基于Erlang-HSMM的设备剩余寿命预测研究

对于关键复杂设备进行健康诊断和设备剩余寿命预测,提出了一种基于爱尔朗分布和隐半马尔可夫模型的联合剩余寿命预测模型(Erlang-HSMM,E-HSMM)。首先,提出了改进的前后向算法、维特比算法和BaumWelch算法,有效地降低了模型的计算复杂度;其次,基于爱尔朗分布改进设备的健康状态逗留时间,将状态逗留时间分为已遍历和未遍历两个部分,提出新的健康状态逗留时间的概率分布;最后,针对状态监测数据,利用失效率理论构建设备剩余寿命预测模型。通过美国Caterpillar公司液压泵的状态监测实际数据进行评价与验证,实验结果表明,E-HSMM模型对设备的状态诊断和剩余寿命预测更加符合实际状况,比传统的隐半马尔可夫模型(HSMM)更有效。

Research on residual life prediction of equipment based on Erlang-HSMM

This paper proposed a combined residual life prediction model(Erlang-HSMM)based on Erlang distribution and hidden semi-Markov model for health diagnosis and life prediction of key complex equipment.Firstly,it proposed new forward-backward algorithm,Viterbi algorithm and Baum-Welch algorithm,which could effectively reduce the computational complexity of the model.Secondly,based on the Erlang distribution,it improved the health state residence time of the equipment.It divided the state residence time into two parts,traversed and untraversed,and proposed the probability distribution of the new state residence time.Finally,according to the state monitoring data,it used failure rate theory to establish the residual life prediction model of the equipment.Through the evaluation and verification of the actual monitoring data of the hydraulic pumps of caterpillar,the experimental results show that the improved HSMM based on Erlang distribution is more in line with the actual conditions for the condition diagnosis and residual life prediction of the equipment,and is more effective than the traditional HSMM.