基于人工神经网络的火灾预警系统

Fire Alarm System Based on Artificial Neural Network

针对传统烟雾报警器阈值报警,不能及时发现火灾的缺点,设计一套基于人工神经网络的火灾预警系统。该系统在STM32F767芯片上搭建长短时循环神经网络,使用激光粉尘传感器和DS18B20温度传感器采集烟雾、温度变化曲线,并输入到神经网络中进行分析,能够达到在火灾尚未严重时,就进行报警,克服阈值判断的局限性,具有灵活的分析能力。通过实验室历史数据的反复训练,系统可以准确进行预警,比传统阈值烟雾报警器更具灵活性,效率也更高。

Aiming at the disadvantages of traditional smoke alarm threshold alarm,the disadvantages of fire cannot be discovered in time,and a fire alarm system based on artificial neural network is designed.The system builds a long and short-cycle neural network on the STM32F767 chip.The laser dust sensor and the DS18B20 temperature sensor are used to collect smoke and temperature curves,and input into the neu⁃ral network for analysis.It can be used to alarm when the fire is not serious.The limitations of threshold judgment have flexible analysis ca⁃pabilities.Through repeated training of laboratory historical data,the system can accurately provide early warning,which is more flexible and efficient than traditional threshold smoke alarms.