电梯设备安全事故率的广义代数差分预测模型  

Generalized algebraic difference approach prediction model for elevator equipment safety accidents rate

作　　者：晋良海[1,2,3] 陈颖 杨应柳 邵波 JIN Lianghai;CHEN Ying;YANG Yingliu;SHAO Bo(College of Hydraulic&Environmental Engineering,China Three Gorges University,Yichang Hubei 443002,China;Safety Production Standardization Review Center of China Three Gorges University,Yichang Hubei 443002,China;Hubei Anhuan Technology Co.,Ltd.,Yichang Hubei 443002,China)

机构地区：[1]三峡大学水利与环境学院,湖北宜昌443002 [2]三峡大学安全生产标准化评审中心,湖北宜昌443002 [3]湖北安环科技有限公司,湖北宜昌443002

出　　处：《中国安全科学学报》2023年第1期64-69,共6页China Safety Science Journal

基　　金：国家自然科学基金资助(52179136,72204141)。

摘　　要：为预测电梯设备安全事故发生率,采用广义代数差分法(GADA)建立动态差分预测模型。首先,采用GADA设置多自由参数,构建系列动态差分方程;其次,采用代数差分法(ADA)设置单自由变量,构建对照方程;然后,耦合分析统计指标和图形形态,厘定电梯设备安全事故率的最优GADA预测模型;最后,利用近16年我国电梯设备安全事故数据验证分析最优预测理论模型的有效性。结果表明:我国电梯设备的安全事故发生率呈整体降低的趋势,但万台事故率的下降趋势有所减缓;采用GADA由指数方程推导的模型偏差为0.0022,均方根误差(RMSE)为0.045352,拟合度为0.8868,是最佳预测模型,预测值相对误差为5.9%。该模型可根据电梯设备登记数、万台事故数等实时数据进行更新,预测我国电梯设备的安全事故发生率。To predict the occurrence rate of elevator equipment accidents,the generalized algebraic difference approach was used to establish a dynamic difference prediction model.Firstly,the generalized algebraic difference approach was used to set multi-free-parameters and constructed a series of dynamic difference equations.Secondly,the algebraic difference approach was used to set single-free-parameter and constructed a control equation.Then,coupled with the analysis of statistical indicators and graphical patterns,the optimal generalized algebraic difference prediction model for the safety accident rate of elevator equipment was determined.Finally,the data of elevator equipment accidents in China over the past 16 years were used to verify the effectiveness of the optimal prediction theoretical model.The results show that the accident rate of elevator equipment in China presents an overall decreasing trend,but the decreasing trend of the accident rate per 10000 units slows down.Using the generalized algebraic difference approach,the model derives from the exponential equation with the deviation of 0.0022,the root mean square error(RMSE)of 0.045352 and the fitting degree of 0.8868 is the best prediction model,and the relative error of its prediction value is 5.9%.The prediction model can be updated according to the real-time data such as the amount of registered elevator equipment and the number of accidents rate per 10000 units,and can effectively predict the accident rate of elevator equipment in China.

关 键 词：电梯设备 安全事故 广义代数差分法(GADA) 代数差分法(ADA) 预测模型 万台事故率 

分 类 号：X952[环境科学与工程—安全科学]