A Real-Time Safe and Economical Dispatch Method and System for DC Distribution Network Based on Marginal Sensitivity

Abstract

This invention discloses a real-time safe and economical dispatch method and system for DC distribution networks based on marginal sensitivity. The method includes: solving for optimization results at the baseline operating point and under the maximum and minimum photovoltaic scenarios; identifying marginal generating units and boundary constraints, and constructing a linear equation set for marginal sensitivity; solving for the photovoltaic fluctuation range as the operating domain when the set of marginal generating units remains unchanged; iteratively constructing a continuous operating domain sequence covering the entire photovoltaic fluctuation range and extracting the marginal sensitivity coefficients of each domain; in the online stage, locating the operating domain based on the actual photovoltaic output, calling the corresponding coefficients to quickly calculate the adjustment amount of each unit, and executing dispatch. This invention transfers complex optimization offline, requiring only linear computation online, balancing voltage safety and economical dispatch, significantly reducing the real-time computation burden, and overcoming the defect of traditional sensitivity methods failing across the safety domain boundary.

Description

Technical Field

[0001] This invention relates to the field of DC distribution network optimization scheduling technology, specifically to a real-time safe and economical scheduling method and system for DC distribution networks based on marginal sensitivity. Background Technology

[0002] DC distribution networks, due to their simple structure, strong controllability, and ease of distributed power source integration, have become one of the important forms of future smart grids. With the increasing penetration rate of photovoltaics in distribution networks, the strong randomness and volatility of its output pose serious challenges to the real-time safety and economic operation of the system. Therefore, how to perform real-time dispatching of adjustable resources (such as gas turbines, energy storage, and power exchanged with the upstream grid) at the lowest operating cost while ensuring voltage safety has become a critical issue that urgently needs to be addressed in the field of DC distribution network operation and control.

[0003] Currently, the optimal dispatching of distribution networks with uncertainties mainly faces the following two problems: Traditional optimization methods struggle to balance computational speed and model accuracy. While deterministic optimization methods (such as optimal power flow models based on second-order cone relaxation) offer high accuracy, their time-consuming solutions make them unsuitable for real-time scheduling requirements at the second or minute level. Stochastic and robust optimization methods, though capable of handling uncertainty, introduce numerous scenarios and constraints, drastically increasing computational burden and rendering them unsuitable for direct real-time online decision-making.

[0004] Existing fast scheduling methods cannot simultaneously guarantee safety and economy. The academic community has developed two types of methods to replace traditional online optimization: one is the safety domain method, which can define the safe operating boundary of the system offline and ensure the safety of real-time operation, but it is difficult to directly generate the economically optimal power allocation rule; the other is the marginal sensitivity method, which can analytically express the linear relationship between unit output and power imbalance, and can achieve fast allocation, but its effectiveness depends heavily on the safety domain in which the current reference operating point is located. Once the reference operating point crosses the safety domain boundary, the sensitivity relationship becomes invalid. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time safe and economical dispatching method and system for DC distribution networks based on marginal sensitivity, so as to solve the problems of difficulty in balancing safety and economy and heavy online optimization computation burden in the prior art.

[0006] To achieve the above objectives, the technical solution provided by this invention is: a real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity, comprising the following steps: S1: Obtain the load forecast and photovoltaic output forecast for a future set time period; based on the load forecast and photovoltaic output forecast, solve the DC distribution network optimization model with the goal of minimizing the distribution network operating cost, obtain the benchmark operating point of adjustable power supply output, and the scheduling results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits constraints of adjustable power supply output; S2: Based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario, the adjustable power sources whose output has not reached the limit are identified as the marginal unit set, and the constraints that the node voltage reaches the limit are identified as the boundary constraint set; based on the identified marginal unit set and boundary constraint set, a set of linear equations characterizing the marginal sensitivity of the marginal unit output to the change in photovoltaic output is constructed. S3: Based on the linear equations of marginal sensitivity, solve for the range of allowable fluctuations in photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged, and use it as the operating domain corresponding to each benchmark operating point; S4: Determine whether there is an uncovered interval between the current operating domain and other operating domains in the photovoltaic power output fluctuation range; if so, take the boundary point of the uncovered interval as the new operating point, re-identify the marginal unit set and boundary constraint set corresponding to the new operating point, construct the corresponding marginal sensitivity linear equation system, and repeat steps S3 to S4 until a continuous operating domain sequence covering the entire photovoltaic power output fluctuation range is obtained; extract and store the marginal sensitivity coefficient corresponding to each operating domain; S5: During the real-time operation phase, obtain the actual photovoltaic output value of the DC distribution network, determine its current operating domain based on the actual photovoltaic output value, call the corresponding marginal sensitivity coefficient, calculate the output adjustment amount of each marginal unit, and execute the scheduling.

