无偏灰色模糊马尔可夫链在设备衰退预测中的应用

Application of unbiased grey-fuzzy-Markov chain in equipment degradation prediction

针对传统灰色马尔可夫模型在固有灰色偏差和抗干扰性方面的不足,提出一种基于无偏灰色模糊马尔可夫链的设备衰退趋势预测模型。首先,引入无偏灰色理论,建立无偏灰色模型,预测设备健康状态的总体趋势;然后,根据此趋势,采用模糊集合理论进行模糊状态划分,从分类方法上改进传统灰色马尔科夫模型,同时克服马尔可夫状态矩阵运算量大的缺点;最后,进行模糊马尔可夫残差修正,确定设备健康状态组合预测值。与传统灰色马尔可夫模型相比,该模型可有效提高预测精度。以屏蔽泵的健康状况数据为样本进行设备衰退趋势预测,结果表明:该模型对第13、14和15个周期的设备健康状态的预测,残差偏移率分别为0.24%、0.10%和-0.05%,对应传统灰色马尔科夫模型的残差偏移率0.46%、0.11%和0.08%,预测精度更高,能够有效实现设备衰退趋势的精确预测。

To eliminate the inherent grey bias and improve the anti-jamming performance of the standard grey-Markov prediction model, a prediction model of equipment degradation based on Unbiased grey-fuzzy-Markov chain method is proposed. Firstly, the unbiased grey theory is introduced to establish an unbiased grey model for the prediction of the overall trend of equipment health conditions. Then according to the trend, fuzzy classification is brought in the fuzzy states division to improve the traditional clas sification method and overcome the disadvantage of large volume calculation of state transition matrix in Markov chain. Finally, the ultimate prediction values are determined through the fuzzy Markov deviation correction. By taking the degradation data of a canned motor pump as samples, the improved model is applied to predict its degradation. The results show that deviation rates of the 13th .14thand 15th cycle are 0. 24%, 0. 10% and --0.05%, respectively. They are better than the results of traditional grey- Markov model, which are 0. 46%, 0. 11% and 0. 08%. Therefore, compared with the standard grey-Markov model, this unbi ased grey-fuzzy-Markov chain model is more efficient and accurate.