一般非线性函数的广义统计预测模型的研究与应用

摘要

A factor decomposition was carried out on general nonlinear function by applying calculus fundamental principle, discriminance method of "constribution size" was obtained about independent variables to the nonlinear function value, and "generalized statistics prediction model" was put forward about general nonlinear function. The blasting pile morphology prediction in Heidaigou Open-pit Coal Mine shows the absolute accuracy of the prediction is above 95%. The comparative analysis with the use of neural network to predict the results in the same period shows that the accuracy of the"generalized statistical prediction model" is far above the accuracy of neural network prediction model.%应用微积分基本原理,对一般非线性函数进行了因子分解,得出了自变量对非线性函数值的“贡献大小”的判别方法,提出了针对一般非线性函数的“广义统计预测模型”.对黑代沟露天煤矿爆堆形态的预测表明其预测绝对准确度达到95％以上,与同期使用神经网络预测结果进行的对比分析表明,“广义统计预测模型”准确度远远高于神经网络预测模型.