一种提高处理器运算能力的控制方法研究  

Control Method to Improve the Computing Power of the Processor

作　　者：江帆[1] 张晨洁[2] JIANG Fan;ZHANG Chenjie(Suzhou Industrial Park Industrial Technology School,Suzhou 215123,China;School of Electronics and Information Engineering,Changchun University of Science and Technology,Changchun 130022,China)

机构地区：[1]苏州工业园区工业技术学校,江苏苏州215123 [2]长春理工大学电子信息工程学院,长春130022

出　　处：《兵器装备工程学报》2020年第10期239-243,254,共6页Journal of Ordnance Equipment Engineering

基　　金：第四期江苏省职业教育教学改革研究重点课题(ZZZ20);吉林省教育厅“十三五”科学技术项目(JJKH20170618KJ)。

摘　　要：针对未知情况下非线性系统处理资源具有时变特性,研究了一种Anytime控制算法。通常的算法在可用处理资源允许的情况下,计算未来尽可能多的时间步长上的暂定控制输入序列,但有时处理资源不足以计算控制输入。为解决该问题,采用Anytime算法计算出未来控制输入的序列,然后缓冲这些序列以备后续使用。假设处理器的可用性是相关的,以便控制输入的数量在任何时间步长下均可采用马尔可夫链描述。利用基于随机李雅普诺夫函数的方法,得到闭环随机稳定的充分条件。在所得到的充分条件下提出一种新的模型,用于分析当处理器可用性是马尔科夫的情况下所得到的闭环系统,并通过仿真验证了该模型的稳定性。Aiming at the time-varying characteristics of processing resources for non-linear systems under unknown conditions,an anytime control algorithm was proposed.At each time step,the available processing resources permit,the tentative sequence of control inputs was computed,but sometimes the processing resources were insufficient to compute the control inputs.In order to solve this problem,the algorithm calculated the sequence of future control input,and then buffered these sequences for subsequent use.It is assumed that processor availability is relevant so that the number of inputs can be controlled by a Markov chain at any time step.By using the method based on random Lyapunov function,sufficient conditions for closed-loop stochastic stability were obtained.A new model was proposed to analyze the closed loop system when the processor availability is Markov under sufficient conditions.The stability of the model was verified by simulation.

关 键 词：马尔科夫 输入序列 李雅普诺夫 ANYTIME算法 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]