基于电源纹波的功率负载旁路检测方法

传统功率器部件的工作状态往往需要大量的传感器资源,并占用较多的无线/有线复用通信资源。为了实现功率器件的智能状态检测,降低使用传感器的复杂性,增强系统的可靠性,提出了一种基于电源状态检测的功率器部件旁路多点状态的检测方法。该方法的创新点体现于在不修改原有电路结构的情况下,旁路采集电源纹波作为功率器件工作状态的重要特征。在此基础上应用信号处理算法分析电源纹波的频域信息并构建数据集,进一步通过机器学习方法区分功率器件的各自工作状态。通过对大功率LED二极管PWM调制的纹波进行探测,对提出的相关方法进行了实验验证。实验结果表明,该方法检测性能优越、对环境不敏感,实现了更加智能、可靠、可拓展的功率器部件智能检测。

Power load bypass detection method based on power supply ripple

Real-time condition detection and evaluation of power devices plays an important role in industrial production to ensure product quality, economic efficiency and safety. The traditional operating status of power devices often requires a large number of sensor resources and occupies more wireless/wired multiplexed communication resources. In order to achieve intelligent state detection of power devices, reduce the complexity of using sensors, and enhance the reliability of the system, this paper proposes a power device component bypass multi-point state detection method based on power state detection. The innovation of the method is to bypass the power supply ripple as an important feature of the power device operating state without modifying the original circuit structure. Based on this, a signal processing algorithm is applied to analyze the frequency domain information of the power ripple and construct a data set to further distinguish the respective operating states of the power devices by machine learning methods. In this paper, the proposed correlation method is experimentally verified by detecting ripple of PWM modulation of high-power LED diodes. The experimental results show that the method has superior detection performance and is insensitive to the environment, achieving a more intelligent, reliable, and scalable intelligent detection of power device components.