Biblio

Brushless synchronous generators (BSG) are typically used to produce an island network whose voltage is close to the nominal voltage of the generator. Generators are often used also in test-field applications where also zero output voltage is needed. The exciter construction and magnetic remanence may lead to a situation where the non-loaded generator terminal voltage cannot be controlled close to zero but a significant voltage is always generated because the exciter remanence. A new brushless synchronous generator excitation and de-excitation converter topology for test applications is proposed. The purpose is to achieve full voltage control from zero to nominal level without modifications to the generator. Insulated-gate bipolar transistor (IGBT) and Field-Programmable Gate Array (FPGA) technology are used to achieve the required fast and accurate control. In the work, simulation models were first derived to characterize the control performance. The proposed converter topology was then verified with the simulation model and tested empirically with a 400 kVA brushless synchronous generator. The results indicate that the exciter remanence and self-excitation can be controlled through the exciter stationary field winding when the proposed converter topology controls the field winding current. Consequently, in highly dynamical situations, the system is unaffected by mechanical stresses and wear in the generator.

In photovoltaic (PV) systems, machine learning-based methods have been used for fault detection and diagnosis in the past years, which require large amounts of data. However, fault types in a single PV station are usually insufficient in practice. Due to insufficient and non-identically distributed data, packet loss and privacy concerns, it is difficult to train a model for diagnosing all fault types. To address these issues, in this paper, we propose a decentralized federated learning (FL)-based fault diagnosis method for insulated gate bipolar transistor (IGBT) open-circuits in PV inverters. All PV stations use the convolutional neural network (CNN) to train local diagnosis models. By aggregating neighboring model parameters, each PV station benefits from the fault diagnosis knowledge learned from neighbors and achieves diagnosing all fault types without sharing original data. Extensive experiments are conducted in terms of non-identical data distributions, various transmission channel conditions and whether to use the FL framework. The results are as follows: 1) Using data with non-identical distributions, the collaboratively trained model diagnoses faults accurately and robustly; 2) The continuous transmission and aggregation of model parameters in multiple rounds make it possible to obtain ideal training results even in the presence of packet loss; 3) The proposed method allows each PV station to diagnose all fault types without original data sharing, which protects data privacy.

Series-connected IGBTs, when properly controlled, operate similarly to a single device with a much higher voltage capacity. Integrating series IGBTs into a Modular Multilevel Converter (MMC) can reduce its complexity without compromising the voltage capacity. This paper presents the circuit design on the sub-modular level of a MMC in which all the switching devices are series-connected IGBTs. The voltage sharing among the series IGBTs are regulated in a self-balancing manner. Therefore, no central series IGBT controller is needed, which greatly reduces the sensing and communication complexities, increasing the flexibility and expandability. Hardware experiment results demonstrate that the series IGBTs are able to self-regulate the voltage sharing in a fast and accurate manner and the system can operate similarly to a sub-module in a MMC.