摘要
构建了火灾报警系统设计性能评价指标体系,包括7个一级评价指标和21个二级评价指标。运用主成分分析法(PCA)对二级评价指标进行压缩提取,作为概率神经网络(PNN)的输入,建立系统设计性能评价模型。PNN在用于模式分类时,可以得到贝叶斯最优结果,提高了评价体系的精确度。通过具体案例,说明采用PCA和PNN结合对火灾报警系统进行设计性能评价,可以提高模型准确性。
The fire alarm system design performance assessment index system was built, including 7 first - grade indexes and 21 second-grade indexes. The second-grade indexes were extracted by principal components analysis （PCA） as the input of proba- bilistic neural network （PNN）, and the system model of design performance evaluation was established. Bayes best results can be obtained by PNN when it used for detection and pattern rec- ognition. By specific case, it was proved that using PCA and PNN to assess the design performance of fire alarm system can improve the model accuracy.
出处
《消防科学与技术》
CAS
北大核心
2016年第3期380-383,共4页
Fire Science and Technology
基金
国家自然科学基金项目(11000601-123-17
61473069)
关键词
火灾自动报警系统
性能评价
主成分分析法
概率神经网络
fire alarm system
performance assessment
princi-pal components analysis
probabilistic neural network
作者简介
曲娜（1979-），女，辽宁营口人，沈阳航空航天大学安全工程学院讲师，硕士，主要从事火灾报警技术与风险评估方面的研究，辽宁省沈阳市沈北新区道义南大街37号，110136。
