Application of modified and optimized unbiased GM(1, 1) model in gas accidents predictio

For problem that unbiased GM(1, 1) model may lead to large prediction error or even prediction failure due to individual outlier, modified and optimized unbiased GM(1, 1) model was established. The model excludes outlier and uses new data obtained by once interpolation method instead of the outlier, and finally fits the predictive value by residual modification. The model was used to predict death toll in coal mine gas accidents, the prediction result showed that the modified and optimized unbiased GM(1, 1) model greatly improves fit and prediction accuracy of death toll in gas accidents, the posterior difference accuracy level is good, the absolute value of relative error of the predictive value is 0.070 4, only 40.8% of the value before optimization.