权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Identification of Large Scaled System Based on Knowledge Engineering and Development of Abnormal Diagnosis System

基于知识工程的大规模系统识别及异常诊断系统开发

基本信息

批准号：
03555092
负责人：
KOBAYASI Yasuhide
金额：
$ 0.96万
依托单位：
Kure National College of Technology
依托单位国家：
日本
项目类别：
Grant-in-Aid for Developmental Scientific Research (B)
财政年份：
1991
资助国家：
日本
起止时间：
1991 至 1993
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-03555092/
关键词：
Abnormal Diagnosis System Identification Knowledge Engineering Modeling

项目摘要

In order to do an abnormal diagnosis, 1) abnormal detection, 2) abnornmal identification, 3) abnormal prediction, 4) presentation of a recovery law, of four processes are necessary. At first as establishing a various sensor against a system, in order to do a detection of abnormally of it, We measure a physical quantity, chemical qunatity in various ways and compare the state variables of a normal system. Then the model regarding a normal and an abnormal system is necessary. and it is the most important at an abnormal diagnosis to model the system exactly.We considered a following problem, at this research about each process 1)-4). 1) Detection law of parameter change for an already known of abnormal model structure. 2) Decision law of the most suitable input for a system identification. 3) Identificatin of an abnormal model against a on-linear system. 4) Identification law of the non-linear system to apply a knowledge base.Because a system that we are aimed at is a complicated non-linear system in large way, structure of an abnormal model is unknown. Therefore, we asume it, as structure of an abnormal model is unknown and consider a system identification. We propose a method that we detect early as much as possible to abnormally. We were able to make an abnormal diagnosis caability against a large scale system improve.We sum up a result of these and would like to construct the abnormal diagnosis ytem that is based on knowledge information.

要进行异常诊断，必须经过以下四个步骤：1)异常检测，2)异常识别，3)异常预测，4)异常恢复规律的提出。首先对一个系统建立各种传感器，为了对系统的异常进行检测，我们用各种方法测量一个物理量、化学量，比较一个正常系统的状态变量。那么，关于正常系统和异常系统的模型是必要的。在异常诊断中，系统的准确建模是至关重要的。在本研究中，我们考虑了以下问题：1)-4)。1)已知异常模型结构参数变化的检测规律。2)系统识别最合适输入的决策规律。3)针对非线性系统的异常模型识别。4)对非线性系统的辨识规律应用知识库。由于我们所研究的系统在很大程度上是一个复杂的非线性系统，异常模型的结构是未知的。因此，我们假设异常模型的结构是未知的，并考虑系统识别。我们提出了一种尽可能早发现异常的方法。对大规模系统的异常诊断能力进行了改进。总结了这些研究成果，提出了构建基于知识信息的异常诊断系统的设想。