2023, 47(15):133-141. DOI: 10.7500/AEPS20220224003
Abstract:With the increasing proportion of renewable energy in power systems, the energy transmission change between regions is becoming more and more violent. Therefore, it is necessary to study the power adjustment methods for transmission sections in large power grids. However, due to the limitations of the traditional algorithms, such as the problems of the non-convergence of power flow and relying on expert experiences, it cannot well overcome the difficulties of poor convergence when the adjustment target changes greatly. Therefore, a power flow adjustment method for transmission section combining artificial experiences and deep reinforcement learning is proposed. Firstly, the basic concepts of deep reinforcement learning are introduced, and the generator pre-selection and power compensation mechanism is proposed. Secondly, the reinforcement learning state, action space, reward function, and deep neural network framework are designed. The knowledge experience is introduced in the model training process, and the action space of the agents is effectively reduced. Finally, the effectiveness of the method is verified by practical cases of IEEE 39-bus system and Northeast China power grid.
2023, 47(15):111-121. DOI: 10.7500/AEPS20220317003
Abstract:The use of cross-provincial transmission channels for wide-area optimal configuration of renewable energy resources is an important means to achieve the low-carbon transition of interconnected power grids. However, the traditional "provincial dispatch centre application through telephone, regional dispatch centre manual modification" inter-network real-time regulation mode greatly restricts the flexible regulation ability of interconnected power grids. To this end, an engineering practical real-time optimal dispatching method for renewable energy accommodation across provinces and regions has been proposed, and corresponding system development and deployment have been completed. This system effectively solves the problems faced by the cross-provincial accommodation of renewable energy in traditional regulation modes, such as delayed perception of supply and demand, incomplete optimization of channel allocation, and difficulty in verifying power flow sections, by calculating channel transmission margin, sending and receiving power matching revising channel plans, and verifying power flow safety. The analysis of the system operation results during the trial operation period indicates that the computational efficiency of the system meets the practical engineering application requirements and can promote the wide and deep utilization of renewable energy resources in different typical scenarios.
2023, 47(15):142-150. DOI: 10.7500/AEPS20220411001
Abstract:Wind power can provide the reserve by de-loading operation. Due to the uncertainty of the actual output of wind power, the reserve provided by wind power may deviate from the predetermined value, so the wind power reserve is not completely reliable. Therefore, a reserve optimization model for power systems considering the reliability of wind power reserve is proposed. Firstly, based on the probability distribution information of wind power output, the variation of wind power reserve is analyzed. Then, the wind power reserve is included in the calculation of curtailed wind costs and load loss costs. The proposed model is transformed into a mixed-integer linear programming problem by further eliminating the integral sign and linearization. Finally, the effectiveness of the proposed model is illustrated based on the IEEE-RTS 24-bus system case. The results show that considering the reliability of wind power reserve, the curtailed wind costs and load loss costs in the system can be evaluated more precisely, so that the reserve in the system can be configured more precisely, which can avoid overly conservative or radical results.
2023, 47(15):67-79. DOI: 10.7500/AEPS20220430001
Abstract:As flexible load resources, electric vehicles and 5th generation (5G) base stations provide abundant reserve capacity for the power grid, which can be regulated by certain strategies and participate in the auxiliary services of the power grid. However, in the case of the participation of massive distributed flexible resources in the future, the centralized optimization dispatching method will have the shortcomings of privacy leakage and heavy computing burden. On this basis, a real-time evaluation model of 5G base station backup battery and electric vehicle dispatchable capacity is first established. Then, an undirected graph network is used to model the communication system and energy storage components. And a dispatching strategy based on balanced state of charge is proposed. The privacy protection and dispatching control of massive resources are realized by adopting the distributed consensus algorithm. Furthermore, a two-stage real-time dispatching process scheme is proposed, which first evaluates the power boundary and then distributes consensus dispatching. Finally, the effectiveness of the scheme is verified by simulation.
