A study on fire prediction method using air quality measurement sensors of smart indoor parking lot

We proposed a technology for predicting the degree of fire anomaly by machine learning using the air quality sensor of the indoor parking lot. Recently, artificial intelligence models for early detection of fires in parking spaces are being developed. In order to develop an artificial intelligence model, it is necessary to first collect air quality sensor data when a fire occurs. In this study, it was found that CO2, PM2.5, and VOC have a major influence on fire through fire tests using air quality sensors. In this paper, an autoencoder-based anomaly detection method was used for fire prediction, and when the collected data exceeds the threshold, the risk of fire is predicted to be high.