Abstract:To accurately recommend power sales packages to users and help them select the package that best meets their needs, a deep interest evolution network（DIEN）algorithm based on deep learning is proposed. First, a comparison of several recommendation models is conducted to assess the performance of DIEN. Subsequently, an analysis of the structure of the interest evolution layer and the model’s hyperparameters is performed. Then, aiming at the“long tail effect”observed in the application of DIEN model in electricity market domain,two gating mechanisms are introduced between the interest extraction layer of the original model and the user vector. Finally, the feasibility of the proposed method is verified through a case analysis. Results show that proposed method can improve the electricity package adaptation rate of electricity users, enhance the market competitiveness of the power company and bidirectional benefits for both of the power company and users.

Abstract:Due to the tight coupling of electric and heating systems, combined heat and power dispatching has become a hot topic. By using the flexibility of heating systems in the aspects of sources, networks and loads, additional space for wind power penetration in the power systems has been provided. However, the available wind power output is difficult to forecast accurately, and its probability distribution cannot be obtained beforehand, while robust dispatch tends to be overly conservative. To address this issue, a data-driven adaptive robust dispatching method is proposed for integrated electric and heat systems. Combining the advantages of stochastic programming and robust optimization, the method simulates worst-case probability distribution scenarios using historical data to achieve a balance between economic efficiency and conservatism of dispatch strategy. The effectiveness of the proposed method is validated through simulation tests on a system consisting of a 6-bus electric network and a 6-node heating network.

Abstract:Under the background of new power system construction, line anomaly diagnosis is more significant for realizing evaluation of line health status and line loss management in rural power stations. In order to solve the problem of the lack of digital line anomaly diagnosis method in the current rural network station area, a line anomaly diagnosis method adaptive based on DBSCAN-PNN is proposed. Firstly, the calculation results of the virtual loop impedance of the abnormal user are obtained. Secondly, the k-nearest neighbor method is used to adaptively select DBSCAN parameters, and combined with the expert prior knowledge rules formed by various impedance anomalies, the sample data set of typical rural power station line anomalies is constructed. Further, the sample data set is divided into training set and test set according to a certain proportion, which is sent into the PNN classification model for training and testing, and the typical anomaly classification results are output. Finally, a case analysis is carried out based on four typical abnormal cases in a certain area, and the results showe that this method can realize the rapid and accurate identification of typical line anomalies diagnosis in the rural power grid low-voltage station area, and assisted in supporting lean operation and maintenance management of line loss.

Abstract:Accurate awareness of power system topology can enhance the assessment of weak links and facilitates development of load regulation strategies. To address the real-time acquisition challenge of dynamic changes in network topology, a topology identification method for distribution networks based on graph attention network（GAT）and multi-layer perceptron（MLP）is proposed. Firstly, the active distribution network is abstracted into a graph model, and the GAT adaptively learnes the relationships between different nodes. Additionally,multi-head attention is employed to calculate the fusion features of each node in the graph. Subsequently, the fused features of nodes and edge sets in the topology are fed into the MLP to learn the relationship between node features and the state of edge connections. The topological graph-level identification results are obtained by integrating all edge states within the network. Finally, the effectiveness of the proposed method is verified in IEEE 33-node and IEEE 123-node distribution networks. The robustness of the proposed method under different noise levels is analyzed. Simultaneously, the proposed model is compared with traditional machine learning and deep learning algorithms to determine its superiority.

Abstract:In view of the problem of real-time data acquisition errors in new energy plants, the data of new energy plants has mass and mutually coupled characteristics, a deep learning-based method for identifying and correcting bad data from new energy plants is proposed.Firstly, a LSTM identification model is constructed to identify the real-time bad data, and the bad data of the real-time identification is obtained. Secondly, the BP correction model optimized by the firefly algorithm is constructed to correct the bad data identified and obtain reliable data of the operation of the new energy station. The accuracy and effectiveness of the proposed method are verified by analyzing the real historical data of a typical wind farm.

