A method for evaluating flexibility and economic dispatch of virtual power plant considering wind and solar uncertainty

Abstract

The present application relates to a kind of virtual power plant flexibility evaluation and economic scheduling method considering wind light uncertainty, comprising: establishing the linearization network model of distribution network, to characterize the linear mapping relationship of node injection power and node voltage amplitude, branch current and balanced node exchange power;Establish the equipment operation constraint of renewable energy source unit and adjustable unit;Based on historical data, uncertainty is handled using distribution robust optimization method, and uncertainty operation constraint is constructed;Based on linearization network model, equipment operation constraint and uncertainty operation constraint, determine the flexibility adjustable domain of virtual power plant at point of common coupling;With the minimum operation cost as the goal, with flexibility adjustable domain, equipment operation constraint and linearization network model as constraint, generate optimal scheduling instruction.The present application solves the problem that scheduling plan reliability is poor, economy and security are difficult to consider due to the coupling complexity of renewable energy output uncertainty and grid physical constraint.

Description

Technical Field

[0001] This invention relates to the field of power system operation and dispatching technology, and in particular to a method for assessing the flexibility and economic dispatching of virtual power plants that takes into account the uncertainties of wind and solar power. Background Technology

[0002] With the continuous advancement of new power system construction, the penetration rate of renewable energy sources such as wind power and photovoltaics in the power system continues to increase. While the large-scale integration of these clean energy sources improves energy utilization, their inherent randomness, volatility, and intermittency also pose significant challenges to the safe and stable operation and economically optimized dispatch of the power system. Against the backdrop of increasingly significant fluctuations on both the source and load sides and the widespread integration of distributed resources, the traditional single-source dispatch mode is no longer sufficient to meet the actual needs of modern distribution systems for flexible adjustment and multi-resource coordinated operation.

[0003] Virtual Power Plants (VPPs), as an advanced operational organization method, aggregate dispersed resources such as distributed renewable energy, distributed generators, energy storage devices, and adjustable loads through information communication and coordinated control technologies. This allows for unified modeling, optimization, and scheduling, thereby improving the observability, controllability, and dispatchability of distributed resources. Participating in grid dispatch and market transactions as a whole, VPPs are of significant value in enhancing renewable energy absorption capacity, improving system flexibility, and enhancing operational economics.

[0004] However, existing virtual power plant dispatching methods have significant shortcomings. First, most of these methods focus on economic optimization, aiming to minimize electricity purchase and sale costs or maximize operational benefits, while neglecting the overall flexible adjustment capabilities of the various resources within the virtual power plant. Especially when wind and solar power output is subject to significant uncertainty, relying solely on deterministic predictions for dispatching can easily lead to insufficient reserve capacity, imbalances in regulation, or serious deviations from the plan in actual operation, thereby compromising the safety, economy, and external responsiveness of the virtual power plant.

[0005] Secondly, the flexibility of a virtual power plant is not simply the sum of the adjustment capabilities of its individual devices, but rather a complex interplay of factors including the upper and lower limits of distributed unit output, ramp-up capabilities, energy storage state of charge, network power flow constraints, and node voltage constraints. Currently, research on the flexibility of virtual power plants largely focuses on qualitative analysis or evaluation of local indicators, lacking quantitative assessment methods that comprehensively consider the uncertainties of wind and solar power, network operational constraints, and the coupling relationships among multiple resources.

[0006] Crucially, how to effectively integrate the results of flexibility assessment with the economic scheduling model during actual scheduling to achieve collaborative optimization of various distributed resources remains a core technical problem that urgently needs to be solved in this field.

[0007] Therefore, there is an urgent need in this field for a new method that can comprehensively consider the uncertainties of wind and solar power, accurately quantify and assess the flexible adjustment capabilities of virtual power plants, and use the assessment results for economic dispatch, in order to address the shortcomings of existing technologies and improve the ability of virtual power plants to cope with new energy fluctuations and their safe operation.

[0008] It should be noted that the information disclosed in the background section above is only for understanding the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0009] The technical problem addressed in this application is "how to solve the problem of poor reliability of virtual power plant dispatching plans and difficulty in balancing economy and security due to the uncertainty of output from renewable energy sources such as wind power and photovoltaics and the complexity of the coupling of physical constraints of the power grid".

[0010] The technical solution adopted in this application to solve the above-mentioned technical problems is as follows.

[0011] This application provides a method for assessing the flexibility and economic dispatch of a virtual power plant considering the uncertainties of wind and solar power. The virtual power plant is connected to the distribution network and includes distributed energy resources. The distributed energy resources include renewable energy units with uncertain output and adjustable units with adjustable output. The renewable energy units include distributed wind turbines and distributed photovoltaics. The method includes the following steps: A linearized network model of the distribution network is established. The linearized network model is used to characterize the linear mapping relationship between node injected power and node voltage magnitude, branch current and slack node exchange power. Establish equipment operation constraints for renewable energy units and adjustable units; Based on historical data, the partial Brussels bar optimization method is used to handle the uncertainty of renewable energy unit output and construct corresponding uncertainty operation constraints. Based on the linear network model, equipment operation constraints, and uncertain operation constraints, the flexible adjustable domain of the virtual power plant at the point of common coupling is determined. The flexible adjustable domain is used to characterize the power regulation capability that the virtual power plant can provide within the scheduling cycle. With the goal of minimizing operating costs, and constrained by the adjustable flexibility domain, equipment operation constraints, and linearized network model, an optimization solution is performed to generate the optimal scheduling instructions for each unit in the virtual power plant.

[0012] In some embodiments, the linearized network model is a single-phase linear power flow model.

[0013] In some embodiments, the step of establishing a linearized network model of the distribution network includes: establishing a linear mapping relationship between node injected power and node voltage amplitude; establishing a linear mapping relationship between node injected power and branch current; and establishing a linear mapping relationship between node injected power and balancing node exchange power.

[0014] In some embodiments, the adjustable unit includes a micro gas turbine and an energy storage system.

[0015] In some embodiments, the steps for establishing equipment operation constraints include: establishing upper and lower limits for output and ramp rate constraints for distributed wind turbines and distributed photovoltaics; establishing upper and lower limits for output and ramp rate constraints for micro gas turbines; and establishing charging and discharging power constraints, energy state constraints, and charging and discharging mutual exclusion constraints for energy storage systems.

[0016] In some embodiments, the steps of using the sub-Bruker optimization method to handle the uncertainty of renewable energy unit output include: obtaining the empirical mean vector and covariance matrix of the prediction error based on historical sample data; constructing a distributed fuzzy set constrained by the empirical mean vector and covariance matrix; and transforming the sub-Bruker chance constraint into a deterministic linear safety margin constraint based on the one-sided Chebyshev-Cantelli inequality.

[0017] In some embodiments, the step of determining the flexible adjustable domain includes: aggregating the resources of the virtual power plant into virtual generators and virtual batteries; and using a boundary shrinking algorithm to iteratively solve the aggregation parameters of the virtual generators and virtual batteries to obtain the boundary of the flexible adjustable domain.

