基于改进事件触发与1-D CNN的MMC冗余子模块开路故障实时诊断策略

Real-time open-circuit fault diagnosis strategy for MMC redundant sub-modules based on improved event-triggered and 1-D CNN

在模块化多电平换流器(MMC)中存在大量的子模块(SM),其可靠性直接影响整个系统的安全稳定运行。本文引入了具有容错能力的新型冗余SM,并在此基础上提出了一种基于改进事件触发与一维卷积神经网络(1-D CNN)的MMC冗余SM开路故障实时诊断策略。首先,选定MMC中特定SM构成的集成单元的电容电压作为事件触发对象,大大减少了传统诊断策略中需要观测的电容数量;其次,从有效地减少计算负担角度出发,改进事件触发流程,并结合1-D CNN算法有条件地对集成单元电容电压和桥臂电流波动情况分别进行检测,构建出MMC故障实时诊断系统;最后,综合考虑故障集成单元与故障桥臂位置定位出开路SM,进而完成新型冗余SM条件下MMC的容错运行。利用Matlab/Simulink搭建的19电平MMC模型,验证了所提实时诊断策略的有效性。

There are a large number of sub-modules(SM)in the modular multilevel converter(MMC),and its reliability directly affects the safe and stable operation of the whole system.In this paper,a new redundant SM with fault tolerance is introduced,and on this basis,a real-time open-circuit fault diagnosis strategy for MMC redundant sub-modules based on improved event-triggered and one-dimensional convolutional neural network(1-D CNN)is proposed.Firstly,the capacitor voltage of the integrated unit composed of specific SM in MMC is selected as the event-triggered object,which greatly reduces the number of capacitors that need to be observed in the traditional diagnosis strategy.Secondly,from the perspective of effectively reducing the computational burden,the event-triggered process is improved,and the 1-D CNN algorithm is used to detect the fluctuation of the capacitor voltage and the bridge arm current of the integrated unit,and the MMC fault real-time diagnosis system is constructed.Finally,considering the fault integration unit and the fault bridge arm position,the open-circuit sub-module is located,and then the fault-tolerant operation of MMC under the new redundant SM condition is completed.The effectiveness of the proposed real-time diagnosis strategy is verified by the 19-level MMC model built by Matlab/Simulink.