Abstract:Building a new power system with new energy as the main body is an important means to achieve the goal of carbon peak and carbon neutrality. The new energy in new type power system will become the main power source, and the new energy with high penetration will profoundly change the form, characteristics and mechanism of the power system. A fusion method combining power flow equation and deep neural network is proposed to solve the topology and line parameter estimation method that best matches the measured value. By analyzing massive information data, the operation law of the power network is explored through data relations, which is used for fine topology identification and line parameter estimation of the distribution network. Firstly, the topology and line parameters are estimated by linear regression method, and the initial identification parameters are obtained, and the initial identification parameters are denoised. Then, feature screening is performed on the measured data based on the deep neural network, and the selected feature categories are one-to-one corresponding to the corresponding topology structure. Training data sets are constructed, and offline training is conducted, and the trained model is finally obtained, thus obtaining the accurate topology structure. Finally, the simulation results are carried out in IEEE 33-node distribution network, which proves the effectiveness and strong engineering practicability of the proposed method.

Abstract:To address the problem of low prediction accuracy due to the influence of multidimensional factors such as weather factors and calendar information on air conditioning load and the difficulty in sufficiently extracting the time series characteristics of load data, an air conditioning load prediction model for public buildings using long short-term memory（LSTM）recurrent neural networks based on an adaptive sliding window is proposed. The model first analyzes the influencing factors of air conditioning load in public buildings. Considering that traditional time series prediction models often perform poorly when dealing with non-stationary data, an adaptive sliding window mechanism is innovatively introduced. This mechanism can dynamically adjust the window size to better capture the variations in temperature and historical air conditioning load data, thereby improving the effectiveness of data preprocessing. Furthermore, given the complexity and long-term and short-term dependencies of air conditioning load variations, a multi-layer LSTM network architecture is designed to achieve accurate prediction of air conditioning load in public buildings. Taking the load data of a specific region as an example, proposed model achieves higher fitting ability and better prediction results when an appropriate sliding window size is selected.

Abstract:Under the“double carbon”goal, the power supply subject of the new power system will gradually transform from fossil energy to renewable energy, which greatly increases the pressure on supply and demand for power systems. Virtual power plants can gather massive,decentralized, and diverse distributed resources through advanced regulatory technology to form a subject of flexible regulation, which provides a feasible path for ensuring the safe and reliable operation of the new power system. Firstly, the concepts and types of virtual power plant is introduced. Secondly, key technologies of existing virtual power plants are analyzed from four aspects：resource aggregation, cooperative control, optimized scheduling and commercial operation. Then typical project cases of virtual power plants at home and abroad are summarized. Finally, challenges faced by virtual power plants under the new power system are pointed out, and the development direction of virtual power plants under the new power system is expected from three aspects：external specific quantitative analysis, low-carbon aggregation regulation, and new commercial operations.

Abstract:DC microgrids have the characteristics of low damping and weak inertia, making them vulnerable to external disturbances and difficult to maintain stable of DC bus voltage. To address this limitation, a hybrid energy storage coordination control strategy with inertial support for DC microgrids has been proposed. This strategy is based on the complementary advantages of hybrid energy storage and introduces virtual inertia control on the battery converter side. By establishing a functional expression between its control coefficient and DC voltage, it can provide a certain amount of inertial support to the system based on the size of voltage fluctuations, thereby improving the weak inertia of DC microgrids. Finally, a experimental platform based on RT-Lab is built to verify the effectiveness of the above control method. The results show that the strategy can achieve smooth switching between operating states of each micro-source unit and rapid stabilization of the DC bus voltage.

