ABSTRACT. The number of power terminals is constantly increasing, and the amount of data to be processed in cloud computing centers is growing exponentially, which requires the computing capacity and data security of computing centers to be continuously improved. The emergence of edge computing solves this problem well. During the application process of edge computing in power systems, power terminals are designed to complete more calculation and storage work. Besides, data security of power terminals is becoming more and more important. That is because the power terminals are widely distributed in a complex and diverse environment. The communication channel is open and easy to destroy, causing serious information security threats. This paper collects global research on data encryption, privacy protection, auditing, and other aspects of data security through the research and analysis of security requirements, as well as research on terminal identity authentication, and eventually reviews various methods.

ABSTRACT. Electric double layer capacitor (EDLC) has seen widespread applications in energy storage for smooth operation of power systems. However, successful integration of EDLC into energy internet requires comprehensive understanding of its electrical response under various operation conditions. This work employs molecular dynamic simulations to study the physical mechanism behind the short-term memory effect reported in discharging process of EDLC. The physical meaning of constant phase element for EDLC is also revealed with equivalent circuit of transmission line model of resistors and capacitors.

ABSTRACT. With the development of the Energy Internet (EI), its applications have gradually spread from industrial uses to smart homes. Specifically, home Internet of Things(IoT) devices have become popular in the field of smart homes. In this paper, we propose an interpretable model that can be applied on the IoT devices. When Chinese characters are grouped into words, the meaning may vary. Inspired by the observation, we convert character-level Bi-directional Transformer (BERT) to word-level, which we call it multi-granular BERT (MLGB). It constructs the n-gram representation of different lengths within a model. It also learns the self-attention between n-grams during both pre-training and task-specific fine-tuning to learn both the word representation and word-word self-attention at the same time. As a diagnostic task, we evaluate our model on two Chinese text pair classification tasks and observe the model’s behavior. The MLGB retains the BERT’s accuracy on the tasks while demonstrates more interpretable word-level self-attention. Multi-granularity may also have served as a regularization of attention that alleviates the non-identifiability issue of self-attention.

ABSTRACT. The massive data produced in energy Internet (EI) faces the challenge involved by disturbance of noise, especially the measurement noise and noise attacks by hackers. Traditional noise estimation mainly focuses on the non-white additional or multiplicative noise estimation with definite wave types, and others, e.g., the phase noise in direction-of-arrival of antenna array, or the frequency bias in orthogonal frequency division multiplexing, etc., which usually use traditional estimation technologies. In this paper, a novel noise level estimation algorithm is proposed based on artificial neural network prediction. In an EI scenario, the proposed algorithm only needs to know the noise amplitude of historical data when training the model. At the time of execution, the algorithm estimates the noise level of existing data based on the latest noisy historical data, and the algorithm can be used for most noise types. Through numerical simulations, we found that its performance is apparently improved compared to the low passband filter method.

ABSTRACT. The uneven air gap between the inner cavity of the stator and rotor’s outer circle is one of the main vibration sources of large hydro-generator sets. Unbalanced magnetic tension between the stator and rotor forms a frequency-exciting disturbance force on the rotor and stator. In this paper, in order to explore the relationship between the uneven air gap and the magnetic field strength as well as electromagnetic stress, magnetic induction intensity of inner air-gap magnetic field of the generator and the stress on the rotor surface are analyzed for six working conditions of the hydroelectric generator, including centering and eccentricity of the rotor. The influence of the hydroelectric generator’s air gap change rotor pole stress is studied with finite element software based on the structure of the generator.

ABSTRACT. In the increasingly severe situation of network security, the blocking of external IP based on regional characteristics, which requires manpower to judge and operate, is becoming inadequate. Aiming at the practical problems existing in the network security defense of power enterprises, this paper proposed a risky IP identification algorithm based on hybrid genetic ensemble learning. The algorithm comprehensively used the improved genetic algorithm and the selective ensemble algorithm to establish a risky IP identification and prediction model. At the same time, A variety of network security information data were widely used to test the algorithm. The results show that the ensemble learning algorithm based on hybrid genetic can effectively identify risky IP and has higher recognition accuracy.

ABSTRACT. The construction of digital power grid is the key task of China Southern Power Grid Corporation. However, with the application of new technologies such as "cloud, big data and intelligence", the risks of network security are becoming diversified and complicated. In order to improve the defense and detection ability of advanced attacks, especially the crypted attacks, this paper proposes a detection technology of encrypted malicious traffic, through the use of deep neural network for feature learning and recognitionc, so as to improve the situation awareness ability in security of digital grid network.

ABSTRACT. This paper proposes a novel control parameters optimization method to enhance the dynamic performance of MMC-based DC distribution systems. The small signal model of DC distribution systems consisting of MMC main circuits, control systems, AC systems, and DC system is established. The participation factor analysis shows that the small signal stability of MMC-based DC distribution systems is strongly influenced by the control system. The sensitivity analysis-based control parameters optimization method is proposed to adjusting the control parameters for stability enhancement. The effectiveness of the proposed method is further validated by stability assessment and time-domain simulations on a three-terminal MMC-based DC distribution system.