面向复杂机电系统多源状态信息的运行可靠性评估方法研究

According to the reliability estimation of complex electromechanical system，the traditional reliability estimation methods are limited for some reasons, such as the difficulties in time-dynamic process description and non-individual characteristics. Degradation-based reliability estimation methods are required to know the professional background knowledge and degradation mechanism, or make assumptions about the degradation tracks and the probability distribution of the state features. To solve the above problems, the operational reliability estimation methods for complex electromechanical system based on multi-source condition information are proposed. The efficient LMD decomposition algorithm is constructed and the low resonance component demodulation analysis of RBSD is adopted to extract the complex signal modulation characteristics. Some salient features are selected from signal parameters which having a precise definition and a clear trend change as data base of the reliability assessment. The PHM, PCM, logistic regression and Logistic+PCM composite model are respectively proposed to implement operational reliability under the conditions of the large sample and small sample. The key factors affecting the system reliability are identified based on the analysis of electromechanical system dynamics, tribology theory and degradation failure principle. The project aims at assessing and forecasting the performance degradation process of the key equipment online, dynamically and initiatively. The operational condition-based reliability estimation methods can also effectively improve accuracy and credibility of the assessment to ensure the efficient and safe operation of the equipment in service period. The methods can also guide the auxiliary repair, reduce and eliminate accidents as possible, save repair cost and provide scientific management and maintenance policy. It is the richness and development of reliability assessment theory.

在分析复杂机电系统可靠性时，传统可靠性评估方法在描述时间、动态过程等方面存在困难，个体针对性不强；退化可靠性评估方法需要了解设备的退化机理，或对状态参数的退化路径和概率分布做过多假设。针对以上问题，提出了基于多源状态信息的运行可靠性评估方法。通过构建高效的LMD分解算法和RBSD低共振分量解调分析，提取复杂信号调制特征；将具有准确定义和明显趋势性变化的信号参数作为状态特征指标，为可靠性评估提供数据基础；提出PHM、PCM、Logistic回归模型和Logistic+PCM复合模型分别解决大样本和小样本条件下的运行可靠性分析问题；根据机电系统动力学、摩擦学理论和退化失效原理，找出影响系统可靠性的关键因素；在线、动态、主动地评估预测关键设备性能退化过程，提高其可靠性分析与评估的可信度；为设备的管理和维护提供科学的决策依据，尽可能减少和消除事故，节约维修费用；丰富和发展可靠性评估理论。

项目围绕复杂机电系统，从系统运行状态出发，通过监测设备测取系统的多源动态信号，提取反映设备性能退化过程的显著特征指标，结合可靠性建模技术，形成基于运行状态的可靠性方法及相关技术，主要的特色与创新点如下：（1）设备运行过程中的退化机理研究。研究复杂机械系统关键部件的轴承和齿轮摩擦学问题。基于各向异性分形几何理论，建立固定结合部法向接触力学模型。采用二变量WM函数模拟各向异性三维分形轮廓表面。建立了法向载荷、最大静摩擦力、静摩擦系数的改进分形模型。根据整个柔性结合面的结构函数给出识别柔性结合面分形维数、分形粗糙度的理论及实验方法，对静摩擦系数的理论解进行了理论和定量实验佐证。（2）多源状态信号特征提取方法。针对设备故障信号中常见的调制性特点，提出了基于LMD解调的特征提取方法，通过构造不同信噪比水平的加噪仿真信号对LMD相位展开算法及其噪声影响性进行了分析，给出了LMD波形和瞬时频率提取的信噪比阈值。针对设备故障信号中常见的冲击性特点，提出了一种小波降噪与RSSD相结合的振动信号特征提取技术。先通过小波阈值降噪方法明显减小信号中的噪声，随后对降噪后的信号进行共振稀疏分解，将信号分解为具有持续振荡特性的高共振分量和具有瞬态冲击特性的低共振分量。最后通过对分解所得到的低共振分量采用Hilbert包络解调方法提取冲击故障特征。采用BITD、Teager能量算子和对角切片谱相结合的方法、提出了基于形态滤波和时延自相关的时频切片等分析方法。采用多结构元素差值形态滤波和时延自相关方法对信号进行降噪，用FSWT分解降噪后的轴承振动信号，然后根据轴承故障特征频率选择时间频率切片区间，进行细化分析来提取故障特征。（3）运行可靠性评估方法。在大样本条件下，提出了基于Logistic回归模型的设备运行可靠性评估方法，可以直接建立设备运行状态信息与可靠度间的关联关系，通过设备的状态特征指标和对应的设备状态即可实现建模和实施评估。提出了PCM可靠性评估方法，通过基本协变量函数，响应协变量可对系统故障率函数进行不断更新，更加准确地揭示了状态监测数据与设备运行安全性和可靠性的映射关系。在小样本条件下，提出了小样本运行可靠性评估方法，包括Logistic+PCM 复合模型，直接从历史状态数据求解初始故障率函数，减少了主观因素减少了主观因素引入对评估精度造成的影响，可更加准确地判定系统的时间动态特性。

内容获取失败，请点击重试