基于奇异值谱熵复杂度的CRDM滚轮磨损程度识别研究

针对反应堆控制棒驱动机构（CRDM）中的滚轮丝杠运动副磨损状态难以评估的问题，引入复杂度概念，并基于奇异值谱熵构建复杂度指标，实现了滚轮磨损程度的识别。对观测时间序列进行相空间重构并计算奇异值，再结合信息熵的概念得到奇异值谱熵。经仿真、分析和验证，结果表明，该复杂度指标对于信号的不确定性较为敏感，可以有效的反映出不同磨损状态下信号频域的变化，进而判断出滚轮的磨损程度；滚轮随着运行时间增加，振动信号的高频成分显著增加，低频成分变化很小。因此，应用奇异值谱熵作为评价指标可以有效地识别滚轮的磨损状态。 

Roller Wear Degree Recognition of Control Rod Drive Mechanism Based on the Complexity of Singular Spectrum Entropy

Aiming at the problem of the wear condition monitoring of the roller in the control rod drive mechanism in reactors, the concept of complexity is introduced. As one of the complexity, the singular spectrum entropy is used to recognize the wear condition. After the reconstruction phase space of the observed data, the singular is calculated. Combined with the concept of entropy of information, the singular spectrum entropy is defined. The singular spectrum entropy is sensitive to the uncertainty of the observed data and shows the change of the components in the frequency domain which is related to the wear condition. By the experiment of the control rod drive mechanism in its whole life, it is found that the high-frequency components increase and the low-frequency components change little, with the running time increasing. The experiment proves that the method of singular spectrum entropy is quite valid in wear condition monitoring and recognition.