[0018] In some embodiments, the iterative solution steps using the boundary shrinkage algorithm include: solving a set of linear programming problems to obtain initial aggregate boundary parameters for the virtual generator and virtual battery; searching for infeasible points within the current aggregate boundary; if infeasible points exist, searching for the nearest feasible boundary point in the real feasible domain; updating the boundary parameters based on the nearest feasible boundary point to shrink the current aggregate boundary; iteratively executing the steps of searching for infeasible points and updating boundary parameters until there are no infeasible points within the current aggregate boundary, thus obtaining the final flexible adjustable domain boundary parameters.

[0019] In some embodiments, operating costs include: micro gas turbine power generation costs, renewable energy curtailment penalty costs, and energy storage system operating costs.

[0020] In some embodiments, a virtual power plant economic dispatch system considering the uncertainty of wind and solar power is also provided, applied to a virtual power plant connected to a distribution network. The virtual power plant includes distributed energy resources, which include renewable energy units with uncertain output and adjustable units with adjustable output. The renewable energy units include distributed wind turbines and distributed photovoltaics. The system includes: The linearized network modeling unit is configured to establish a linearized network model of the distribution network. The linearized network model is used to characterize the linear mapping relationship between node injected power and node voltage magnitude, branch current and slack node exchange power. The equipment constraint modeling unit is configured to establish equipment operating constraints for renewable energy units and adjustable units; The uncertainty processing unit is configured to handle the uncertainty of renewable energy unit output based on historical data and using the sub-Bruker optimization method to construct corresponding uncertainty operation constraints. The flexibility assessment unit is configured to determine the flexible adjustable domain of the virtual power plant at the point of common coupling based on the linearized network model, equipment operation constraints, and uncertain operation constraints. The flexible adjustable domain is used to characterize the power regulation capability that the virtual power plant can provide within the dispatch cycle. The economic dispatch unit is configured to minimize operating costs, and to perform optimization solutions based on the adjustable flexibility domain, equipment operation constraints, and linearized network model, generating optimal dispatch instructions for each unit within the virtual power plant.

[0021] The present invention has the following beneficial effects: This invention establishes a linearized network model of the distribution network, which characterizes the linear mapping relationship between node injected power and node voltage amplitude, branch current, and slack node exchange power. This transforms complex, nonlinear power grid physical constraints (node ​​voltage, branch current) into linear mathematical relationships that can be embedded in the optimization model. By using linear approximations, the coupling relationship between the power grid operating state and equipment injected power is explicitly expressed. Power grid safety operation constraints (such as voltage exceeding limits and line overload) are transformed from difficult-to-handle implicit conditions into explicit, computable linear inequality constraints. This lays the computational foundation for subsequent optimization solutions, significantly reducing model complexity and solution difficulty, and ensuring that the scheduling scheme meets power grid safety requirements. Requirements: By "establishing equipment operation constraints for renewable energy units and adjustable units," the physical operating boundaries of all distributed energy units are defined, ensuring that the optimized scheduling instructions do not exceed the actual capacity range of any renewable energy unit or adjustable unit (such as maximum / minimum output, maximum ramp rate, energy storage capacity, etc.), thus guaranteeing the equipment feasibility of the scheduling scheme; By "using the distributed bar optimization method to handle the uncertainty of renewable energy unit output based on historical data and constructing corresponding uncertainty operation constraints," a distributed fuzzy set is constructed using the first-order and second-order statistical information (mean, covariance) of historical data, rather than assuming a specific probability distribution, and optimization theory is used to transform it into a deterministic constraint. The system can robustly handle renewable energy output forecasting errors, and the generated operating constraints are robust to potential "mismatches" in the true distribution of uncertainties, thus avoiding dispatch risks caused by overly optimistic or pessimistic forecasts and significantly improving the reliability of the dispatch scheme. By "determining the flexible adjustable domain of the virtual power plant at the point of common connection based on a linearized network model, equipment operating constraints, and uncertain operating constraints, the flexible adjustable domain is used to characterize the power regulation capability that the virtual power plant can provide within the dispatch cycle," an approximate quantification of the overall regulation capability of the virtual power plant is achieved. After comprehensively considering all internal constraints (equipment, grid) and external uncertainties, the external interface (point of common connection) of the virtual power plant is solved using an aggregation method. By defining the feasible power domain of a point, a reliable and quantifiable flexibility boundary (both upward and downward adjustment capabilities) is obtained. This "flexible adjustable domain" provides a clear and safe optimization space for economic dispatch. By "optimizing the solution with the goal of minimizing operating costs, and with constraints of the flexible adjustable domain, equipment operation constraints, and linearized network model, the optimal dispatch instructions for each unit in the virtual power plant are generated." Within the safe, reliable, and feasible "feasible domain" constructed by the aforementioned steps, mathematical optimization is performed with the goal of minimizing costs. The resulting "optimal dispatch instructions" are the result of pursuing economic optimization while satisfying all safety and reliability constraints, thus achieving a unity of economy, safety, and reliability.

[0022] Ultimately, the combined effect of all the aforementioned technical features is to provide a novel method for assessing the flexibility and economic dispatch of virtual power plants. This method, through the organic integration of linearized network modeling, equipment constraint modeling, debulking uncertainty handling, and flexible adjustable domain assessment, can generate an optimal dispatch command that is economically optimal, absolutely safe and reliable, and directly executable. This fundamentally solves the industry problem of poor dispatch plan reliability and difficulty in balancing economy and safety caused by the uncertainty of renewable energy output (especially wind and solar uncertainty) and the coupling complexity of power grid physical constraints.

[0023] Other beneficial effects of the present invention will be further described below. Attached Figure Description

[0024] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a technical framework diagram of the method in the embodiment; Figure 2 This is an example of the IEEE 33-node network topology after the integration of distributed energy resources. Figure 3 Example: Predicted output curve of distributed wind turbine; Figure 4 The example shows the predicted output curve of a distributed photovoltaic system.

[0025] Figure 5 The implementation example considers the flexibility boundary of a virtual power plant under conditions of wind and solar uncertainties; Figure 6 A comparison diagram of the flexibility boundaries of virtual power plants considering wind and solar uncertainties in the examples; Figure 7 The virtual generator ramp rate boundary is used as an example. Figure 8 This is a virtual battery energy boundary for an example. Figure 9 The virtual power plant dispatch curve and its flexibility upper and lower bounds are shown in the example. Figure 10 Examples are provided; results of distributed energy scheduling under four scheduling scenarios. Detailed Implementation

[0026] The embodiments of the present invention will be described in detail below. It should be emphasized that the following description is merely exemplary and not intended to limit the scope and application of the present invention.

[0027] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of the present invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0028] This invention recognizes that existing virtual power plant scheduling methods suffer from poor scheduling plan reliability and difficulty in balancing economy and security. The root cause lies in their failure to recognize that the aggregate flexibility of a virtual power plant can be mathematically quantified as a low-dimensional projection of its high-dimensional feasible operating domain onto the point of common coupling (PCC). This feasible operating domain is strongly coupled with output uncertainty, equipment operating constraints, and complex power grid physical constraints. Therefore, this invention, through a profound understanding of robust optimization separation and dimensionality reduction of this high-dimensional feasible operating domain, proposes an "evaluation-then-optimization" scheme: first, it employs sub-bulbar optimization to handle uncertainties and establishes a linearized network model and equipment constraints; then, it uses robust optimization to separate the network constraints of different distributed energy sources, solving for a conservative, flexible, adjustable domain characterizing the overall regulation capability of the virtual power plant (i.e., a reliable projection of the feasible operating domain); finally, it performs economic optimization within this safe and reliable domain to resolve scheduling objective conflicts caused by uncertainty and constraint complexity.

