建筑施工事故非线性灰色伯努利模型预测

Construction Accidents Forecast Based on Nonlinear Grey Bernoulli Model

为提高建筑施工事故灰色预测模型精度,在传统GM(1,1)模型基础上,建立非线性灰色伯努利模型(NGBM),并采用粒子群优化(PSO)算法对参数进行优选。以2001—2011年全国建筑事故死亡人数统计数据为基础,运用该模型对2012—2013年的相应人数进行预测,并与GM(1,1)模型和灰色Verhulst模型的结果相对比。结果表明,NGBM拟合精度最好,平均相对误差仅为2.65%,验证了模型的可行性和准确性。

In order to improve the predictive precision of grey predictive models for construction acci-dents, a NGBM based on traditional GM（ 1,1 ） model was built, and a PSO algorithm was applied to opti- mize its parameters. Based on the construction accident death toll statistics from 2001 to 2011, the death tolls for 2012 and 2013 were forecast using NGBM, and its results were compared with those by using GM （ 1,1 ） model and the grey Verhulst model. The results show that the NGBM is more accurate than another two models, with a mean absolute percentage error of only 2. 65%, validationg the feasibility and accuracy of the model.