Abstract:Under the influence of domestic dual carbon targets, international green trade barriers, the demand for carbon reduction on the user side has sharply increased. As an important bridge connecting new power systems and electricity loads, microgrids can effectively balance local distributed energy output and promote green and low-carbon transformation on the user side, meeting their carbon reduction needs. The economic operation strategy of microgrids considering the demand for carbon reduction is mainly studied. Firstly, a microgrid operation model that includes wind and solar power, energy storage, electric vehicles, and flexible loads is constructed. Then carbon emission costs and system operation costs are comprehensively considered to construct an optimization model. Finally, based on the CPLEX solving tool, the results show that the proposed method can adapt to fluctuations in new energy output, changes in electricity prices, and changes in carbon prices, meet the carbon reduction needs of users within the microgrid, and significantly improve the low-carbon economic operation level of the system.

Abstract:In order to plan distribution system under the background of electric energy substitution, firstly considering the technical situation of various potential users such as electric oil substitution, electric coal substitution, electric gas substitution, a game model of distribution network expansion planning with multi-agent self-drive optimal decision-making is constructed, and a closed-loop decision-making chain of“expansion planning-market transaction-substitution decision”is formed. Then, the distribution system operator planning model considering operation control and the economic energy consumption model of users with multiple electric energy substitution potential are introduced respectively. A two-layer model of distribution network expansion planning including comprehensive electric energy substitution is formed, and the solution method of the model is proposed. Finally, an improved IEEE 33-node distribution system is selected for analysis. This method can reduce the total cost by about 30%, taking into account the flexibility to cope with load growth and multi-agent interaction. The results show that this method can effectively improve the operating margin to cope with the load growth and avoid the reduction of power reliability caused by the secondary programming. It is a new method of distribution network planning for the two-carbon target.

Abstract:With the rapid development of energy and information communication technology, the participation of novel demand resources in system operation and control has become an important part of grid operation. Due to the inherited differences and uncertainties of demand response, accurately quantifying the response performance of demand response resources has become the important tasks of grid operation. For this a demand response performance evaluation system is establisged from the perspective of power system operation and control, defines various indicators to assess the timeliness, accuracy, reliability and utilization of demand response and provides corresponding calculation methods. An example demand response case is used to demonstrate how to calculate the performance indicators quantitatively.Through comparison of the response performance indicators of three different responses, it is verified that the proposed quantitative evaluation indicators of response performance conform to the intuitive judgment of response performance and also provide more comprehensive evaluation information. The response performance indicators can accurately evaluate the response performance of novel demand response resources, and provide the basis for system operations, settlements and compliance.

Abstract:Under the background of“dual carbon”and new power systems, the installed capacity of new energy has increased year by year,new loads such as data centers and 5G base stations have continued to grow, and the user-side load patterns and load characteristics have undergone major changes, while superposing the influence of external environment and other factors, and the power supply and demand situation is grim. In order to ensure the safe and stable operation of power grid and tap the user’s adjustable potential, a dynamic evaluation method of multi-element customer load adjustable potential based on BP neural network based on particle swarm optimization is proposed.The influence mechanism of the adjustable potential of multi-customer load is analyzed, the feature label system is constructed, the relevant features of the adjustable potential are selected by the grey relational degree analysis method, and the adjustable potential assessment model is built to realize the assessment of the load control ability of multi-customer load. The accuracy of the assessment model is verified according to the actual response results, and the user load and demand response data are regularly updated. Dynamic evaluation of userside adjustability potential enables the model to adapt to changing user behavior.

Abstract:Under the strong guidance of the new energy vehicles to the countryside policy, sales of electric vehicles in rural areas have grown rapidly. However, rural power grid is widely distributed, power supply lines are long, and charging load is relatively dispersed and difficult to predict. To this end, a charging load prediction model for electric vehicles in rural areas based on modified graph temporal convolutional network（MGTCN）is proposed. Firstly, a rural power grid graph structure matrix is constructed based on graph convolutional neural network to characterize the spatial information of user charging characteristics and reduce the dimension of input data. Secondly, a temporal convolutional network is introduced to perceive the time series information of charging data and mine the time series features that affect load forecasting. Then, an MGTCN algorithm based on attention mechanism is proposed for charging demand forecasting. The attention mechanism assigns different weights to each feature, and the model can adaptively learn network parameters. Finally, the effectiveness of proposed method in predicting electric vehicle charging load in rural areas is verified based on the example results, and the impact of charging load on rural power grids under different electric vehicle penetration rates is further analyzed.