[0029] In some embodiments, this invention proposes a method for assessing the flexibility and economic dispatch of virtual power plants considering the uncertainties of wind and solar power. This method targets virtual power plants (VPPs) that include distributed energy resources (DERs) such as wind turbines (WT), photovoltaics (PV), micro turbines (MT), and energy storage systems (ESS). It comprehensively considers the volatility and uncertainty of wind and solar power output, and, under the premise of meeting distribution network operation constraints and equipment operating conditions, achieves a quantitative assessment of the aggregate flexibility boundary of the VPP and the coordinated economic dispatch of multiple resource types. This method can effectively assess the upswing and downswing capabilities of the VPP and connect the flexibility assessment and economic dispatch processes. It is applicable to VPP operation scenarios with multiple distributed energy resources, providing technical support for their optimized dispatch and contributing to improving the renewable energy absorption capacity and the decarbonization level of the power system.

[0030] In some embodiments, the technical framework diagram of the method is as follows: Figure 1 As shown.

[0031] 1. Distribution network model

[0032] Within the normal operating range, the voltage deviation and voltage drop of the distribution network are relatively small, and the changes in power injection by power flow variables exhibit an approximately linear characteristic. A linear model can significantly reduce computational complexity while maintaining high accuracy. Therefore, for the distribution network connected to the VPP, a single-phase linear power flow model is established to characterize the impact of distributed renewable energy, ESS, and MT resources on node voltage, branch current, and the power exchanged by the upstream grid.

[0033] Specifically, with the distribution network balancing node and the rest A single-phase radial distribution network is formed by several unbalanced nodes. The network topology and line parameters are described using node admittance matrices. The system node admittance matrix is ​​expressed as follows: (1) in, Indicates the self-admittance of the equilibrium node. This represents the admittance submatrix of the non-equilibrium nodes. and These represent the mutual admittance matrices between the balancing node and the other nodes, respectively.

[0034] Let the voltage at the slack node be The voltage vector of the non-equilibrium node is The node complex power injection vector is ,in, and These represent the active power injection and reactive power injection at each node, respectively. From the network equations, we can obtain... (2) Furthermore, from the nodal complex power relationship achievable (3) Therefore, the power flow equations of a single-phase distribution network can be written in the following fixed-point form: (4) in, (5) This represents the node voltage vector under zero injection conditions.

[0035] To obtain a linear power flow model suitable for distribution networks, a reference voltage vector is selected. The reference voltage is the rated voltage. pu. The nonlinear terms in the above fixed-point equations By approximately fixing the reference voltage, a linear approximate model of the nodal voltage phasors can be obtained: (6) make Then there is (7) in, , This is the node voltage constant term under zero injection conditions; Sensitivity matrix for the effect of node injected power on node voltage.

[0036] (8)

[0037] here, Represents the reference node voltage vector The complex conjugate, This represents a diagonal matrix constructed from vector elements. Since the rated voltage is used as the reference in the specific implementation of this invention, then... (9) (10) The sensitivity matrix can now be further simplified to... (11) This allows for the establishment of an explicit linear mapping relationship between the voltage of each node in the distribution network and the changes in the output of distributed resources at each node, facilitating its embedding into subsequent flexibility assessment and economic dispatch models.

[0038] To facilitate direct constraint of node voltage amplitude in optimized scheduling, a linear approximation model of the voltage amplitude can be further established. Based on the small voltage deviation assumption, we have... (12) in, Sensitivity matrix for the relationship between injected power and node voltage amplitude. For the node voltage magnitude constant term: (13) (14) Therefore, the node voltage safety constraint can be written in the form of a linear inequality: (15) To describe the power exchange between the VPP and the upstream grid, the slack node injection current is: (16) The complex power at the equilibrium node is (17) Substitute into the linear voltage model , can be obtained (18) in, The sensitivity matrix for injecting power into nodes and exchanging power with balancing nodes. For constant terms: (19) (20) Therefore, the active and reactive power exchange between VPP and the upstream grid can be expressed as follows: (twenty one) (twenty two) This allows us to obtain an explicit relationship between the power purchased and sold by the upper-level power grid and the output of distributed resources within the distribution network, providing a basis for the economic dispatch of VPPs.

[0039] To describe the line power flow and branch current constraints, assume the distribution network has a total of There are 1 branch, and the diagonal matrix of branch admittances is: (twenty three) set up For the branch-unbalanced node incidence matrix, Let the branch-balance node correlation vector be denoted as , then the branch current vector can be expressed as . (twenty four) Substituting into the linear voltage model, we can obtain (25) in, (26) (27) Therefore, the line current safety constraint can be written as follows: (28) in, Representation matrix The OK, Representing vectors The Each element.

[0040] This achieves a linear mapping between branch current and node injected power. In a specific implementation of this invention, line current constraints are constructed based on the above approximate relationship of branch current to ensure the safe operation of the distribution network during VPP multi-resource coordinated scheduling.

[0041] The distribution network model established in this invention establishes a linear coupling relationship between the active and reactive power outputs of distributed renewable energy, MT and ESS resources and the voltage of distribution network nodes, branch currents and the exchange power with the upper-level grid. While preserving the physical meaning of network operation constraints, it reduces the solution complexity of VPP flexibility assessment and optimization scheduling models, and provides a unified network constraint for subsequent VPP flexibility assessment and economic scheduling.

[0042] 2. Mathematical Model of Distributed Energy

[0043] This invention establishes mathematical models for WT, PV, MT, and ESS, respectively, to mathematically describe the adjustability and operational boundaries of various types of distributed energy resources. WT, PV, and MT primarily reflect their output upper and lower bounds and ramp rate characteristics, while ESS reflects their bidirectional power regulation and energy transfer capabilities across time periods. These four distributed energy resource models collectively form the basis for VPP flexibility assessment and economic dispatch models.

[0044] All the distributed energy models described above are constructed based on discrete time periods, assuming the scheduling period is divided into... The time period, the first The output of various equipment in each time period is used as an optimization variable in subsequent scheduling solutions.

[0045] (1) Mathematical model of distributed wind turbines

[0046] Let the first Taiwan WT during the time slot The actual active power output is Its maximum available output during the corresponding time period is Unproductive efforts The maximum and minimum reactive power outputs are respectively and Then we have: (29) (30) To reflect the power regulation limitations between adjacent WT time periods, ramping constraints are established: (31) in, and The first The lower and upper limits of the active ramp rate of the Taiwan WT.

[0047] (2) Mathematical model of distributed photovoltaic

[0048] For PV, let the first... PV in time period The actual active power output is The maximum available output for the corresponding time period is Unproductive efforts The maximum and minimum reactive power outputs are respectively and Then the following condition is met: (32) (33) To limit the variation in active power regulation between adjacent PV time periods, a ramping constraint is established: (34) in, and The first The lower and upper limits of the active ramp-up rate for each PV.

[0049] (3) Mathematical model of micro gas turbine

[0050] For MT, considering its strong schedulability, it is treated as a controllable conventional power source in the model, mainly including output constraints and ramp rate constraints.

