论文部分内容阅读
目的探讨基于BP神经网络的组合模型在煤工尘肺发病工龄预测中的应用。方法采用SPSS18.0中的BP神经网络模型和多重线性回归模型对数据进行分析预测,运用最小二乘加权的方法对二模型进行加权拟合,采用平均相对误差对各模型预测结果进行分析,比较各模型的精确度,从而评价各模型的预测性能。结果由BP神经网络进行预测可以得出发病工龄预测值和真实值大致位于从原点起始的45°线上,符合理想状态下值的分布情况。在对数据进行多重线性回归分析后,得到R=0.967,R2=0.935,对方程进行检验,F=1367.408,P=0.000,表明可以应用此多重线性回归方程进行预测分析。BP神经网络模型,多重线性回归模型,组合模型的均方根误差分别为0.057、0.057、0.052;平均绝对误差分别为1.4、1.46、1.38;平均相对误差分别为0.17、0.12、0.02。结论实证表明,该组合模型的预测精确度比常规BP神经网络模型要好。组合模型的算法概念明确,计算简便,有较高的拟合和预测精度。
Objective To explore the application of combined model based on BP neural network in predicting the age of coal workers’ pneumoconiosis. Methods BP neural network model and multiple linear regression model were used to analyze and predict the data. The least square method was used to weigh the two models. The average relative error was used to analyze the model predictions. The accuracy of each model to evaluate the predictive performance of each model. The results predicted by BP neural network can predict the predicted value of the length of service and the true value is located roughly from the origin of 45 ° line, in line with the distribution of ideal values. After multiple linear regression analysis of the data, R = 0.967 and R2 = 0.935 were obtained and the equation was tested, F = 1367.408, P = 0.000, indicating that this multiple linear regression equation can be used for predictive analysis. BP neural network model, multiple linear regression model and combined model have root mean square error of 0.057, 0.057 and 0.052, respectively; mean absolute errors are 1.4, 1.46 and 1.38, respectively; mean relative errors are 0.17, 0.12 and 0.02 respectively. Conclusion The empirical results show that the prediction accuracy of the combined model is better than that of the conventional BP neural network model. The concept of the combined model algorithm is clear, the calculation is simple, and the fitting and prediction accuracy are high.