数据流定性化方法研究

基金项目: 国家自然基金（编号：69974038）资助项目；国家高性能计算基金（00215）支持

详细信息

RESEARCH OF FORMING THE QUALITATIVE TREND FROM THE MEASUREMENT STREAM

Department of Computer Science and Technology University of Science and Technology of China Hefei 230027

摘要: 数据流定性化处理，是在考虑噪声影响的情况下，对数据流进行处理，最终将数据流化分为单调变化的区段．这是定性仿真应用中的一个关键问题．本文在分析基于线性最小方差拟合的数据流定性化方法的基础上，改进其定性核心函数．提出了基于偏度系数的数据流定性化方法．并通过具体数据分析，证明了该方法的有效性．
- 数据流定性化 /
- 定性核心函数 /
- 偏度系数
Abstract: The qualitative trend is generated through a process which breaks the measurement stream into monotonic segments. It is important for the applications of qualitative simulation. After analyzed the method which using linear least-squared fit, we modified the qualitative kernel function and presented the method which using deflecting modulus. In the end of the article, we demonstrated the effectiveness of the method.
- qualitative trend /
- qualitative kernel function /
- deflecting modulus

[1]	Kuipers B.J.(1986).Qualitative simulation.Artif.Intell.,29,289～338
[2]	白方周,张雷.定性仿真导论,中国科学技术大学出版社,1998,218～244
[3]	Herbert Kay and Lyle Ungar.Estimating Monotonic Funcions and Their Bounds using MSQUID.Technical Report TR AI99-280,University of Texas at Austin,1999
[4]	Herbert Kay and Lyle H.Ungar.Deriving Monotonic Function Envelopes from Observations.In Working Papers from the Seventh International Workshop on Qualitative Reasoning about Physical System (QR-93),pages 117-123,Oreas Island,Washington,1993
[5]	H.Kay,B.Rinner and B.Kuipers.Semi-Quantitative System Identification.Technical Report TR AI99-279,University of Texas at Austin,1999
[6]	B Rinner,B Kuipers.Monitoring Piecewise Continuous Behaviors by Refining Semi-Quantitative Trackers,16th IJCAI-99