GM（1,1）模型改进算法及其应用研究

摘要

Original series pattern,1-AGO sequence′s smooth degree and its initial value are main factors affecting the prediction accuracy of GM（1,1） model.To enhance the prediction accuracy,according to the investigation of original method on improving smooth degree of series,a new grey improved algorithm based on weighted and minimum average relative error is put forward.Comparative results with original method show that this new algorithm is more accurate and useful for other general series.The traditional and improved algorithms are applied to practical machine fault diagnosis and life prediction.According to the experimental results,it can be concluded that the new grey improved algorithm can offer very valuable information for preventative maintenance of machinery equipment.%原始序列规律、一次累加生成序列光滑度及其初始值是影响GM（1,1）模型预测精度的主要原因.为了获得较高的预测精度,在原有改善序列光滑度方法的基础上,给出一种基于加权和最小平均相对误差的灰色改进算法.该改进算法不但能够提高拟合及预测精度,而且拓展了传统GM（1,1）预测模型的适用范围.将传统方法与改进算法应用于实际设备故障诊断和寿命预测,结果表明,改进算法对于机械设备的预知维修具有较好的参考价值.