Abstract:Identifying power load curves is essential to ensure the safety and energy efficiency of the power grid. However, existing algorithms for power load curve identification tasks often suffer from issues such as low recognition accuracy and robustness. To tackle these issues, a multi-model fusion ensemble learning method for power grid load curve recognition is proposed. Temporal convolutional network （TCN）, transformer, and light GBM models are adaptively improved to predict load curve categories, considering three dimensions：local,global and structural features. Then, predictions through stacking ensemble learning（EL）to refine overall accuracy are adaptively fused.Additionally, a truncated Gaussian distribution（TGD）data augmentation strategy is introduced, which models intra-class signal fluctuations to alleviate data category imbalances, thereby enhancing the robustness of the recognition model. Through simulation analysis, compared with methods such as XG Boost, LSTM, and MLP, this approach shows a significant improvement in power load classification accuracy.

Abstract:Sustainable development of green energy is an important path for China to achieve the 3060 dual carbon target, but problems such as large fluctuations in new energy output and lagging subsidies have affected the development of new energy. Green certificate trading is an effective means to improve the revenue of new energy generation under the regulation of market mechanisms. Operation optimization strategy of virtual power plants under the green certificate trading mode is studied. First, on the purchasing side, power generation companies regard maximizing sales and green certificate profits as the goal, while virtual power plant operators regard minimizing purchase and green certificate costs as the goal, a master-slave game model is constructed between virtual power plants and power generation companies. Then, on the sales side, virtual power plant operators aim to maximize electricity sales and green certificate revenue, while users aim to minimize comprehensive costs, a dynamic game model between virtual power plants and users is constructed. Finally, CPLEX solver is used to solve the model. Examples verify that participating in green certificate trading can simultaneously expand the profits of power generation companies and virtual power plants, and better promote the cooperative relationship between power generation companies, virtual power plant operators, and users.

Abstract:With the growing development and advancement of the carbon and electricity markets, power producers should accurately capture their own carbon and electricity coupling characteristics. Based on this, bidding for power trading and carbon trading can observably enhance their returns. A model that considers the joint carbon and electricity bidding strategy of power producers in the medium and longterm trading scale is foucused on. A medium-and long-term carbon-electricity joint bidding model for thermal power producers is proposed.In addition, a rolling optimization method is proposed to cope with errors in medium-and long-term time-scale forecasts. The error between actual carbon emissions and carbon allowance margins is periodically revised during the bidding process. The case studies validate the effectiveness of the model and strategy, and the consideration of joint carbon-electricity optimization will substantially reduce the purchase cost of carbon allowances for the bidding power producers, and thus enhance corporate profits.

Abstract:With the continuous promotion of the construction of new power systems, the proportion of new energy installations is increasing, the demand for ramping capability is growing, and the importance of flexible ramping capability for maintaining the stable operation of power systems is gradually becoming apparent. At present, there is still much room for improvement in theoretical research and market construction of ramping auxiliary service market in China, while the United States has more mature experience in the construction of ramping auxiliary service market. It is necessary to study and analyze the determination of ramping demand price curve, which is regarded as a key and complicated part of the climbing market construction and operation. Firstly, the basic concepts related to the demand price curve of the U.S. ramping auxiliary service market are introduced. Based on this, the methods of determining the demand price curves of the ramping auxiliary service market in California and Midwest power markets are analyzed in detail, and their differences are compared and analyzed. Finally, taking into account the market construction situation and demand in China, feasible suggestions for constructing the ramping auxiliary service market are proposed in terms of market construction, demand calculation and demand price curve setting.