最佳灰色回归组合模型及其在中国火灾预测中的应用

发布时间：2018-07-11 17:00

本文选题：灰色模型 + 回归模型　；参考：《数学的实践与认识》2008年06期

【摘要】：火灾每年给国家和人民生命财产造成巨大损失.火灾现象具有随机性、模糊性,是个复杂的灰色系统行为.研究火灾发生规律及发展趋势,具有实用价值.为此,首先给出最小二乘估计(LSE)意义下的最佳组合预测模型的定义,并求得组合模型的权的公式和证明权的唯一性.其次,用回归分析方法建立多个回归模型,并按以下三条标准:①回归指数(或相关系数)r大、②系统误差s小、③模型精度p高,选定最佳非线性回归模型;用灰色理论建立多个灰色模型,并按以下三条标准:①后验差比值c小、②小误差概率P大、③预测关联度ξ大,选定最佳灰色模型;再用最小二乘法将最佳回归模型与最佳灰色模型有机地结合起来建立的中国火灾最佳灰色回归组合预测模型.最佳灰色回归组合预测模型综合利用前两者提供的不同的有用信息,改善了单一模型的局限性,提高了模型的预测精度,减少了预测误差,使预测效果更佳.组合模型预测中国年火灾起数处于动态增长过程.
[Abstract]:Fire causes great loss of life and property to nation and people every year. The fire phenomenon is random and fuzzy, and it is a complex grey system behavior. It is of practical value to study the regularity and development trend of fire. For this reason, the definition of the optimal combination prediction model in the sense of least square estimation (LSE) is given, and the weight formula and the uniqueness of the proof weight of the combined model are obtained. Secondly, several regression models are established by regression analysis, and the best nonlinear regression model is selected according to the following three standard: 1 regression index (or correlation coefficient) r big error s small error. Several grey models are established by using grey theory, and the best grey model is selected according to the following three standard: 1 posterior error ratio c small error probability p big 3. The best grey regression prediction model for fire in China is established by combining the best regression model with the best grey model by using the least square method. The best grey regression combination forecasting model synthetically utilizes the different useful information provided by the former two models, improves the limitation of the single model, improves the prediction accuracy of the model, reduces the prediction error, and makes the prediction effect better. The combined model predicts that the beginning of annual fire in China is in the process of dynamic growth.
【作者单位】：福建农林大学计算机与信息学院
【基金】：福建省教育厅项目(JB04122)资助福建农林大学项目(011727)资助
【分类号】：D631.6

【相似文献】