文献类型：期刊文献

英文题名：Intelligent fire monitor system based on information fusion technology

作者：Cao, Liting[1]

第一作者：曹丽婷

通讯作者：Cao, L.

机构：[1] College of Automation, Beijing Union University, Beijing 100101, China

第一机构：北京联合大学城市轨道交通与物流学院

通讯机构：[1]College of Automation, Beijing Union University, Beijing 100101, China|[1141751]北京联合大学城市轨道交通与物流学院;[11417]北京联合大学;

年份：2008

卷号：4

期号：5

起止页码：2111-2116

外文期刊名：Journal of Computational Information Systems

收录：EI(收录号：20092212098587);Scopus(收录号：2-s2.0-65649134302)

语种：英文

外文关键词：Alarm systems - Backpropagation - Cladding (coating) - Decision making - Fire alarm systems - Fires - Information fusion - Intelligent buildings - Monitoring - Neural networks

摘要：Fire detection is a non-structured problem, and it is very hard to describe by exact mathematic model. The difficulty of detection increased because the actual circumstance is very complex. This paper presents a fire monitor system based on multi-sensor information fusion. There are three fusion layers in this system; they are signal fusion, character fusion and decision-making fusion. Signal fusion is primary information process. The differing character information of a certain property will be distilled in signal fusion layer by optimizing and choosing state signals and process parameters. Character fusion is intermediate information process. The most important parameter will be distilled in character fusion layer by fusing the character information. Decision-making fusion is altitude information process. Right decision will be obtained in decision-making fusion layer by fusing multi-kind information source. The three layer forward structured BP (Back-Propagation) network is adopted in information fusion. Levenberg-Marquardt arithmetic of neural network is used to train the BP network. Temperature sensor, smog sensor and CO sensor was used to detect the information of fire. The simulation results show that the training results very approach the anticipant probabilities. The precise degree of identification of LM arithmetic is very high. The mistake alarm is overcome and the belief degree of fire alarm is advanced. The system has strong accommodating ability and anti-jamming ability. It can be put in practice in intelligent building to realize fire alarm and fire protection automatically. ? 2008 Binary Information Press.