基于灰色马尔可夫链的核设备退化趋势预测

核设备的状态退化趋势预测是确定其在役检查以及维修计划的重要依据，但由于核设备样本小、退化数据缺乏、退化轨迹具有波动性，难以采用传统的概率统计模型对其退化趋势进行精确预测。为此，本文提出应用灰色马尔可夫链模型对核设备退化趋势进行预测的方法，该方法充分利用GM（1，1）和马尔可夫链的优点，能够有效提高核设备退化趋势预测的精度。并以屏蔽泵的退化数据为样本，精确预测了屏蔽泵的退化趋势，同时与GM（1，1）模型的预测结果进行了对比。结果表明，灰色马尔可夫链模型的预测精度更高，能够对核设备的退化趋势进行精确预测。

Prediction Degradation Trend of Nuclear Equipments Based on GM(1,1)-Markov Chain

The degradation trend prediction results are important references for nuclear equipments in-service inspection and maintenance plan. But it is difficult to predict the nuclear equipments degradation trend accurately by the traditional statistical probability due to the small samples, lack of degradation data and the wavy degradation locus. Therefore, a method of equipment degradation trend prediction based on GM(1, 1)-Markov chain was proposed in this paper. The method which makes use of the advantages of both GM(1, 1) method and Markov chain could improve the prediction precision of nuclear equipments degradation trend. The paper collected degradation data as samples and accurately predicted the degradation trend of canned motor pump. Compared with the prediction results by GM(1, 1) method, the prediction precision by GM(1, 1)-Markov chain is more accurate.