Search

When faults such as blocking and commutation failure occur in a high-voltage direct current （HVDC） transmission system， they can cause transient voltage fluctuations near the sending-side converter station and lead to the disconnection with power grid. Firstly， the operating principles， response characteristics， and key parameter influences of the grid-forming energy storage open-loop control strategy are analyzed. Then， the reactive response characteristics of the synchronous condenser are analyzed， and a comparison is made between the grid-forming energy storage reactive power control model and the excitation regulation control model. Finally， a semi-physical simulation system is established， including EHVDC， photovoltaics， synchronous condenser， and grid-forming energy storage， to conduct comparative experiments on the effect of suppressing AC transient overvoltage under DC commutation failure. The results indicate that grid-forming energy storage and distributed synchronous condenser with the same capacity can achieve similar effects in terms of reactive transient response speed and the ability to suppress AC transient overvoltage.

Single-task recommendation algorithms have problems such as sparse data, cold start and unstable recommendation effect. Multi-task recommendation algorithms can jointly model multiple types of user behaviour data and additional information, to better explore the user’s interests and needs in order to improve the recommendation effect and user satisfaction, which provides a new way of thinking to solve a series of problems existing in single-task recommendation algorithms. Firstly, the development background and trend of multi-task recommendation algorithms are sorted out. Secondly, the implementation steps of the multi-task recommendation algorithm and the construction principle are introduced, and the advantages of multi-task learning with data enhancement, feature identification, feature complementation and regularization effect are elaborated. Then, the application of multi-task learning methods in recommendation algorithms with different sharing models is introduced, and the advantages and disadvantages of some classical models and the relationship between tasks are summarized. Then, the commonly used datasets and evaluation metrics for multi-task recommendation algorithms are introduced, and the differences and connections with other recommendation algorithms in terms of dataset evaluation metrics are elaborated. Finally, it is pointed out that multi-task learning has shortcomings such as negative migration, parameter optimization conflicts, poor interpretability, etc., and an outlook is given to the combination of multi-task recommendation algorithms with reinforcement learning, convex function optimization methods, and heterogeneous information networks.

Deep learning methods based on time series prediction are widely applied to solving fluid dynamics problems. However, the application characteristics of Long Short-Term Memory (LSTM) and Gate Recurrent Unit (GRU) in fluid mechanics are lacking in the existing research. Therefore, this paper proposes two flow field prediction models combining the Proper Orthogonal Decomposition (POD) method, referred to as POD-LSTM and POD-GRU. Their performance in flow field prediction is compared with numerical simulation data and experimental data respectively. The numerical simulation was obtained by using OpenFOAM-v9, and the experimental data were obtained from Particle Image Velocimetry (PIV). The results show that LSTM performs excellently in predicting velocity field, while GRU excels in predicting pressure field. On the other hand, under the same computational resources and cost, the training time required by GRU is shorter than that of LSTM. Furthermore, as the number of POD modes increases, the overall deviation of LSTM network predictions from actual flow field values tends to increase. In contrast, GRU networks show the opposite trend. This study provides a detailed description of the performance of POD-LSTM and POD-GRU in fluid mechanics. The proposed prediction models demonstrate strong generalization capabilities, further exploring the potential of deep learning methods in fluid mechanics.

The content of ethanol in wine can affect its quality. Some edible alcohol contains small amounts of methanol, and its accumulation can cause serious health problems. Therefore, it is very important to test the contents of methanol and ethanol in alcohol products. In this paper, a rapid method for the determination of methanol and ethanol by ambient corona discharge ionization mass spectrometry was developed. The method is based on the interaction of water radical cation with methanol and ethanol, respectively. In the positive ion mode, the water dimer radical cation (H2O)2+· (m/z 36) with ambient corona discharge ionization source was prepared by using ambient corona discharge ionization source, and it was introduced into the mass spectrometer with methanol or ethanol through a dual-channel low-energy corona discharge ionization device. The two reacted with each other to form the corresponding water radical cationic complex [CH3OH+H2O]+· (m/z 50) and [CH3CH2OH+H2O]+· (m/z 64), which realized the direct and rapid qualitative and quantitative analysis of methyl alcohol and ethanol in different kinds of alcoholic beverages. The result showed that the limits of detection and the limits of quantification of methanol and ethanol were 0.011 and 0.004 mg/L, 0.038 and 0.014 mg/L, respectively. The correlation coefficients of the target ionic strength curves obtained by the interaction of different concentrations of standard methanol and ethanol with water free radical cation could reach to 0.990 4 and 0.991 1, respectively. By using this method, no methanol was detected in common alcoholic drinks, and the ethanol content in liquor was higher than beer. In the actual work, the method has good practicability through the experimental exploration and verification of different wines. This method does not require sample pretreatment, has the advantages of simple operation and fast analysis, and is especially suitable for the detection of aqueous system. This study has important guiding significance for the quantitative detection and analysis of methanol and ethanol in complex system, provides technical support for the detection of alcoholic beverages and food, and also provides an effective strategy for the quantitative detection of the content of small and medium molecular alcohols in actual samples.

Microdroplets have many unique features such as small size and high electric field, therefore they can be accelerated to initiate chemical reactions and even promote some reactions that cannot be carried out in liquid phase. N, N-Dimethylaniline (DMA) is commonly used in medicine as a raw material for the synthesis of pharmaceutical intermediates. At present, the study involving in the detection of DMA is mainly focused on quantitative detection, and there is few reports on oxidation reaction and gas phase dissociation reaction of DMA. In this paper, the oxidation behavior and dissociation reaction of DMA and related compounds were studied by microdroplet spray mass spectrometry and tandem mass spectrometry. It was found that the ultra-high electric field spontaneously generated by microdroplets at the air-water interface can enable the oxidation reaction of DMA and related compounds to form abundant radical cation DMA+·. By changing the reaction distance, carrier gas pressure, injection velocity, sample concentration and substituent groups, the influence of experimental conditions and substituent modes on the generation of DMA+· was explored. The result showed that the presence of electron donor groups is benefit for the stability of radical cations. After the dissociation of DMA+·, the main product ions were generated by elimination of hydrogen radical and a successive elimination of HC≡N via methyl migration reaction. In addition, when the electron-donating CH3 group was connected with DMA, the intensity of related radical cation increased obviously, indicating that the electron-donating groups promoted the migration reaction. In conclusion, the stable transient intermediate DMA+· that is difficult to prepare by other means, was obtained by microdroplet reaction in this study. This method is helpful to deepen the understanding of the chemical promoted oxidation reaction characteristics of water droplets and enriches the rearrangement reaction types of aniline compounds in gas phase dissociation. It is expected that the prepared active species such as high reactivity radical cations will be applied to chemical industry or pharmaceutical fields in the future.

According to a small-step real-time simulation study on active neutral-point clamped （ANPC） three-level converters， existing electromagnetic transient modeling methods are inferior in model accuracy， computational efficiency， simulation scale， etc. To this end， an efficient electromagnetic transient simulation model is proposed. The model is built based on the associated discrete circuit （ADC） of the ANPC three-level bridge arm， and the internal nodes are eliminated by the WARD equivalence theory to construct an externally equivalent multi-port Norton equivalent circuit that replaces the bridge arm for solving the electromagnetic transient simulation of the network and achieving node reduction and simulation acceleration without spoiling modeling accuracy. A simulation system of the ANPC three-level converters based on this equivalent model is built in MATLAB/Simulink platform. Compared with the detailed model simulation results， the equivalent model can accurately simulate the steady-state and transient operating conditions and significantly accelerate the simulation.