[0051] Let the first Taiwan MT during the time period Contributing to the cause Unproductive efforts Then the following condition is met: (35) (36) in, For the first Taiwan MT during the time period Those who have contributed to the higher realms; and These represent the maximum and minimum reactive power output, respectively.

[0052] To reflect the limitation of the unit's dynamic characteristics on the continuous adjustment capability of MT output, a ramp rate constraint is established: (37) in, and The first The lower and upper limits of the ramp rate for Taiwan MT.

[0053] (4) Mathematical model of energy storage system

[0054] For ESS, let the first... ESS in time period The charging power and discharging power are respectively and The upper limits of their maximum charge and discharge power are respectively and Then we have: (38) (39) ESS net active effort output is: (40) The upper and lower limits satisfy: (41) in, Indicates ESS discharge. This indicates that the ESS is charging.

[0055] Let the first ESS in time period Unproductive output Its maximum and minimum reactive power output are respectively and Then we have: (42) Let the first ESS in time period The energy state is Its charging efficiency and discharging efficiency are respectively and The duration is Then its energy state recursive equation is: when hour, (43) when hour, (44) The ESS energy state must also satisfy upper and lower bound constraints: (45) in, and The first The minimum and maximum energy states allowed by each ESS.

[0056] To ensure that the ESS still has continuous scheduling capability after the scheduling period ends, an energy regression constraint is further set at the end of the ESS period: (46)

[0057] 3. Using sub-Brussels bar modeling to handle wind and solar uncertainties.

[0058] Because renewable energy output is stochastic, the operational constraints of a Virtual Power Plant (VPP) need to include the uncertainty introduced by prediction errors. Unlike the assumption that prediction errors follow a given parameterized probability distribution, this invention employs distributed robust modeling to address this uncertainty. Specifically, a distributed fuzzy set is constructed based on historical samples, and the chance constraints are required to meet a pre-set confidence level under any probability distribution within the distributed fuzzy set. This reduces the model's reliance on precise distribution assumptions and enhances its robustness to distribution mismatches.

[0059] set up This represents the power prediction error vector, i.e. , Let the historical samples be the empirical mean vector and covariance matrix, respectively. and Then a distributed fuzzy set is defined as follows: (47) in, This represents the support set of prediction errors estimated from historical data. The support set of prediction errors is defined as the central 95th percentile empirical interval for each device and time period: (48) (49) (50) That is, the 2.5% and 97.5% empirical quantiles of the error samples are used, and the intersection with the physical output boundaries of WT and PV is taken, with a support set confidence level of 95%.

[0060] Accordingly, the general form of stochastic running constraints can be expressed as the following distributed bar chance constraints: (51) in, For the decision variable vector, , and The constant coefficients, Given a confidence level.

[0061] For ease of derivation, we define a scalar random variable. (52) Equation (51) can be rewritten as (53) Based on distributed fuzzy sets Definition of random variable The mean satisfies (54) And its variance satisfies (55) Again (56) Then equation (53) is equivalent to the requirement (57) This invention employs a one-sided Chebyshev-Cantelli inequality to deterministically reconstruct the boundary of random uncertainty. This method relies only on the first and second moments of the prediction error, i.e., its mean and variance, thus not requiring the error to follow a normal distribution. It is suitable for situations such as WT and PV where the prediction error distribution is unknown or the sample size is limited. For any scalar random variable with a finite mean and variance, the one-sided Chebyshev-Cantelli inequality can be used to obtain... (58) Therefore, as long as the following conditions are met (59) This guarantees that equation (57) holds true. From equation (59), we can further obtain... (60) Combining the upper bound of variance in equation (55), the following sufficient condition holds: (61) Substituting into equation (56), we can obtain a conservative deterministic reconstruction of the pluripotent chance constraint (51): (62) in (63) Equation (62) shows that the original stochastic constraint is replaced by a deterministic safety margin. This safety margin consists of two parts: the nominal impact term of the prediction error. And the degree of discreteness characterized by second-order moment information in the worst case. In this way, a robust and manageable constraint form for distribution mismatch can be obtained without precisely knowing the true probability distribution of the prediction error. This method relies only on the first and second-order statistical information of historical samples, making it more suitable for scenarios where the true distribution is difficult to identify accurately.

[0062] This method determines the support set of prediction errors based on historical prediction error samples, using the 95th percentile empirical interval at the center of each device and time period, and further intersects it with the physical feasible boundary of new energy output. This method can reduce the impact of extreme values ​​in the samples on the width of the support set and avoid over-conservatism caused by using historical extreme values. The confidence level of the support set can be adjusted between 90% and 99% according to the system risk preference.

[0063] 4. Virtual Power Plant Flexibility Assessment Method

[0064] To quantify the up- and down-regulation capabilities that a Virtual Power Generation (VPP) can provide within a scheduling cycle, a VPP flexibility assessment model considering wind and solar uncertainties and temporal coupling constraints is established. Geometrically, VPP flexibility aggregation essentially projects the high-dimensional feasible domains of all distributed energy sources onto the power trajectory space of the point of common junction (PCC). Since accurate projection of the high-dimensional feasible domain onto the low-dimensional power trajectory space and explicit polyhedral representation typically involve high computational complexity, to reduce computational complexity, based on the physical characteristics of DERs, they are divided into generator-type DERs (GDER) and battery-type DERs (BDER). GDERs are characterized by "power boundary + ramping constraints," while BDERs are characterized by "bidirectional power boundary + energy state coupling constraints across time periods." This invention classifies WT, PV, and MT as GDERs and ESS as BDERs. GDERs can be aggregated into virtual generators (VG), and BDERs can be aggregated into virtual batteries (VB).

[0065] Assume the scheduling time domain includes In each discrete time period, the active and reactive power outputs of each distributed energy source within the VPP constitute the decision variables. Combining the constraints of distributed energy devices, the linear power flow constraints of the distribution network, and the uncertainties of WT and PV output, the feasible operating domain of VPP can be uniformly represented as: (64) (65) (66) in, For time period The decision variable vector, This represents static constraints such as power upper and lower limits, node voltage, and branch current within each time period. This represents cross-time-period coupled constraints such as ESS energy state and unit ramp-up. This represents the active power output trajectory of the VPP at the point of common junction.

[0066] Based on the physical characteristics of distributed resources, WT, PV, and MT can be classified as generator-type resources, while ESS can be classified as battery-type resources. Therefore, the active power flexibility trajectory of VPP is decomposed into the sum of two parts: virtual generator and virtual battery, i.e.: (67) in, The output trajectory of the virtual generator. Output trajectory for virtual battery.

[0067] For GDER, its aggregation flexibility region is described using power upper and lower bounds and ramping constraints: (68) (69) in, and These represent the time periods of GDER. The upper and lower output boundaries, and These represent its ability to climb uphill and downhill, respectively.

[0068] For BDER, its aggregation flexibility region is described using power upper and lower bounds and energy state constraints: (70) (71) (72) (73) (74) in, and These represent the time periods of BDER. The charge and discharge power boundary, and These represent the upper and lower bounds of the BDER energy, respectively.

[0069] Based on this, the flexibility area of ​​VPP can be represented as: (75) in, and These are the aggregation flexibility areas for virtual generators and virtual batteries, respectively; , Indicates discharge. This indicates charging. From this, the adjustable power boundary of the VPP can be obtained throughout the entire scheduling cycle, providing a flexible regional scheduling basis for subsequent economic scheduling.

