基于生物进化论消噪算法的齿轮传动系统故障特征多尺度提取研究

It is known that the early fault in gear transmission system is hard to detect, the root cause of this problem is that the early fault vibration characteristics are weak, and the signal to noise ratio of early fault vibration characteristics is low affected by the large amount of non-stationary coupled operation noise in the observation signals. Therefore, aiming to eliminate the non-stationary coupled noise in the observation signal of the gear transmission system and extract early fault-related weak vibration features, three parts are studied based on evolutionary adaptive filter method and multi-scale decomposition theory. 1) Research on the non-stationary coupled noise characteristics analysis and characterization method, and construct the parametric modeling of the non-stationary coupled noise of the gear transmission system, provide the foundation for the further elimination of noise; 2) Conduct evolutionary adaptive filter denoising research, and establish the model of filter population dynamic reproduction, realize global optimal suppression and elimination of non-stationary coupled noise; 3) Carrying out multi-scale composite extraction and identification research of fault vibration characteristics, Construct signal adaptive multi-scale decomposition and reconstruction model, achieve accurate extraction and identification diagnosis of weak vibration characteristics of early faults. Based on the research of this project, it is expected to provide a new theoretical basis and technical support for the realization of efficient and highly accurate gear transmission system signal noise elimination and fault diagnosis.

齿轮传动系统早期故障难以准确诊断，其根本原因是早期故障振动特征微弱，又受环境与运行条件等因素影响，齿轮传动系统观测信号中包含了大量的非平稳耦合运转噪声，导致早期故障振动特征信噪比低。因此，本项目拟基于生物进化论自适应滤波方法和信号多尺度分解原理，围绕齿轮传动系统非平稳耦合噪声消除与早期故障微弱振动特征提取问题，1)开展齿轮传动系统非平稳耦合噪声特性分析与表征方法研究，完成齿轮传动系统非平稳耦合噪声的参数化建模与表征，为噪声的准确消除奠定基础；2)开展生物进化论自适应消噪算法研究，提出滤波器种群动态繁殖迭代方法，实现非平稳耦合噪声的全局最优消除；3)开展故障振动特征多尺度提取研究，构建信号自适应多尺度分解与重构模型，实现早期故障微弱振动特征的准确解调提取和识别诊断。通过本项目的研究，期望为实现高效和高精准的齿轮传动系统振动信号噪声消除和故障诊断提供新的理论基础和技术支撑。

本项目基于生物进化论自适应滤波方法和信号多尺度分解原理，围绕齿轮传动系统非平稳耦合噪声消除与早期故障微弱振动特征提取问题，1)开展了齿轮传动系统非平稳耦合噪声特性分析与表征方法研究，构建了误差影响下的齿轮系统局部故障激励产生和传递机理，分析了故障特征与干扰信号之间的关系和分布特征；2)开展了生物进化论自适应消噪算法研究，提出了基于动态生物进化策略和多尺度进化的种群全局优化方法，实现了消噪滤波器系数全局最优的寻找，提升了早期微弱故障的信噪比；3)开展故障振动特征多尺度提取研究，发明了多尺度时频协同解调的微弱特征提取方法，构建了信号自适应多尺度分解与重构模型，实现早期故障微弱振动特征的准确解调提取和识别诊断。本项目在国内外高水平期刊上发表论文共计14篇，申请国家发明专利3项，成果支撑获得中国机械工业科学技术发明一等奖1项。