2023, 47(15):80-90. DOI: 10.7500/AEPS20220707011
Abstract:The increasing penetration rate of renewable energy in the power grid causes a serious lack of flexibility in the power system. The existing methods for dealing with the uncertainty of supply and demand in power system flexibility have the problem of overly conservative or risky. To address the problem, a data-driven distributionally robust optimal dispatch model is proposed in this paper. Firstly, considering the spatial and temporal correlations of wind and photovoltaic (PV) power, the output power set of wind and PV is constructed based on Copula theory. The flexibility demand of the power system is quantified by combining the scenario method and the interval method. The flexibility regulation factor is introduced to characterize the capability of various resources to participate in flexibility regulation, and the flexibility supply and demand balance constraint is established. Secondly, considering the flexibility supply capability of demand-side resources such as electric vehicles, with the objective of minimizing the operation cost of flexible resources and the penalty cost of grid flexibility deficiency, a data-driven two-stage distributionally robust model is established. To reduce the conservativeness, the comprehensive norms are included to constrain the probability distribution, which can also reduce the appearance probability of extreme cases of flexibility demand. For the solution of the two-stage robust model, the theory of zero-sum game is used in the paper to decouple the model into a master problem and a subproblem, which are iteratively solved by using the column-and-constraint generation algorithm. Finally, the case verifies that the proposed model has a positive effect on improving the flexibility adequacy and the economy of the power system compared with the traditional uncertainty model.
2023, 47(15):91-99. DOI: 10.7500/AEPS20220808005
Abstract:Aiming at the problems of insufficient power balance capacity and lack of regulation flexibility in the power grid with high proportion of renewable energy, a flexible economic optimal dispatch strategy of hydrogen-oxygen dual-cycle for comprehensive regulation of renewable energy and thermal power is proposed. First, the mechanism of hydrogen-oxygen dual-cycle improving regulation capacity of power system operation is introduced, and the model for the flexible regulation characteristics of water electrolytic devices, oxygen-enriched deep-regulation units and gas-fired generators are established. Then, considering the energy storage and flexible load, an flexible economic optimal dispatch model of hydrogen-oxygen dual-cycle for comprehensive regulation of renewable energy and thermal power is established. Through the quantification of load and renewable energy abandonment cost and random fluctuation risk, the comprehensive cost objective function is constructed considering economy and flexibility. Finally, the effectiveness of the hydrogen-oxygen dual-cycle mechanism of improving the system operation and regulation is verified with the modified IEEE 39-bus system.
2023, 47(15):122-132. DOI: 10.7500/AEPS20220822006
Abstract:The extensive integration of renewable energy and flexible resources has increased the complexity of power system dispatching. In order to meet the non-stationary fluctuation of flexibility demand in the system, based on the differentiation of flexible resource response characteristics, this paper proposes a day-ahead adaptive time-scale dispatching strategy for the power system considering the flexibility supply-demand amplitude-frequency matching. Firstly, based on the interval number theory, the uncertainty of the error rate in predicting the power of renewable energy in the future is described, and the uncertainty scenario of flexibility demand is obtained. Secondly, the mean variational mode decomposition technique is used to decompose and analyze the sequence of flexibility requirements, and a physical model of the response of flexible resources is established. Then, the fluctuation period attribute of the decomposition submode and the response period attribute of the flexible resource are used to construct a flexibility supply-demand amplitude-frequency matching framework, and an optimal adaptive time-scale dispatching model is established based on the matching results. Finally, an analysis is conducted on the optimized IEEE 39-bus system and the actual power grid system to verify the rationality of the proposed method.
2023, 47(15):55-66. DOI: 10.7500/AEPS20220829005
Abstract:The boundaries of dispatchable resources are expanded when the multi-energy system participates in power dispatch, which helps improve energy utilization efficiency. To deal with the difficulty in collaborative optimal calculation caused by the nonlinear characteristics of the multi-energy system model, and the difficulty in order issuing caused by the privacy protection and participation threshold, a depiction and aggregation method of replacement constraints for the flexible regulation capability is studied. First, aiming at the difficulty in feasibility assurance caused by the non-convex characteristics of a single energy subsystem, the inscribed convex approximation depiction method of the flexible regulation capability based on the boundary inward modification is proposed. Then, aiming at the difficulty in considering the juggling aggregation calculation accuracy and efficiency of flexibility regulation capability for multi-energy system clusters, the accurate and efficient aggregation method of flexible regulation capability based on generalized zonotopes is proposed. For the polynomial-zonotope-type generalized zonotopes, the transformation and inverse transformation strategy of flexible regulation capability is proposed. Finally, the simulation results show that the proposed method ensures the feasibility of alternative constraints and greatly improves the calculation accuracy on the premise of meeting the requirements of operation dispatch time window.
