A Comparative Study On Forecasting The Work Year Of Coal Workers’ Pneumoconiosis Based On The Combined Model Of Neural Network Model

Objective To study the pros and cons of prediction performance of the combinedmodel based on the Neural Network Model when use the models to forecast the workyear of Coal Workers’ Pneumoconiosis.Method The research data was analyzed by SPSS19.0software. In order to determinethe operating parameters of all the models, we need to repeat training and study of BPneural network and RBF neural network. The true value and predict value of all modelswere applied comparative t-test to perform statistical analysis. Primary performanceanalysis of the design is made through the scatter diagram. The weight coefficient ofcombination model was defined by standard difference method. Standard error、averagerelative error and mean absolute error were applied to analyze the predicting outcomes ofthe two models in order to achieve the aim of comparing the prediction performance.Results The scatter diagram was made by the output results and real value, besidesfewer splashes, the model had global liner trend distribution of all this models. The bestone is the combination model combined by this three singleness model among all thisscatter diagrams, in the second place is the combination model combined by BP neuralnetwork and Multiple linear regression model;The worst on is RBF neural network.There was no significantly difference in true value and predicted value of all this models.The standard error of BP neural network、RBF neural network、Multiple linearregression model、the combination model combined by BP neural network and Multiplelinear regression model、the combination model combined by BP neural network andRBF neural network、the combination model combined by RBF neural network andMultiple linear regression model and the combination model combined by this threesingleness model was5.31、7.48、4.89、2.06、5.92、6.15and3.21；The averagerelative error was-0.51%、2.80%、-0.57%、0.45%、0.68%、0.87%and0.50%；Themean absolute error was3.16、5.57、4.93、1.19、3.51、3.01and4.40. In order tofurther validate the predictable performance of this models, we use ten simulationsamples to test. The standard error of BP neural network、RBF neural network、Multiple linear regression model、the combination model combined by BP neuralnetwork and Multiple linear regression model、the combination model combined by BP neural network and RBF neural network、the combination model combined by RBFneural network and Multiple linear regression model and the combination modelcombined by this three singleness model was3.13、6.48、4.57、1.57、4.92、5.15and1.69; The average relative error was-0.43%、1.97%、-0.59%、0.07%、0.53%、0.85%and0.19%；The mean absolute error was2.83、5.31、4.15、0.73、3.79、4.23and3.63.Conclusion The study shows that in the prediction of the work year of Coal WorkersPneumoconiosis, the prediction performance of combination model was superior to thosesingleness models, the combination model combined by BP neural network and Multiplelinear regression model has the best prediction performance. It’s prediction accuracy ishigh and the prediction results are reliable.