[0070] (1) Robust optimization separates network constraints

[0071] Since the GDER and BDER within the VPP are connected to the same distribution network, their active and reactive power outputs are coupled through node voltage constraints, branch current constraints, and point of common coupling power constraints. Directly aggregating all resources for overall flexibility would result in a high model dimensionality and large solution complexity. Therefore, this invention introduces the concept of robust optimization to separate the network constraints of the GDER and BDER, effectively absorbing the coupling effects of the two types of resources in the distribution network into their respective constraint sets, thereby constructing decoupled network constraints that can be used for flexibility assessment.

[0072] Let the first The decision variable vector for each time period is divided into three parts: the GDER active power output vector. BDER active power output vector and all resource reactive power output vectors Then the linear inequalities for each time period, which include network operation constraints and device static constraints, can be uniformly expressed as: (76) Furthermore, the constraint terms related only to the BDER active variables and the remaining coupled constraint terms can be separated and organized as follows: (77) in, This represents the feasible region constraints that the BDER itself must satisfy. This represents the coupling constraint formed by the combined effect of two types of resources through network flow relationships.

[0073] To obtain independent network constraints specific to the GDER, this invention incorporates the active and passive forces of the BDER. Treated as an uncertain parameter, and required to satisfy BDER feasible output in all scenarios: (78) Both GDER and reactive power control variables can guarantee that network constraints hold, i.e., satisfy: (79) Using robust optimization, the above constraints are equivalently transformed into deterministic linear constraints containing only GDER variables and reactive power variables. For the first coupled constraint... Okay, there are: (80) (81) Solving the above linear programming problem using the Gurobi solver yields the GDER in time intervals. Decoupling network constraints as follows: (82) in, (83) and These are the constraint coefficient matrix and constant term vector of the GDER network obtained after robust optimization transformation, respectively.

[0074] Similarly, to obtain independent network constraints specific to the BDER, the active and passive inputs of the GDER can be... Treated as uncertain parameters, under all scenarios where its own feasible operating conditions are met, the BDER and reactive power variables are required to always satisfy network coupling constraints. After the same robust optimization transformation, the BDER in the time period can be obtained. Decoupling network constraints as follows: (84) in, (85) and These are the coefficient matrix and constant term vector of the BDER network constraints, respectively.

[0075] Based on the robust optimization separation results described above, the network constraints originally formed by the coupling of GDER, BDER, and reactive power regulation are transformed into independent constraint sets applicable to VG and VB aggregation modeling, respectively. Therefore, all technical constraints of GDER can be uniformly written as: (86) All technical constraints of BDER can be uniformly written as: (87) in, In addition to the decoupled network constraints mentioned above, it also includes power boundaries, ramping constraints, and uncertainty constraints for WT, PV, and MT. In addition to the corresponding decoupling network constraints, it also includes ESS charging and discharging power constraints and energy state constraints.

[0076] (2) Boundary contraction algorithm

[0077] After separating the network constraints of GDER and BDER, the corresponding original feasible regions can be obtained respectively. Since the precise projection polyhedron of the original feasible region onto the active output trajectory space of the common connection point is difficult to obtain analytically, this invention adopts a boundary shrinkage algorithm to iteratively identify the equivalent parameters of VG and VB, thereby obtaining the aggregate flexibility boundary that satisfies decomposability feasibility.

[0078] The following explanation uses VG as an example; the parameters of VB can be solved in the same way.

[0079] 1) Step 1: Construct the VG aggregation boundary model

[0080] The flexibility region of VG is represented as: (88) in, This represents the active power output trajectory of VG throughout the entire scheduling cycle. For a fixed constraint structure matrix, This is a boundary parameter vector, whose elements include the upper and lower limits of output for each time period, as well as the uphill and downhill climbing capabilities for adjacent time periods.

[0081] Correspondingly, the original high-dimensional feasible region of VG can be represented as: (89) in, For GDER active and reactive power decision variables, GDER device constraints and decoupled network constraints. for A linear mapping relationship to the output power trajectory at the point of common junction.

[0082] 2) Step 2: Obtain the initial boundary parameters

[0083] First, by solving a set of linear programming problems, the upper and lower bounds of output and the upper and lower bounds of climbing at each time period are obtained, thus yielding the initial parameter vector. The upper bound of output for each time period can be expressed as: (90) (91) The uphill climbing ability in adjacent time periods can be expressed as: (92) (93) Similarly, the lower bound of the force and the downhill climbing ability can be calculated, thereby constructing the initial aggregation region: (94) The initial boundary is typically an approximation of the true projective feasible region.

[0084] 3) Step 3: Search for infeasible points in the current flexibility zone

[0085] In the In the next iteration, the current flexibility region is searched. Are there any infeasible points? Its mathematical criterion can be written as existence. (95) But it doesn't exist. Make (96) (97) Established at the same time.

[0086] If any of the above infeasibility points exist, it means that the current aggregation region still contains output trajectories that the original feasible region could not achieve, and the boundary needs to be further shrunk; if none exist, then the current parameters... These are the final boundary parameters.

[0087] The search for infeasible points in the flexibility region can be constructed as the following extreme value search model: (98) (99) (100) (101)

[0088] 4) Step 4: Find the nearest feasible point

[0089] For the infeasible trajectory points obtained from the search Search for the nearest feasible boundary point in the true projected feasible region. The optimization model can be expressed as: (102) (103) This boundary point is used as the target point for subsequent boundary corrections.

[0090] 5) Step 5: Update boundary parameters and shrink the polyhedron

[0091] Suppose the current infeasible trajectory point The corresponding set of active constraint indexes is ,satisfy: (104) While maintaining the constraint structure matrix Without changing the boundary parameter vector, we correct the boundary parameter vector to obtain a new parameter vector. : (105) (106) (107) (108) And make the nearest feasible boundary point It becomes a boundary point or pole on the new boundary. Therefore: (109) This method gradually eliminates infeasible areas from the current flexibility zone.

[0092] 6) Step 6: Iterate until convergence

[0093] Repeat steps 3 through 5 until there are no infeasible trajectory points within the current aggregation region, i.e., when the optimal value of the extreme value search model is not greater than the preset feasibility tolerance. When the number of iterations reaches the preset maximum number of iterations, the algorithm is considered to have converged; when the number of iterations reaches the preset maximum number of iterations, the iteration stops and an approximate boundary is output. The parameter vector obtained after algorithm convergence. This represents the final flexibility parameter of VG. Using the same method, the parameter vector of VB can be further obtained. The difference lies in that the boundary parameters of VB mainly include the upper and lower bounds of charging and discharging power and the upper and lower bounds of energy state at each time period, but its parameter iterative correction process is the same as that of VG.

[0094] In summary, this invention uses a boundary shrinkage algorithm to gradually correct the boundary parameters of VG and VB, obtaining a VPP flexibility aggregation result with decomposable feasibility, and finally obtaining a low-conservatism equivalent flexibility region that can be used for economic scheduling.

