间歇性强干扰下轮毂电机机械故障特征提取与运行安全评估方法研究

It is a very urgent problem that performance degradation of in-wheel motor based on frequent mechanical faults can imperil operation security of electric vehicle. Aiming at operation security estimation of in-wheel motor, the methods both feature extraction of mechanical faults and operation security estimation of in-wheel motor will be researched under powerful intermittent interference. Then dynamics simulation of in-wheel motor under typical load excitations is utilized to trace back the causes of mechanical faults and explore the degree of fault deterioration together with the experiments of mechanical faults. Artificial hydrocarbon networks (AHNs) is used to filter powerful intermittent interference and extract the features of mechanical faults. The features parameters are employed to represent fault states, and are used to construct the fault diagnosis model of in-wheel motor. Next, it is investigated that three parameters of vibration, average efficiency and temperature will change on the basis of the fault type and fault degree of in-wheel motor, and estimation models of three parameters are built for fault risk by multiple-condition information. Finally, the evaluation information of three parameters in multiple stages is united to evaluate the operation security of in-wheel motor. In this way, innovation and breakthrough in the development of feature extraction for weak non-stationary signal with powerful intermittent interference can be expected, which will improve the research of operation security estimation for in-wheel motor based on condition information.

针对电动汽车轮毂电机机械故障频发导致电机性能下降危及车辆运行安全的实际问题，瞄准轮毂电机运行安全评估，开展间歇性强干扰下轮毂电机机械故障特征提取及运行安全评估的方法研究。从典型载荷激励下轮毂电机动力学仿真研究入手，结合轮毂电机机械故障试验，追溯机械故障形成原因，探索故障恶化过程。基于人工碳氢网络技术，研究间歇性强干扰信号滤除方法，提取轮毂电机机械故障特征，决选能够较好表征故障状态的特征参数，构建轮毂电机机械故障诊断模型；研究故障类型及其程度对轮毂电机振动、平均效率及温升等参数的影响规律，融合多状态信息构建三个参数的故障风险评估模型；基于Bayes定理及三个参数故障风险评估模型多阶段评估信息，建立多模联动的轮毂电机运行安全评估方法。通过本项目研究，可望在间歇性强干扰下微弱非平稳故障特征提取和基于运行状态信息的轮毂电机运行安全评估等方面有所创新和突破，获得有价值的研究成果。

针对电动汽车轮毂电机机械故障频发导致电机性能下降危及车辆运行安全的实际问题，瞄准轮毂电机运行安全评估，开展间歇性强干扰下轮毂电机机械故障特征提取及运行安全评估的方法研究。在轮毂电机振动响应特征研究方面，仿真分析了典型载荷激励下轮毂电机不平衡电磁力的大小和作用点，探明了轮毂电机机械故障的形成原因，研究了其振动响应特征，阐述了间歇性强干扰下轮毂电机轴承故障特征。在故障特征提取方法研究方面，研究典型载荷激励对轮毂电机机械故障特征的影响程度，探索间歇性强干扰下轮毂电机机械故障特征的响应规律，基于人工碳氢网络、共振稀疏分解、非负权系数稀疏表示等技术，滤除间歇性强干扰信号，提取轮毂电机机械故障特征，并优化了汽车真实运行工况下轮毂电机故障特征提取的算法。在故障诊断与运行安全评估方法研究方面，基于改进t分布随机邻域嵌入、狼群算法、综合权重诊断指标法、粗糙集理论，探索轮毂电机故障特征参数的降维、融合和离散化方法，揭示了特征参数集与轮毂电机故障状态之间的映射关系；基于贝叶斯网络、动态贝叶斯网络、人工碳氢网络、隐马尔科夫模型等模式识别方法，构建了多输入多输出的轮毂电机故障诊断系统、多模联动多阶段协同轮毂电机运行安全评估系统。上述研究在间歇性强干扰下微弱非平稳故障特征提取和轮毂电机故障诊断与运行安全评估等方面有所创新和突破，可为研发轮毂电机运行安全评估系统提供基础理论研究和技术支撑。