采煤机调高泵隐半马尔可夫模型磨损故障预测

针对采煤机调高液压系统中调高泵故障很大程度依赖于设计者和用户的知识和经验水平，提出一种基于隐半马尔科夫模型（Hidden Semi-Markov Model，HSMM）的故障预测方法。首先，对采煤机调高液压系统中调高泵压力、流量，调高油缸活塞杆速度、位移进行时域分析。然后,通过对特征值进行K-均值聚类分析确定隐状态数。最后，通过MATLAB和Python编程实现调高泵HSMM故障预测。结果表明：与基于隐马尔可夫模型（Hidden Markov Model，HMM）预测方法相比，调高泵早期故障、后期故障和灾变故障测试样本的预测结果识别率分别提高了10%、5%、20%，基于HSMM的调高泵磨损故障预测方法具有可行性。

Research on Wear Fault Prognostics of Hidden Semi-Markov Model of Shearer Pump

Aiming at the fault of the adjustable pump in the hydraulic system of coal shearer, it depends on the knowledge and experience level of the designer and the user to a great extent. A fault prognostics method based on Hidden Semi-Markov Model (HSMM) is proposed. First of all, time domain analysis of hydraulic system for shearer, which were including pressure and flow of height adjustment pump and velocity and displacement of raise the cylinder piston rod. Then, the number of hidden states was determined by k-mean clustering analysis of the eigenvalues. Finally, HSMM fault prediction was realized by software program-ming of MATLAB and Python. The results show that recognition rate of predicted results is improved, which is compared with the Hidden Markov Model (HMM) prediction method. Thereinto, early fault recognition rate of height adjustment pump is increased 10%, late period fault recognition rate of height adjustment pump is increased 5%, catastrophe fault recognition rate of height adjustment pump is increased 20%. It is concluded that the abrasion fault prognostics method of adjustable pump based on HSMM is feasible.