[0095] 5. Virtual Power Plant Economic Dispatch Method

[0096] This invention proposes an economic dispatch method for VPPs (Virtual Power Generation Platforms) to economically dispatch various distributed energy sources within them. Specifically, based on the aforementioned VPP aggregation flexibility assessment results, an optimal dispatch model is constructed that includes wind power curtailment (WT), photovoltaic (PV), medium-duty (MT), and energy storage (ESS). The objective is to minimize the sum of WT wind curtailment costs, PV solar curtailment costs, MT generation costs, ESS charging and discharging operation costs, carbon emission costs, and distribution network loss costs. Under the conditions of satisfying distribution network safety operation constraints, distributed energy output boundary constraints, ramp rate constraints, and ESS energy constraints, given the total active power output dispatch curve of the VPP at the PCC (Power Constraint Center), the optimal output of each type of distributed energy source is obtained in each dispatch period. Compared with methods that only focus on the external equivalent flexibility boundary of the VPP, this invention further provides an economic de-aggregation mechanism for internal resources.

[0097] (1) Objective function

[0098] This invention aims to minimize the total operating cost of a VPP (Variable Power Plant). The total operating cost includes wind curtailment penalty costs, solar curtailment penalty costs, medium-duty power generation costs, ESS (Emergency Storage System) charging and discharging operation costs, carbon emission costs, and distribution network loss costs. The objective function is expressed as: (110) in, This represents the total operating cost of the VPP within the scheduling period; This indicates the cost of curtailment penalties for WT and PV power generation; Indicates the cost of generating electricity using MT; This indicates the operating cost of the ESS during charging and discharging; Indicates the cost of carbon emissions; This represents the cost of network losses.

[0099] The curtailment penalty costs for WT and PV are expressed as follows: (111) in, and These represent the penalty coefficients for wind curtailment and solar curtailment, respectively. and These represent the power of wind curtailment and solar curtailment, respectively.

[0100] The power of wind and solar power curtailment is: (112) (113) And satisfy: (114) (115) The cost of MT power generation is represented by a quadratic function: (116) in, , and The first The invention defines the secondary cost coefficient, primary cost coefficient, and fixed cost coefficient for a single MT (Mechanical Transport Unit). In this invention, the MT remains operational throughout the entire scheduling cycle, without considering start-up or shutdown decisions. The fixed cost is the unit's online maintenance cost.

[0101] The operating cost of ESS charging and discharging is expressed as follows: (117) in, and These represent the charging and discharging operating cost coefficients of the ESS, respectively. To reflect the lifespan loss of the ESS during its cyclic use, the charging and discharging cost of the ESS can be determined based on the ESS replacement cost, cycle life, and available depth of discharge: (118) (119) (120) in, This represents the replacement cost per unit capacity of ESS. Indicates the ESS cycle life. Indicates the available depth of discharge. and These represent the operation and maintenance costs of the ESS charging and discharging processes, respectively. and These represent the proportions of degradation costs allocated to the charging and discharging processes, respectively.

[0102] The carbon emission cost of MT is expressed as: (121) in, Indicates carbon emission price, Indicates the first Carbon emission factor per unit of electricity generated by a MT (Metal Transport Unit).

[0103] Network loss cost is calculated based on the linearized branch current model, the first... The active power loss of the distribution network during a given time period can be expressed as: (122) in, For the collection of distribution network branches, For the first The per-unit resistance value of each branch. This is the per-unit value of the branch current. This is the system power baseline value. The corresponding network loss cost is: (123) in, This indicates the price calculated based on the amount of electricity consumed by the network.

[0104] (2) Constraints

[0105] 1) VPP total output tracking constraint

[0106] Given the scheduling curve at PCC The node injection power at PCC is Then the combined output of each distributed energy source within the VPP should satisfy...

[0107] - (124)

[0108] 2) Network security constraints

[0109] Node voltage and branch current should meet the following requirements: (125) (126) in, This is the lower limit of the allowable voltage for the node. The upper limit of the allowed voltage for the node. This represents the maximum allowable current for the branch.

[0110] 3) Operational constraints of distributed wind turbines and distributed photovoltaic systems

[0111] set up and Let WT and PV represent the upper bounds of usable output forces obtained after correction using the split-bar method, then we have: (127) (128) (129) (130) (131) (132)

[0112] 4) Operating constraints of micro gas turbines (133) (134) (135)

[0113] 5) Power and energy constraints of energy storage systems

[0114] The upper and lower limits of active and reactive power during ESS charging and discharging are as follows: (136) (137) (138) The ESS energy state evolution equation is (139) ESS energy upper and lower limits constraints are (140) To ensure that the ESS still has continuous scheduling capability after the scheduling period ends, an energy regression constraint is further set at the end of the ESS period: (141) To prevent ESS from charging and discharging simultaneously during the same period, this invention introduces 0-1 variables for constraint. Therefore, the mutual exclusion constraint for ESS charging and discharging can be expressed as: (142) (143) (144) (145)

[0115] 6) Node injection power constraints

[0116] VPP uses node-based modeling internally, then the nodes During the period The net active and reactive power injections can be written as (146)

[0117] (147) After summarizing the above objective function and constraints, the model can be uniformly written as: (148) (149) (150) (151) Equation (149) corresponds to the upper and lower bound constraints of network and equipment output in a single time period, and Equation (150) corresponds to the ramp rate and ESS energy constraints across time periods.

[0118] Best practice

[0119] 1. Basic Overview of Implementation

[0120] This invention uses the MATLAB platform for modeling and solving, utilizes the YALMIP toolbox to build an optimization model, and calls the Gurobi solver to complete flexibility assessment and economic scheduling calculations.

[0121] To verify the effectiveness of the proposed VPP flexibility assessment and economic dispatch method considering wind and solar uncertainties, a case study was conducted using the IEEE 33-node distribution system as the VPP access network. The distribution network adopted a single-phase equivalent linearized power flow model, with node voltage amplitude constraints set at 0.95~1.05 pu, and branch current upper limits determined based on line rated capacity. The confidence level for the distributed bar chance constraint was 0.9, i.e. The boundary shrinkage algorithm has a preset feasibility tolerance. Take 10 -5 The maximum number of iterations is 50.

[0122] In the implementation example of this invention, the scheduling period is divided into 12 discrete time periods, each with a length of 1 hour. The VPP consists of 4 WTs, 4 PVs, 2 MTs, and 4 ESSs. The relevant parameters of various distributed resources are shown in Table 1. Among them, WTs, PVs, and MTs are classified as GDERs, and ESSs are classified as BDERs.

[0123] Table 1. Relevant parameters of various distributed resources

[0124] Each distributed resource is connected to the distribution network according to preset nodes. WT is connected to nodes 6, 24, 27, and 30; PV is connected to nodes 5, 8, 23, and 29; MT is connected to nodes 2 and 26; and ESS is connected to nodes 3, 7, 11, and 28. The corresponding network topology is shown in the figure below. Figure 2 As shown.

[0125] Considering the uncertainty of wind and solar power output, the example uses a distributed bar model to handle the available power output of wind turbines (WT) and photovoltaics (PV). Specifically, a conservative upper bound for available power output is constructed based on historical prediction error samples and combined with error statistical characteristics, serving as the input for subsequent flexibility assessment and economic scheduling. The predicted power output curves for each WT and PV are shown below. Figure 3 and Figure 4 As shown.

