摘要
摆动变量是QSIM产生奇异行为分支的主要原因之一,辨识摆动变量并利用有效的方式消除变量的摆动是提高QSIM运算效率和预测准确度的一个关键技术,这个问题一直是定性仿真的一个难题。认真分析了Kuipers的相关理论后,发现其判定摆动变量的判据与变量的摆动没有直接的因果关系。因此一套摆动变量辨识和利用变量的高阶导数约束变量摆动的改进方法被提出,并以两级级联水箱和多级级水箱系统为例对该方法进行了说明。改进的方法有效地利用了系统常识和QDE中蕴含的定性信息,对系统常识与QDE的结合进行了有益的探索。
Chattering variable is one of important causes that QSIM creates intractable branch, and identifying chattering variables and eliminating the chattering with available method is a key technology improving the operation rate and predicting accuracy of QSIM,it has been the problem of qualitative simulation. We studied carefully 揌igher- Order Derivative Constraints?theory of Kuipers, and found that there is not direct causal relation between the criterion and chattering variables. So a series of improved methods to identify chattering variables and eliminate chattering of variables via the higher-order derivative of the variables were advanced,and the illumination for the methods is made by two-tank and N-tank cascade systems . Meanwhile we tried to use effectually system knowledge and qualitative information implicating in QDE to eliminate the chattering of variables.
出处
《系统仿真学报》
CAS
CSCD
2004年第4期829-832,836,共5页
Journal of System Simulation
关键词
定性仿真
摆动变量
定性推理
模型分解
qualitative simulation
chattering variable
qualitative reasoning
model decomposition