基于多传感器数据融合的机房火灾检测算法

Fire detection algorithm in computer rooms based on multi-sensor data fusion

针对机房传统单传感器报警系统存在漏报率高、准确率低等问题,提出了一种基于多传感器数据融合的机房火灾检测算法。该算法首先采用寻优能力强的麻雀搜索算法(SSA)优化极限学习机(ELM)的预测精度和准确度。其次通过SSA-ELM算法模型对机房内多传感器采集的温度、烟雾浓度、CO浓度进行特征层数据融合,输出各火情概率。最后利用模糊推理将输出的各火情概率和火灾持续时间在决策层中进行特征融合,决策出火情警报等级。仿真实验表明:本文算法能根据多传感器数据融合的结果并结合不同危险等级区域给出合理的警报决策,极大提高了机房火灾检测的灵活性和准确性。

Aiming at the problems of high leakage rate and low accuracy of traditional single-sensor alarm system in computer rooms,we proposed a fire detection algorithm for computer room based on multi-sensor data fusion.Firstly,the algorithm optimizes the prediction accuracy and precision of the extreme learning machine(ELM)using the sparrow search algorithm(SSA)with high optimization seeking ability.Secondly,the feature layer data fusion of temperature,smoke concentration,and CO concentration collected by multiple sensors in an engine room was performed through the SSA-ELM algorithm model to output the probability of each fire condition.Finally,the features of output probability and duration of each fire were fused in the decision layer using fuzzy reasoning to decide the fire alarm level.Simulation experiments show that the algorithm can give a reasonable alarm decision based on the results of multi-sensor data fusion and combined with different hazardous areas,which greatly improves the flexibility and accuracy of fire detection in computer rooms.