城市高层建筑智能火灾多感监测系统研究

Research on fire intelligent multi-sensory monitoring system for urban high-rise buildings

针对城市高层建筑火灾的监测困难与预警准确度低的现状,以ZigBee-WiFi为基础通信网络,给出了多感监测系统网络结构与节点硬件设计。构建基于PSO-ELM的高层建筑智能火灾多感监测模型,完成了实验室条件下的PSO-ELM仿真验证,采用多传感器的100次实验数据样本的训练对该模型进行分析与测试验证。仿真结果表明,使用PSO-ELM优化算法时能够提高监测计算的速度和准确度,而且降低了训练样本数和隐含层节点数变化对训练结果的影响,通过实验仿真得到PSO-ELM的预测结果更接近实际值,而且最大相对误差只有0.6%,其预测效果优于SVR算法和BP神经网络算法。

In allusion to the situation that it is difficult to monitor the fire disaster of urban high-rise buildings and the ear-ly warning accuracy is low,the network structure and node hardware of the multi-sensory monitoring system are designed,which takes ZigBee-WiFi as its basic communication network.The fire intelligent multi-sensory monitoring model based on PSO-ELM was constructed for high-rise buildings,which was verified by PSO-ELM simulation under laboratory conditions.This model is analyzed and tested for verification by 100 times of experimental sample training with multiple sensors.The simulation results show that the PSO-ELM optimization algorithm can improve the speed and accuracy of monitoring calculation,and reduce the in-fluence of the number of training samples and the number of hidden layer nodes on the training results.The PSO-ELM prediction results obtain by experimental simulation is more close to actual value,and the maximum relative error is only 0.6%,whose pre-diction effect is better than SVR algorithm and BP neural network algorithm.