民航事故征候预测模型对比研究

Comparative study on prediction models of civil aviation accident symptoms

为预测民航事故征候数,依据民航行业发展统计公报,选取2010-2019年民航运输总周转量、旅客运输量、运输机队数量、固定资产投资等16个影响因素指标,使用GM(1,1)模型进行拟合,利用灰色关联分析理论进行指标关联度计算,选取指标关联度大的指标构建GM(1,N)模型进行拟合研究,以及使用多元线性回归理论构建回归方程进行对比。模型拟合结果显示,多元回归模型拟合精度最高,平均相对误差为4.62%,GM(1,1)模型拟合的平均相对误差为7.48%,GM(1,2)模型、GM(1,3)模型、GM(1,4)模型拟合精度分别为15.52%、27.86%、15.90%。应用多元回归模型得到中国民航2020年和2021年事故征候数的预测区间分别为616~645和650~711,事故征候趋于增加。

In order to predict the number of civil aviation accident symptoms,according to the statistical bulletin of civil aviation industry development,16 influencing factors such as total turnover of civil aviation transport,passenger transport volume,number of transport fleets and fixed asset investment were selected from 2010 to 2019.The GM(1,1)model was used for fitting,the grey correlation analysis theory was used for calculation of index correlation degree,the GM(1,N)model was constructed by selecting indexes with high index correlation degree for fitting research,and the regression equation was constructed by using multiple linear regression theory for comparison.The results of model fitting show that the multiple regression model has the highest fitting accuracy,with an average relative error of 4.62%,GM(1,1)model of 7.48%,GM(1,2)model,GM(1,3)model and GM(1,4)model of 15.52%,27.86%and 15.90%respectively.The prediction interval of the number of accident symptoms of CAAC in 2020 and 2021 is 616~645 and 650~711 respectively by using the multiple regression model,and the accident symptoms tend to increase.