基于AHP与熵值法构建的火灾预测组合灰色模型  

Construction of combined gray fire prediction model based on AHP and entropy method

作　　者：郑子温 那孜古力·斯拉木 王婧蓉 王旭东[1] ZHENG Ziwen;Nady Slam;WANG Jingrong;WANG Xudong(Key Laboratory of Ministry of Education for Linguistic and Cultural Computing,Northwest Minzu University,Lanzhou 730030,China)

机构地区：[1]西北民族大学语言与文化计算教育部重点实验室,甘肃兰州730030

出　　处：《现代电子技术》2024年第5期118-126,共9页Modern Electronics Technique

基　　金：西北民族大学人才引进科研项目(XBMUYJRC2020009);西北民族大学2021年中央高校基本科研业务费项目申请书(31920210133)。

摘　　要：火灾预测可以帮助消防部门更好地采取预防措施和制定灭火方案,减轻火灾损失。如何通过人工智能方法预测火灾数量、判断火灾发展趋势成为一项重要的研究课题。对城市消防火灾数量进行预测时,文中首先对原始数据序列进行加权滑动均值处理;其次建立了基于背景值优化的灰色模型和无偏优化灰色模型;而后引入了结合等维新息理论的马尔可夫模型,对经过改进的灰色模型进行预测值的残差修正;最后建立了基于层次分析法(AHP)与熵值法的主客观赋权组合模型。针对北京市2012—2019年火灾事故数据进行建模,并对后续两年的火灾发生数量进行数据预测与模型对比验证分析,根据预测结果判断未来火灾数据的变化趋势。实验结果显示,优化模型可以提高预测精度,其中结合AHP与熵值法的组合模型预测精度达到了相对残差最小为0.6105%,后验方差比为0.323%。实验结果证明,优化后的模型可以更好地应用于对火灾事故的预测。Fire prediction can help fire departments to take preventive measures and make fire suppression plans,so as to reduce fire losses.How to predict the number of fires and judge the development trend of fires with artificial intelligence methods have become an important research topic.When predicting the number of urban fire protection,the original data series is subjected to weighted moving average first,then the grey model based on background value optimization and the unbiased optimization grey model are established,and then the Markov model combined with equal⁃dimensional and new information theory is introduced to correct the residual of the predicted value of the improved grey model.Finally,the subjective and objective weighting combination model based on analytic hierarchy process(AHP)and entropy method is established.In this paper,the fire accident data of Beijing from 2012 to 2019 are modeled,and the data prediction and model comparison of the number of fires in the following two years are verified and analyzed,and the change trend of future fire data is judged according to the prediction results.The experimental results show that the optimized model can improve the prediction accuracy.The prediction accuracy of the combination model combining AHP and entropy method reaches the minimum relative residual error of 0.6105%,and its posterior variance ratio is 0.323%.The experimental results show that the optimized model can be applied to the prediction of fire accidents satisfactorily.

关 键 词：火灾事故预测 GM(1 1) 马尔可夫模型 等维新息理论 层次分析法 熵值法 组合模型预测 

分 类 号：TN911.1-34[电子电信—通信与信息系统]