Abstract: In order to improve the judgment accuracy of fire detection, this study proposed an intelligent fire judgment model based on multivariate data fusion for the first time, integrating wind speed, temperature and fire scale, breaking through the traditional single parameter limitation. Firstly, the determination principle and response threshold of smoke sensing and temperature sensing flame detector are analyzed, and the application characteristics and limitations of various detectors are clarified. Subsequently, fire entity tests of wood stacks with various ignition methods and different ignition positions were carried out in a space with a net height of 3 m. The results show that the response time of the detector is positively correlated with the distance of the fire source, that is, the farther the distance, the longer the response time, and the ignition mode has little influence on the response time. The smoke detector can accurately determine the fire when the critical wind speed is 0.15 m/s, and the temperature detector can trigger the alarm when the ambient temperature reaches 67 ℃. Finally, a mathematical model of fire judgment is constructed based on AHP, and the test results are brought into the formula to obtain the comprehensive fire judgment threshold of the test scene， which is Y=0.781. When the actual value exceeds this threshold, it is judged as a fire. The formula combines the parameters of wind speed, temperature and fire size, and verifies that the error range is less than 15%. This study provides the test basis for the accuracy evaluation of the fire detector and has important significance for the comprehensive evaluation of the detector performance.

Key words: fire detector, fire identification, response threshold, fire test, mathematical model