BP神经网络与多重线性回归模型在煤工尘肺发病工龄预测中的比较研究  

A COMPARATIVE STUDY ON FORECASTING THE WORK YEAR OF COAL WORKERS'PNEUMOCONIOSIS OF BP NEURAL NETWORK AND MULTIPLE LINEAR REGRESSION MODEL

作　　者：武建辉[1] 王国立[1] 

机构地区：[1]河北省煤矿卫生与安全实验室河北联合大学公共卫生学院流行病与卫生统计学学科,河北省唐山市063000

出　　处：《中国煤炭工业医学杂志》2014年第12期1992-1995,共4页Chinese Journal of Coal Industry Medicine

基　　金：河北省科技支撑项目(11276911D);河北省卫生厅医学重点项目(20120146);唐山市科技支撑项目(11150205A-3)

摘　　要：目的研究BP神经网络与多重线性回归预测模型对煤工尘肺发病工龄进行预测时各模型的预测性能及其比较。方法 BP神经网络模型和多重线性回归模型对研究数据进行预测分析时采用SPSS18.0对其实现。采用标准误差、平均相对误差和平均绝对误差对两模型的预测结果进行分析,得出各模型的预测效果,进而比较两模型预测性能的优劣。结果多重线性回归模型、BP神经网络模型真实值与预测值之间的差异均无统计学意义,t值分别为0.000、0.168,P值分别为1.000、0.876。多重线性回归、BP神经网络的标准误差分别为2.39、2.31;平均相对误差分别为7%、6%;平均绝对误差分别为1.77、1.67。结论研究表明在煤工尘肺病患者发病工龄的预测中,BP神经网络模型的预测性能优于多重线性回归,并且其预测精度高,预测结果可靠,值得推广应用。Objective To study the prediction performance of BP Neural Network and multiple linear regression model when using them to forecast the work year of Coal Workers’ Pneumoconiosis and then make a comparison.Methods The research data was analyzed by SPSS18.0software.Standard error,average relative error and mean absolute error were applied to analyze the predicting outcomes of the two models.So we could get the predictive effect of each model and then compare the prediction performance of BP Neural Network and multiple linear regression model.Results There was no significantly difference in true value and predicted value of BP Neural Network and multiple linear regression model,t value were 0.168 and 0.000 respectively,p value were 0.876 and 1.000 respectively.The standard error of multiple linear regression model and BP Neural Network were 2.39 and 2.31 respectively;the average relative error of multiple linear regression model and BP Neural Network were 7% and 6%respectively and the mean absolute error were 1.77 and 1.67 respectively.Conclusion The study shows that in the prediction of the work year of Coal Workers’ Pneumoconiosis,the prediction performance of BP Neural Network is superior to multiple linear regression model,and the prediction accuracy is high,the prediction results are reliable,so BP Neural Network is worthy of spreading to application.

关 键 词：BP神经网络 多重线性回归模型 发病工龄 预测 煤工尘肺 职业病 

分 类 号：R135.2[医药卫生—劳动卫生]