2023, 47(15):100-110. DOI: 10.7500/AEPS20220911003
Abstract:The integrated energy system including electric vehicles (EVs) and hydrogen fuel cell vehicles (FCVs) can achieve multi-energy coordination and complementarity, and effectively promote the realization of the "carbon emission peak and carbon neutrality" goal, but temporary changes in vehicle travel plans have adverse effects on the system. Therefore, a multi-layer coordinated optimal strategy for the electric-thermal-hydrogen integrated energy system including the load feedback correction of the vehicle to grid (V2G) is proposed. Based on the coordination and cooperation of three-layer optimization, the charging and discharging management layer of EVs comprehensively considers the variance of the load curve and the dissatisfaction of EV owners, formulates charging and discharging plans and transmits them to the rolling optimization layer of model predictive control. In the rolling optimization process, the number of unplanned vehicles is identified in real time, and the V2G load feedback correction layer is operated independently, so as to eliminate the adverse effects of temporary changes in vehicle travel plans. The simulation results show that the proposed strategy not only achieves a win-win situation between car owners and operators, but also improves operational economy of the system and improves the track results of the planned power flow of the main grid tie-line line.
2023, 47(15):151-161. DOI: 10.7500/AEPS20221101006
Abstract:Wind power scenario generation is an important basis for power grid dispatching and security checking under uncertain conditions. In this paper, a wind power output sequence scenario generation method is proposed. Firstly, based on the average monthly power output, the annual wind power series are divided into several quarters with large differences in distribution characteristics. Secondly, based on the daily characteristic indices of wind power output, the typical-day cluster analysis is carried out based on the annual wind power output sequence, and the joint distribution function of day characteristic indices for each day type is constructed. Thirdly, the Markov chain is used to describe the characteristics of day type transition in each quarter, and based on this, the types and daily characteristics of each day during the future period are obtained by sampling. Furthermore, the 96-point daily wind power output sequence of corresponding days during the future period is generated using the optimization technology. A new method for generating wind power output sequence scenarios according to the specified extreme day types, date and lengths is further proposed. Finally, the effectiveness and feasibility of the proposed method are verified by case analysis.
2023, 47(15):12-35. DOI: 10.7500/AEPS20221128005
Abstract:With the promotion of the strategic goal of “carbon emission peak and carbon neutrality” and the gradual development in constructing new power systems, the renewable energy represented by wind power and photovoltaic keeps developing rapidly. The strong uncertainty and randomness of the large-scale renewable energy pose a huge challenge to the safe and stable economic operation of the power system. First, the situation and challenges faced by active power and frequency control in the power system with high proportion of renewable energy in terms of real-time balancing, frequency regulation resources, market development, and security defense are analyzed in depth. Then, the core issues and research progress in five aspects consisting of multiple control-area coordination control, orderly deployment of multiple types of resources, coordination between economic dispatch and automatic generation control, security event detection and defense, and intelligence and new technologies are introduced. Finally, the key research directions such as active power and frequency control of multiple control agents, collaborative control of massive ubiquitous resources, closed-loop control of auxiliary service markets, improvement of risk defense capabilities, and deepening application of artificial intelligence technology are discussed and prospected.
