CAJ | 학술논문

提出了一种基于粗糙集-支持向量机(Rough Set Support Vector Machine,RS-SVM)的火灾识别算法.首先利用粗糙集理论,将描述火灾特征的6个变量映射为粗糙集的知识系统,再去除冗余信息.对该系统进行属性约简,获取该知识系统的规则集;利用SVM泛化和非线性逼近能力,将以上规则集作为训练火灾识别SVM的样本集,最终得到分类准确、优化的火灾识别算法.实验仿真表明:该算法对火灾识别精度高、速度快、抗扰性好、非线性能力强,且适用范围广,对于火灾及时准确识别具有重要意义.
제출료일충기우조조집-지지향량궤(Rough Set Support Vector Machine,RS-SVM)적화재식별산법.수선이용조조집이론,장묘술화재특정적6개변량영사위조조집적지식계통,재거제용여신식.대해계통진행속성약간,획취해지식계통적규칙집;이용SVM범화화비선성핍근능력,장이상규칙집작위훈련화재식별SVM적양본집,최종득도분류준학、우화적화재식별산법.실험방진표명:해산법대화재식별정도고、속도쾌、항우성호、비선성능력강,차괄용범위엄,대우화재급시준학식별구유중요의의.
An arithmetic for fire identification is proposed based on composite of Rough Set Support Vector Machine(RS-SVM). Firstly,using Rough Set theory,the six variables of fire characteristics mapped to the RS knowledge system are made,the redun-dant information is eliminated,the properties of the system are reduced,then the regulations of this knowledge system are ac-quired.By the generalization and nonlinear approach ability of SVM,the model using the regulations of this knowledge system is trained,ultimately,the accuracy and optimized fire identification algorithms are obtained.The simulation result indicates that the method has better performance of fire identification accuracy,converge speed,nonlinear approaching and immunity.