【摘要】 针对一般GM(1,1)多步预测方法的不足,提出了一种基于代谢递补GM(1,1)的状态预测方法.该方法首先对原始信号进行预处理,再通过递补思想进行多步预测,然后利用更新数据进行代谢预测,最后计算设备状态的预测值与真实值误差,根据预测值的趋势判断设备的未来状态.实例分析结果表明,该方法所需数据样本少且数据训练时间短,后八步的预测精度可满足工程精度,能有效地应用于设备的中长期在线状态预测.更多还原

【Abstract】 Considering the deficiencies of general multi-step predictive method based on GM(1,1)(grey model),a condition prediction method based on metabolic filling GM(1,1) was proposed.First,the multi-step predictive value was computed by filling idea after the original signals were pretreated.Then,the data was predicted metabolically using the refresh data.Finally,the error between the predictive value and the real value of equipment condition was computed,and the future condition of equipment was estimated according to the trend of predictive value.Practical examples demonstrated that only a few samples were required and the training time was short.The predictive precision of the following eight steps can reach the engineering precision.The method can be applied to medium and long term on-line condition prediction of equipment.更多还原