加权朴素贝叶斯算法在消防检测中的应用

Application of weighted naive Bayes algorithm in fire detection

针对当前建筑物消防检测受干扰影响较大,导致火灾预测精度较低的问题,给出一种用于消防检测的改进朴素贝叶斯算法。基于消防检测数据分析,通过信息增益计算加权值,将特征属性附加权重系数对朴素贝叶斯算法进行改进,并在此基础上通过Weka平台,设计并实现了改进朴素贝叶斯算法框架,将其用于消防检测。实验验证,比较朴素贝叶斯算法和其他分类预测方法,改进的朴素贝叶斯算法能有效解决每个特征属性对类别变量影响的关联度量化问题,降低了分类干扰,提高了消防隐患检测准确率。

Aiming at the problem that the fire detection of buildings is greatly affected by disturbance,which leads to the low accuracy of fire prediction,an improved naive Bayesian algorithm is proposed for fire detection. Based on the analysis of fire detection data,an improved naive Bayesian algorithm with additional weighting coefficients of feature attributes is proposed by calculating the weighting value through information gain. On this basis,an improved naive Bayesian algorithm framework is designed and implemented on the Weka platform,which is used in fire detection. Compared with naive Bayesian algorithm and other classification prediction methods,the improved algorithm can effectively solve the problem of associative measurement of the effect of each feature attribute on category variables,reducing the interference of classification,and improving accuracy of fire hazard detection.