摘要
为研究建筑工程安全生产事故死亡人数的变化规律,采用时间序列分析方法,分析了建筑安全事故死亡人数时间序列上的趋势性规律,通过数据预处理和模型的识别与检验,最终建立了安全事故死亡人数预测模型。对全国2005—2014年建筑工程安全生产事故造成的死亡人数进行了分析和预测。结果表明:ARIMA模型各年预测值与实际值误差率为0.393,相比灰色模型和BP神经网络模型误差率最小。总体上说,ARIMA模型较适用于随机性较大的数据的趋势预测。
In order to reveal the change rules of death toll from construction engineering accidents,time series is used to analyze the trend rules of accidents and a ARIMA model is set up by data pretreatment and model identification and testing. Based on this model,the change rules of death toll from construction engineering accidents from2005 to 2014 have been analyzed and predicted. The results show that the difference between the predicted value by ARIMA model and the actual one is 0. 393 and it is minimal compared with the grey model and BP neural network model. In general,the ARIMA model is suitable for the trend forecast of the randomness data.
作者
王书明
郭起剑
WANG Shuming GUO Qijian(College of Civil Engineering, Jinling Institute of Technology Nanjing 211169)
出处
《工业安全与环保》
北大核心
2016年第10期60-63,共4页
Industrial Safety and Environmental Protection
基金
江苏省住建厅科研项目(2014ZD68)
金陵科技学院博士科研启动基金(jit-b-201230)
金陵科技学院校级科研基金项目(jit-2016-jlxm-16)
