ZYNQ平均运算速度达到1.62 GMAC/s,相较目前常用的PC端通用CPU AMD 5800H处理器实现了约5倍加速。
When a certain type of artillery firepower system PHM(prognostic and health management)performs signal preprocessing,it has a large amount of calculation and high real-time requirements.The current embedded general CPU platform computing solution used by PHM cannot meet its computing performance requirements.To this end,a fast processing method of PHM signal based on ZYNQ software and hardware cooperative calculation is studied.This method uses FPGA pipeline to realize convolution calculation,so that a single clock cycle can complete 16 MAC(multiply accumulate calculations),thereby improving the calculation efficiency.Taking the neighborhood average noise reduction algorithm as an example,at a working frequency of 180 MHz,this method can make the ZYNQ average operation speed reach 1.62 GMAC/s,which is about 5 times faster than the current commonly used PC-side general-purpose CPU AMD 5800H processor.
作者
杨硕
狄长安
李大海
张永建
YANG Shuo;DI Chang-an;LI Da-hai;ZHANG Yong-jian(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)
出处
《自动化与仪表》
2021年第12期67-71,共5页
Automation & Instrumentation
关键词
现场可编程逻辑门阵列
硬件加速
信号处理
故障预测与健康管理
故障诊断
健康管理
field programmable gate array(FPGA)
hardware acceleration
signal processing
prognostic and health management(PHM)
fault diagnosis
health management
