Zhiwei Zhang

To achieve a comprehensive state monitoring and guarantee the reliability of dual active bridge DC-DC (DAB) converters, a modeling and parameter monitoring method based on digital-twin technology is proposed. The proposed method only requires non-intrusive sampling of the input and output voltage as well as the inductor current, without the need for an additional sampling circuit. Initially, the discrete mathematical model for the topological circuit of DAB converters has been developed. Here, a load current feedforward minimum current stress optimization (LCF-CSO) method is proposed to reduce the power losses and improve the transient performance of DAB converters. Secondly, the external operating data of the physical DAB converter are collected to drive the digital-twin model, thereby reproducing the characteristics of the real system. Thirdly, an objective evaluation function is constructed, and the key parameters are identified iteratively by particle swarm optimization (PSO) algorithm. As a result, an off-line digital-twin model is developed to enable reliable and non-invasive state monitoring of the capacitors, inductors, and power MOSFETs. Finally, the reliability of the proposed LCF-CSO and, the digital-twin model condition monitoring method is verified by experiments, and the monitoring error is within 5 %, which contributes to effective component state monitoring.

Considering that control parameters are difficult to obtain due to industry confidentiality and intellectual property limitations, the existing white-box digital twin (DT) models of single-phase PWM rectifiers are difficult to apply in practical applications. Therefore, this article develops a gray-box DT model that can overcome the limitations of unknown control parameters and has higher accuracy. Firstly, the dead-time effect and the equivalent series resistance of capacitors are considered in the DT model to improve model accuracy. The operational characteristics of the DT model are closer to the actual system and have more comprehensive health monitoring capability. Then, based on the collected data from actual physical systems under dynamic and static operating conditions, the developed DT model is implemented on an independent server. The key parameters of the model are accurately monitored by using a hybrid algorithm of particle swarm and whale optimization. Finally, the proposed model and its parameter monitoring capability have been validated under different operating conditions, parameters, and control schemes, and the error impact has also been analyzed. This study contributes to the application of non-invasive health monitoring in practical power converters.

Reliability of the key components is crucial for power electronic systems. Digital twin technique that has received increasing attention, provides an innovative solution for high-efficient, noninvasive, and reliable condition monitoring of power converters. This article aims to introduce such an emerging technique to power converters for parameter monitoring, and first presents a digital-twin modeling method for three-phase three-level active-NPC (3L-ANPC) inverters. To this aim, a digital twin model including the key parameters of 3L-ANPC inverters is proposed. Second, according to the external behaviors of the actual physical circuit and the digital twin model, combined with the parameter degradation characteristics, the corresponding objective evaluation function is designed and constructed. Then, the improved particle swarm optimization algorithm is used to update and iterate the key parameters and optimize the model, so as to narrow the gap between the twin model and the actual circuit. The key parameters monitoring of 3L-ANPC inverters is achieved. Finally, a three-phase three-level ANPC inverter experimental platform is established, and the effectiveness of this proposed method and the accuracy of parameter monitoring are verified by experimental results.