目前生产的火灾探测器大多针对不同火灾类型,可靠性和通用性较差。将信息融合技术应用在火灾监控系统中,利用ART—2(自适应共振理论模型)和三层前馈网络对多传感器信息进行两级融合,综合多种火灾参数来进行监测、报警,提高了系统的可靠性和通用性,减少了误报警的发生。着重介绍了神经网络的选择、设计及训练过程。实验证明系统准确且可靠。 Currently produced fire detectors are mostly for different types of fire, reliability and versatility is poor. The application of information fusion technology in fire monitoring system, the use of ART-2 (adaptive resonance theory model) and three-layer feedforward network for multi-sensor information fusion at two levels, integrated a variety of fire parameters to monitor, alarm, improved System reliability and versatility, reducing the incidence of false alarms. The selection, design and training process of neural network are emphatically introduced. Experiments show that the system is accurate and reliable.