基于多源多维信息优化及集成信息融合的永磁同步电机故障诊断研究

Permanent magnet synchronous machine (PMSM) has been widely used in electrical vehicle, wind power and etc, where the high degree of reliability is an important requirement for PMSM. Fault diagnosis technology is an important way to improve the reliability of PMSM. In order to solve the diversity between failure causes and fault symptoms of PMSM, this project introduces information fusion technology into fault diagnosis of PMSM, and a fault diagnosis technology for PMSM is proposed based on multi-source multi-dimension information optimization and ensemble information fusion to improve the accuracy and reliability of fault diagnosis. This project mainly focuses on the fundamental scientific problems, including fault characteristics and mechanism of PMSM, extraction of fault features in the time domain, frequency domain and time-frequency domain of the multi-source signals (vibration signal, stator current signal and zero sequence signal), optimization of multi-source multi-dimension information and integration of information fusion techniques. The theoretical and technical innovations are achieved, and the relatively complete fault diagnosis theory and technical framework are established for PMSM. Hence, this investigation will provide the theoretical foundation and technical support for further research and practical application of fault diagnosis of PMSM. Meanwhile, it will enrich the research contents for fault diagnosis of PMSM. This project not only has an important scientific significance and academic value, but also has a promising application prospect.

永磁同步电机已在电动汽车和风力发电等领域得到了广泛应用，可靠性问题是制约其在各领域应用的重要因素，故障诊断技术是提高其可靠性的重要途径。针对永磁同步电机故障起因与故障征兆之间的多元性，以提高故障诊断的准确性和可靠性为目标，本项目将信息融合技术引入到永磁同步电机的故障诊断，进而提出基于多源多维信息优化及集成信息融合的永磁同步电机故障诊断技术。本项目旨在研究永磁同步电机的故障特征和机理、多源信号（振动信号、定子电流信号和零序信号）的时域、频域和时频域上特征信息的提取、多源多维信息的优化、信息融合技术的集成等基础科学问题，取得理论与技术创新，形成较为完整的永磁同步电机故障诊断理论与技术体系。因此，本项目将为永磁同步电机故障诊断的后续研究和实际应用提供理论基础和技术支持，同时也将丰富永磁同步电机故障诊断的研究内容，不仅具有重要的科学意义与学术价值，而且应用前景广阔。

电机是一个复杂的机电耦合系统，其故障起因与故障征兆之间往往呈现出多元性。由于单一信号能够提供的特征信息量很有限，难以全面地表征电机的运行状态，同时信息融合能像人脑处理复杂问题一样，利用计算机技术，通过对多源信息的融合处理，获得比单一信息更全面、更丰富的综合信息。因此，本项目针对电机故障起因与故障征兆之间的多元性，以提高永磁同步电机故障诊断的准确性和可靠性为目标，开展了基于多源多维信息优化及集成信息融合的永磁同步电机故障诊断研究。本项目重点了研究永磁同步电机的故障机理和故障特征、多源信号的时域、频域和时频域上特征信息的提取、多源多维信息的优化、信息融合技术及其集成等关键理论与技术，并在此基础上取得了一系列的研究成果：（1）揭示了永磁同步电机定子绕组故障和永磁体退磁故障的发生机理及故障特征演变规律；（2）提出了基于零序分量和定子电流融合的永磁同步电机故障诊断方法；（3）提出了基于模型预测控制的永磁同步电机定子绕组匝间短路故障诊断方法；（4）提出了基于混合降维的多源故障特征信息优化算法；（5）提出了基于卷积神经网络的永磁同步电机故障特征优化与故障辨识技术；（6）提出了基于混沌自适应引力搜索和粒子群算法优化神经网络的永磁同步电机故障诊断方法；（7）提出了基于证据理论的多信息融合的永磁同步电机故障诊断方法；（8）初步拓展研究了永磁同步电机故障容错控制，提出了永磁同步电机定子绕组不对称故障容错控制方法。本项目在上述研究成果的基础上，总结和提炼出了一般性、规律性的结论和方法，提升了永磁同步电机故障诊断理论研究水平，同时促进其在工程实际中的应用。