Intelligent research and judgment model of fire threshold based on clustering analysis

With the continuous progress of modern society and the rapid improvement of scientific and technological innovation ability, people's living standards are improving day by day. However, people enjoy a convenient life at the same time, the fire hazard is greatly increased. Once a fire occurs, it will bring serious harm. According to the statistics of the World Fire Statistics Center (WFSC) of the United Nations, there are about 7 million fires in the world every year, with about 65000 to 75000 deaths. In order to avoid fire and reduce the loss caused by fire, a good intelligent fire alarm system and a timely alarm mechanism are very necessary. Therefore, the reliability and failure rate of fire detectors become the key parameters that need to be balanced, in this paper, the actual number of fires in the city from June 1 to June 18 was determined to be 418 through the screening of MATLBA software, and the reliability and failure rate of each early-warning component were quantitatively analyzed. After that, CRITIC and Delphi methods were used to combine weight assignment. Through Cronbach's Alpha test, TOPSIS algorithm was improved. The evaluation coefficient was defined by Soergel and Kulczynskl. The improved TOPSIS-SK model was used to evaluate various types of components. According to the evaluation scores, the government was helped to select reliable fire detector types as point smoke detector and manual alarm button, the biggest innovation of this paper is to improve the TOPSIS comprehensive evaluation, and introduce the Zeger abundance and Kakshinski distance to simplify the calculation process; Secondly, the established intelligent research and judgment model of alarm threshold has strong universality and is of great significance to reality.