2023, 47(15):162-169. DOI: 10.7500/AEPS20221130009
Abstract:At present, frequency regulation of wind-farm-level grid connection faces significant challenges. In order to reduce modeling complexity, using actual operation data of grid connection points of the wind farm, a time-domain differential dynamic modeling method for primary frequency regulation characteristics under all operation conditions within the output range of wind farms is proposed. Firstly, the mechanism is analyzed and the input and output variables are determined to complete data sampling and processing. Then, discrete condition sampling and nonlinear evaluation methods are proposed, and a subspace model under single operation condition is established to evaluate the dynamic nonlinearity of frequency regulation response under all operation conditions through gap measurement. Thereafter, in order to efficiently approximate the complex nonlinear frequency regulation characteristics under all operation conditions, a hybrid finite difference regression vector is constructed, and its stretched differential dynamic space is convexly divided. Modeling samples are evenly selected in each sub operation domain, and long and short-term memory neural network with attention mechanism is used to model under all operation conditions. Finally,through the verification of the frequency regulation operation data of grid connection points of the wind farm, it shows that the proposed finite difference dynamic modeling method under all operation conditions has the approximation ability of arbitrary precision, and can effectively characterize the complex nonlinear response characteristics of primary frequency regulation for the wind farm.
2023, 47(15):36-45. DOI: 10.7500/AEPS20221130016
Abstract:The rapid increase in the proportion of renewable energy in new power systems will lead to a serious lack of regulation capacity on the power generation side. The demand response based on customer directrix load (CDL) can normalize the motivation of load shaping, thereby improving the balance capability of the power grid and promoting the accommodation of renewable energy. A network-load cooperative game mechanism based on Nash bargaining is proposed to address the issue of profit allocation in CDL-based demand response. To address the intractability of the Nash bargaining, its approximation is proposed and established, and its distributed solution method is proposed to ensure the information privacy of each entity. Taking the electric vehicle cluster as an example, a flexibility probability prediction model for adjustable load is proposed to reflect its full-time shaping capability. Finally, a regional power grid case that includes flexible resources such as electric vehicles and heat pump water heaters as well as high proportion of renewable energy is established. The simulation results show that both types of loads have good shaping capability. Based on the proposed profit allocation mechanism, the CDL-based demand response can significantly enhance the incentives of aggregators, effectively promote renewable energy accommodation, and achieve a win-win situation for the power grid and the users
2023, 47(15):46-54. DOI: 10.7500/AEPS20221209004
Abstract:To construct a knowledge graph of renewable energy accommodation, firstly, the accumulated massive dispatching operation data of the power grid in the form of dynamic quaternions explicitly expresses the spatio-temporal correlations of dispatching operation data. The local spatio-temporal graph is quickly searched and extracted by sliding time windows to construct sub-graph data sets. Then, the spatio-temporal synchronous graph convolutional network extracts high-dimensional features from the local spatio-temporal graphs to fully excavate the spatio-temporal correlations of the historical data. The model is guided by the mechanism knowledge stored in the knowledge graph of renewable energy accommodation and trained in parallel by multiple subgraphs to improve the learning efficiency. Finally, simulation and experimental validation are conducted based on a provincial grid case in Northwest China. The results show that the proposed method can effectively avoid complicated mathematical modeling and solving, and has higher evaluation accuracy and speed compared with traditional methods.
2023, 47(15):3-11. DOI: 10.7500/AEPS20230609001
Abstract:The large-scale integration of renewable energy has brought great technical challenges to the operation of power systems. The traditional deterministic dispatch modes ignore the quantification management of safety risks under uncertainties. While, the robust optimal dispatch is over-conservative. The probabilistic dispatch and control technique based on the stochastic optimization has the characteristics of risk controllability and statistical optimality. However, there are problems such as unclear concepts and connotations, difficulties in probabilistic modeling, and low efficiency in solving the complex chance constrained models. The practical application of the probabilistic dispatch technique faces difficulties. Therefore, the definition and technical connotation of risk-quantified probabilistic dispatch are proposed, and the followed basic principles in dealing with the uncertainty of the renewable energy are illustrated. Then the key techniques of risk-quantified probabilistic dispatch are stressed, which include: 1) mathematical basis for probabilistic modeling of the renewable energy; 2) probabilistic reserve decision and unit commitment; 3) probabilistic day-ahead dispatch considering renewable energy curtailment and orderly power utility; 4) stochastic robust adaptive real-time dispatch. Finally, the application effects of the proposed method in provincial power grids are briefly analyzed, and the future research topics are prospected.