Search

High temperature superconducting (HTS) synchronous condenser is a novel large-capacity reactive power compensation device. Its reactive power compensation capacity and accuracy are mainly determined by the excitation system, making the excitation power supply a crucial part of the HTS synchronous condenser system. This paper presents a dual power supply excitation system for HTS synchronous condenser. The principle and control strategy of its topology are introduced, and simulation analysis is carried out. The interleaved parallel control strategy of the excitation system based on the state-space averaging method is investigated, and superconducting experiments on low-voltage power supply are conducted. Experimental results show that this control strategy can significantly reduce the exciting current ripple. For example, the current ripple is reduced to about 0.008%. When the HTS synchronous condenser operates in constant excitation mode and forced excitation mode respectively, the required voltage and current for the excitation winding differ in magnitude and accuracy. The proposed excitation system not only solves this problem but also improves the power supply utilization rate. It provides stable and high-precision excitation current, enabling the HTS synchronous condenser to accurately compensate reactive power and enhance the stability of the power grid.

Temperature superconducting (HTS) synchronous condenser is a reactive power compensation device with the advantages of low loss, high efficiency and wide reactive power range. Therefore, it has a very good application prospect in power grid. To solve the problems of low power factor and low conduction angle of constant excitation function for a long time due to single power supply excitation system, this paper proposes a high-low voltage switching excitation system, which separates the forced excitation and constant excitation modes of the excitation system into two power supplies. Based on the introduction of principle of excitation system, the design of its parameters and simulation analysis are completed, and a set of 400kW device is designed. On this basis, a series of corresponding superconducting experiment are carried out. The experimental results verify the principle of the power supply topology, and also show the feasibility and reliability of the proposed excitation system.

In software engineering, deep learning models are increasingly deployed for critical tasks such as bug detection and code review. However, overfitting remains a challenge that affects the quality, reliability, and trustworthiness of software systems that utilize deep learning models. Overfitting can be (1) prevented (e.g., using dropout or early stopping) or (2) detected in a trained model (e.g., using correlation-based approaches). Both overfitting detection and prevention approaches that are currently used have constraints (e.g., requiring modification of the model structure, and high computing resources). In this paper, we propose a simple, yet powerful approach that can both detect and prevent overfitting based on the training history (i.e., validation losses). Our approach first trains a time series classifier on training histories of overfit models. This classifier is then used to detect if a trained model is overfit. In addition, our trained classifier can be used to prevent overfitting by identifying the optimal point to stop a model’s training. We evaluate our approach on its ability to identify and prevent overfitting in real-world samples. We compare our approach against correlation-based detection approaches and the most commonly used prevention approach (i.e., early stopping). Our approach achieves an F1 score of 0.91 which is at least 5% higher than the current best-performing non-intrusive overfitting detection approach. Furthermore, our approach can stop training to avoid overfitting at least 32% of the times earlier than early stopping and has the same or a better rate of returning the best model.