[0126] In the economic dispatch of VPP, the operating cost of MT (Metal Transport Unit) is represented by a quadratic function to characterize its fuel consumption increasing with power output. Penalty costs are set for wind and solar curtailment to encourage the priority consumption of renewable energy sources such as WT (Wind Power Generation) and PV (Photovoltaic Power Generation). Corresponding operating costs are set for the charging and discharging processes of ESS (Energy Shielded Storage Unit) to reflect the operation and maintenance costs and lifespan degradation costs during the ESS's cyclical use. Simultaneously, to reflect the requirements of low-carbon dispatch, the model further introduces the carbon emission cost of MT, calculating the economic cost of carbon emissions based on unit output, emission factors, and carbon prices. Furthermore, considering the impact of line losses on dispatch economics during distribution network operation, the model includes network loss electricity in the objective function at an equivalent electricity price. Detailed cost parameters are shown in Table 2.

[0127] Table 2 Key parameters of VPP economic dispatch cost

[0128] 2. Optimize Result Analysis

[0129] Considering the uncertainties in WT and PV output, the VPP flexibility boundary solution results are as follows: Figure 5 As shown.

[0130] Depend on Figure 5 It can be seen that throughout the entire dispatch cycle, the total flexibility boundary of VPP is jointly constituted by VG flexibility and VB flexibility. VB provides relatively stable upward and downward adjustment capabilities, while VG further expands the upward adjustment space of VPP. Specifically, the total upward adjustment capability of VPP reaches a relatively high level from period 5 to period 8, while it is relatively low at the beginning and end of the period, indicating that the contribution of VG to overall flexibility exhibits obvious time-varying characteristics as the available output of renewable energy WT and PV changes over time. In contrast, the flexibility range corresponding to VB is relatively stable overall, indicating that ESS can provide relatively stable two-way adjustment support in all periods.

[0131] The comparison of VPP flexibility boundary solution results without considering the uncertainties of WT and PV output is shown in the figure below. Figure 6 As shown.

[0132] from Figure 6 It can be seen that, throughout the entire scheduling cycle, after considering wind and solar uncertainties, the upper adjustment boundary of VPP shifts significantly downward, while the lower adjustment boundary changes relatively little. This indicates that the output prediction errors of WT and PV mainly weaken the upward adjustment capability of VPP. This is because, when considering uncertainties, the upper bound of available output of WT and PV is reduced after conservative correction, thus shrinking the overall upward adjustment flexibility boundary. In contrast, the downward adjustment capability is more affected by the absorption capacity of ESS, minimum output constraints, and basic operating conditions, so its variation is relatively limited. If wind and solar uncertainties are not considered, the obtained flexibility boundary, although wider, may overestimate the actual adjustable capability of VPP; while the flexibility boundary obtained after considering wind and solar uncertainties, although relatively narrower, is more in line with actual operating conditions and can provide a more reliable boundary basis for VPP safe scheduling and risk control.

[0133] Depend on Figure 7 It can be seen that the ramp rate boundary of VG is not a fixed constant between adjacent time periods, but changes with time. The results show that the ramp capability reaches a high level near time periods 5 and 6, indicating that VG has a strong rapid ramp capability during this period; while the absolute value of the ramp boundary is large in time periods 7 and 8, indicating that the system also has a strong ramp capability.

[0134] Depend on Figure 8 It can be seen that the energy boundary of the VB exhibits a characteristic of first expanding, then stabilizing, and finally gradually shrinking at the end of the scheduling cycle. In the initial stage, the feasible energy range gradually expands, indicating that the aggregated ESS can form a large energy flexibility space through charging and discharging adjustments. In the middle stage, the upper and lower boundaries remain relatively stable, indicating that the ESS has the ability to continuously provide bidirectional adjustment support. As the scheduling process approaches the end, the feasible energy range gradually shrinks and eventually returns to the initial energy state. This is because the model sets a constraint that the energy state of the ESS at the end of the scheduling period equals the energy state at the initial moment. This constraint can prevent the ESS from overcharging or discharging at the end of the scheduling period, ensuring that it still has the ability to continue operating after completing the current scheduling cycle, thereby improving the rationality and sustainability of the scheduling results.

[0135] Based on the VPP flexibility assessment results, four VPP scheduling scenarios were constructed. Different scheduling curves exhibit different distribution characteristics within the boundary regions: Scheduling curve 1 generally operates close to the upper boundary, indicating that the VPP undertakes a high outbound or upward transmission task in this scenario; Scheduling curve 2 approaches zero power or even locally approaches the lower boundary, reflecting that the VPP has a strong absorption capacity or a low net output level in this scenario; Scheduling curve 3 is located in the middle of the flexibility region, demonstrating a relatively stable operation; Scheduling curve 4 is basically located in the middle of the flexibility region, but with relatively large fluctuations. The VPP scheduling curves and their corresponding upper and lower flexibility bounds are shown below. Figure 9 As shown.

[0136] Figure 10 The output allocation results of various distributed energy sources within a VPP are presented under four scheduling scenarios. It can be seen that different scenarios correspond to different total output targets, and the coordination methods of WT, PV, MT, and ESS also differ significantly. In Scenario 1, the overall total output of the VPP is at a high level, with WT, PV, and MT jointly undertaking the power supply task. ESS discharges in the initial stage to supplement output and switches to charging at the end to balance the energy state. In Scenario 2, the VPP exhibits negative net output in the early stage, mainly relying on ESS charging to absorb energy. The total output approaches zero in the middle stage, and at the end, ESS discharges along with a small amount of WT output to meet scheduling requirements. In Scenario 3, the overall total output is relatively stable, mainly supported by WT and PV, with ESS performing small-scale charging and discharging in some periods to achieve power balance. In Scenario 4, the total output fluctuates significantly. ESS charges during low-output periods and discharges during high-output periods, coordinating with WT, PV, and MT to respond to rapidly changing scheduling demands. Overall, the combined output of distributed energy resources in each scenario can track the given VPP scheduling curve, indicating that the proposed economic scheduling method can reasonably allocate the output of various resources according to different operational needs and give full play to the energy transfer and power regulation functions of ESS.

[0137] Table 3 presents the cost composition and total cost of VPP under four scheduling scenarios. It can be seen that the total cost varies significantly across different scenarios. Scenario 2 has the highest total cost, mainly due to the large proportion of wind and solar curtailment costs, indicating a low level of renewable energy absorption in this scenario. Although the costs of MT, ESS, carbon emissions, and network losses are relatively small, they still cannot offset the increased costs caused by large-scale wind and solar curtailment. Specifically, MT output is 0, but since MT is always running, it requires online maintenance costs, so the MT cost is not zero. Carbon emission costs are only related to actual power generation, so when MT output is 0, the carbon emission cost is 0. Scenario 1 has the lowest total cost, with zero wind curtailment costs and low solar curtailment costs. Its cost mainly consists of MT power generation costs, carbon emission costs, and ESS operating costs, indicating that this scenario can meet high power output demands while effectively absorbing renewable energy. The total costs of scenarios 3 and 4 are at a moderate level. Scenario 3 has lower costs for MT, ESS, and carbon emissions, but still incurs some costs associated with wind and solar curtailment. In scenario 4, the cost distribution is relatively balanced, reflecting the operational characteristics of multiple resources, including WT, PV, MT, and ESS, participating in the regulation process. Overall, wind and solar curtailment costs remain the main factor affecting the total cost of VPP. Improving the level of renewable energy consumption and rationally coordinating the output of conventional units and ESS will help improve the overall operational economy of VPP.

