摘要
在定义矩阵列循环平移算子的基础上建立了艇体温度动态测试的数学模型 ,从确保温度测试数据有效性出发 ,提出了基于时间序列预测的容错测试方法。在“预置信息”、“正常误差信息”、“故障信息”三个概念的基础上提出了时间序列信息分离的新思想 ,使分离后的“正常误差信息”趋于平稳 ,再用神经网络在线预测并诊断出“故障信息” ,实现了容错测试和故障诊断。文章提出的基于时间序列预测的算法 ,也适用于其他具有时变“预置信息”
In this paper math model on dynamic measurement of hull temperature is founded according to defining column circle shift operator in matrix In order to insuring validity of the temperature testing data,a new tolerating measure method based on time serial forecast is put forward Base on three concepts of 'known message'?'tolerable error message'?'fault message',the novel idea of time serial message separating is bring up.Separated'tolerable error message'is tend to calm 'fault message'is diagnosed using neural network prediction on line This method is also helpful to the disposing of other data with'known message
出处
《电工技术学报》
EI
CSCD
北大核心
2002年第5期96-100,共5页
Transactions of China Electrotechnical Society
