Hybrid Condition Monitoring for Power Converters: Learning-Based Methods With Statistical Guarantees

Abstract

In this paper, a new learning-based method is proposed for the early detection of changes of parameters in power converters. It circumvents the pertinent shortcomings of previous model-based methods, such as their need for acquiring switching signals, dependence on the control method, or the need for isolating a part of the system during monitoring and thus interfering with the system performance or its start-up time. Additionally, a downside of the learning-based approaches, which is that their performance depends on the quality of the measurements, is addressed through constructing hybrid models that combine the benets of both lines of development. Our approaches are evaluated with several types of features based on wavelet decomposition and empirical mode decomposition. Using these, an ANN-based classier is trained for fault detection. For achieving the nal decision on the presence or absence of a fault state, we propose the use of sequential hypothesis testing. This hybrid approach yields a much higher reliability than instantaneous ANNbased classication, while allowing for a statistically sound cross-temporal information integration and providing a controllable error bound for the probability of misclassications. The proposed method was evaluated on two data-sets recorded from a buck converter and an arm of a modular multilevel converter. The results show that, for both systems, the proposed method is capable of reliably recognizing changes of the system parameters. The potential for practical applications is shown through an implementation on a low-power-consumption microcontroller. INDEX TERMS Fault diagnosis, power converters, DC/DC Converter, MMC, machine learning, signal processing, hybrid models I. INTRODUCTIONC ONDITION monitoring of power converters in energy systems is of great importance, considering the wide range of applications in which they play a critical role, and their long operational time. While there are many dierent power converter topologies, these broadly share the same principles of operation via active and passive elements. Hence, very similar condition monitoring methods can, with some modications, be applied for dierent power converters.The specic condition monitoring technique proposed in this paper is applied to two dierent power converter topologies. The rst is a buck converter, a widely employed DC-DC power converter [1]. The second system is an arm of a of modular multilevel converter (MMC). These converters have gained great interest in academia as well as industry shortly after they were initially proposed [2], due to their modular design, their