[0138] Table 3. Cost composition and total cost of VPP in four scheduling scenarios

[0139] In summary, this invention proposes a method for VPP flexibility assessment and economic dispatch considering wind and solar uncertainties. This method establishes a linear power flow model of the distribution network and multiple types of distributed energy models, employs the bibloc robustness method to handle wind and solar output uncertainties, and utilizes robust optimization and boundary shrinkage algorithms to extract the equivalent flexibility boundaries of virtual generators and virtual batteries. Based on this, an economic dispatch model is constructed that considers MT generation costs, wind and solar curtailment costs, ESS operating costs, carbon emission costs, and network loss costs, achieving coordinated and optimized operation of multiple resources within the VPP.

[0140] The numerical results demonstrate that this invention can effectively characterize the impact of wind and solar uncertainties on the flexibility boundary of a Virtual Power Supply (VPP), avoiding overestimation of the actual adjustability of the VPP. Simultaneously, it can rationally coordinate the output allocation of Wt, PV, MT, and ESS according to different dispatching scenarios, improving the level of renewable energy absorption and minimizing the internal resource operating costs of the VPP while ensuring the safe operation of the distribution network. Therefore, this invention possesses good robustness, practicality, and engineering application value, providing effective technical support for the flexibility assessment and optimized dispatching of VPPs under conditions of high renewable energy integration.

[0141] The background section of this invention may include background information about the problems or environment in which the invention is being developed, and is not necessarily a description of prior art. Therefore, the content included in the background section does not constitute an admission of prior art by the applicant.

[0142] The above description provides a further detailed explanation of the present invention in conjunction with specific / preferred embodiments, and it should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various substitutions or modifications can be made to these described embodiments without departing from the concept of the present invention, and all such substitutions or modifications should be considered within the scope of protection of the present invention. In the description of this specification, the reference to terms such as "an embodiment," "some embodiments," "preferred embodiment," "example," "specific example," or "some examples," etc., indicates that the specific features, structures, materials, or characteristics described in connection with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification and the features of different embodiments or examples. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions, and modifications can be made herein without departing from the scope of protection of the patent application.

Claims

1. A method for assessing the flexibility and economic dispatch of a virtual power plant considering the uncertainties of wind and solar power, wherein the virtual power plant is connected to a distribution network and includes distributed energy resources, the distributed energy resources including renewable energy units with uncertain output and adjustable units with adjustable output, the renewable energy units including distributed wind turbines and distributed photovoltaics, characterized in that, Includes the following steps: A linearized network model of the distribution network is established, which is used to characterize the linear mapping relationship between node injected power and node voltage amplitude, branch current and slack node exchange power; Establish equipment operation constraints for the renewable energy units and the adjustable units; Based on historical data, the partial Brussels bar optimization method is used to handle the uncertainty of renewable energy unit output and construct corresponding uncertainty operation constraints. Based on the linearized network model, the equipment operation constraints, and the uncertain operation constraints, the flexible adjustable domain of the virtual power plant at the point of common connection is determined. The flexible adjustable domain is used to characterize the power regulation capability that the virtual power plant can provide within the scheduling cycle. With the goal of minimizing operating costs, and using the aforementioned adjustable flexibility domain, equipment operation constraints, and linearized network model as constraints, optimization solutions are performed to generate optimal scheduling instructions for each unit within the virtual power plant.

2. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1, characterized in that, The linearized network model is a single-phase linear power flow model.

3. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1 or 2, characterized in that, The steps for establishing the linearized network model of the distribution network include: establishing a linear mapping relationship between node injected power and node voltage amplitude; establishing a linear mapping relationship between node injected power and branch current; and establishing a linear mapping relationship between node injected power and balancing node exchange power.

4. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1, characterized in that, The adjustable unit includes a micro gas turbine and an energy storage system.

5. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 4, characterized in that, The steps for establishing the equipment operation constraints include: establishing upper and lower limit constraints and ramp rate constraints for the output of the distributed wind turbine and distributed photovoltaic; establishing upper and lower limit constraints and ramp rate constraints for the output of the micro gas turbine; and establishing charging and discharging power constraints, energy state constraints, and charging and discharging mutual exclusion constraints for the energy storage system.

6. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1, characterized in that, The steps for handling the uncertainty of renewable energy unit output using the sub-Bruker optimization method include: obtaining the empirical mean vector and covariance matrix of the prediction error based on historical sample data; constructing a distributed fuzzy set constrained by the empirical mean vector and covariance matrix; and transforming the sub-Bruker chance constraint into a deterministic linear safety margin constraint based on the one-sided Chebyshev-Cantelli inequality.

7. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1, characterized in that, The steps for determining the flexible adjustable domain include: aggregating the resources of the virtual power plant into virtual generators and virtual batteries; and using a boundary shrinkage algorithm to iteratively solve the aggregation parameters of the virtual generators and virtual batteries to obtain the boundary of the flexible adjustable domain.

8. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 7, characterized in that, The steps of iteratively solving the boundary shrinkage algorithm include: solving a set of linear programming problems to obtain the initial aggregate boundary parameters of the virtual generator and the virtual battery; searching for infeasible points within the current aggregate boundary; if infeasible points exist, searching for the nearest feasible boundary point in the real feasible domain; updating the boundary parameters based on the nearest feasible boundary point to shrink the current aggregate boundary; iteratively executing the steps of searching for infeasible points and updating boundary parameters until there are no infeasible points within the current aggregate boundary, thus obtaining the final flexible adjustable domain boundary parameters.

9. The virtual power plant flexibility assessment and economic dispatch method considering wind and solar uncertainties according to claim 1, characterized in that, The operating costs include: the cost of generating electricity from micro gas turbines, the cost of penalties for abandoning renewable energy, and the operating costs of energy storage systems.

10. A virtual power plant economic dispatch system considering the uncertainty of wind and solar power, applied to a virtual power plant connected to a distribution network, wherein the virtual power plant includes distributed energy resources, the distributed energy resources include renewable energy units with uncertain output and adjustable units with adjustable output, the renewable energy units include distributed wind turbines and distributed photovoltaics, characterized in that, The system includes: A linearized network modeling unit is configured to establish a linearized network model of the distribution network, wherein the linearized network model is used to characterize the linear mapping relationship between node injected power and node voltage magnitude, branch current and slack node exchange power; The equipment constraint modeling unit is configured to establish equipment operating constraints for the renewable energy unit and the adjustable unit. The uncertainty processing unit is configured to handle the uncertainty of renewable energy unit output based on historical data and using the sub-Bruker optimization method to construct corresponding uncertainty operation constraints. The flexibility assessment unit is configured to determine the flexibility adjustable domain of the virtual power plant at the point of common connection based on the linearized network model, the equipment operation constraints, and the uncertain operation constraints. The flexibility adjustable domain is used to characterize the power regulation capability that the virtual power plant can provide within the scheduling cycle. The economic dispatch unit is configured to minimize operating costs and, with the adjustable flexibility domain, the equipment operation constraints, and the linearized network model as constraints, perform optimization solutions to generate optimal dispatch instructions for each unit in the virtual power plant.

Images