[0007] To optimize the above technical solution, the specific measures also include: In step S1, the solution to the DC distribution network optimization model with the objective of minimizing the distribution network operating cost is achieved using the virtual load iteration method. The specific process is as follows: Initialize parameters by setting the total distribution network loss, effective transmission factor, and virtual load to preset initial values; solve the DC distribution network optimization model under the current parameters to obtain the current adjustable power output value; Furthermore, the total distribution network loss, effective transmission factor, and virtual load are updated based on the obtained current adjustable power output value; the DC distribution network optimization model is re-solved based on the updated parameters to obtain a new adjustable power output value. The DC distribution network optimization model is repeatedly updated and its parameters are updated until the output value of the adjustable power source converges compared with the previous iteration result, at which point the solution ends.

[0008] Furthermore, in step S1, the DC distribution network optimization model takes the minimum operating cost of the distribution network as its objective function, and includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits of adjustable power supply output constraints, specifically:

[0009] st

[0010] ,

[0011] ,

[0012]

[0013]

[0014]

[0015] in, N This refers to the number of nodes in the distribution network system. c i For the first i The cost of adjustable power generation at nodes; For the first i Adjustable unit output at nodes; The number of photovoltaic units; DF i For the first i The effective transmission factor of injected power to each node; For the first i The load on the node; For the first j One photovoltaic power source output; P loss This represents the total losses in the distribution network. and The first i The minimum and maximum adjustable power output of the node; V min , V max These are the minimum and maximum values ​​of the node voltage, respectively. For transmission from the head node to the node i The collection of all routes through which the flow occurs; r k For the line k resistance; For nodes m To the line k The power transfer factor; For the first mAdjustable power output of the node; For the first m The load on the node; For the node m Virtual load; For the first Total losses in the distribution network at any given time; N l This refers to the number of distribution network lines. For the first i Node to Line k Power transfer factor; subscript i ( j ) indicates the first j The node where the photovoltaic power source is located i ; Subscript m ( j ) indicates the first j The node where the photovoltaic power source is located m ; For the first Time Node i Virtual load; For the first Time Node i Virtual load; For the first Time of the first i The effective transmission factor of injected power to each node; To connect at node i A collection of routes.

[0016] In step S2, the construction of a system of linear equations characterizing the marginal sensitivity of the unit output to changes in photovoltaic output specifically involves: The power balance equality constraint and the nodal voltage inequality constraint are rewritten in matrix equation form with relaxation variables:

[0017] Furthermore, the matrix equation can be simplified to the following form:

[0018] Solving the matrix equation yields the marginal variables. MP G , NV The expression for this is the system of linear equations for marginal sensitivity:

[0019] In the formula, , , , , , .

[0020] in, A The order is The square formation, The number of marginal units; The number of nodes with voltage as the upper limit; MP G For the output vector of the marginal unit, NV This is the relaxation vector for node voltages that have not exceeded their limits. NP G For non-marginal unit output vectors, CV This is the relaxation vector for node voltage exceeding the limit. P D For node load vectors, For photovoltaic power generation matrix, This is a virtual load matrix; for ; for ; for ; for ; for ; for The elements of each preset coefficient matrix are determined by the network parameters and the current running point.

[0021] In step S3, the specific process of solving for the allowable fluctuation range of photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged is as follows: The total change in photovoltaic output in the distribution network is allocated to each photovoltaic power source according to a proportional coefficient:

[0022] The sensitivity relationship between the change in photovoltaic power output and the change in marginal variables is as follows:

[0023]

[0024] Furthermore, the sensitivity of the marginal variable to changes in photovoltaic output is:

[0025]

[0026] With the constraints of ensuring that the output of marginal units does not exceed the limit and that the voltage of nodes that have not exceeded the limit does not exceed the limit, the feasible region for adjusting the photovoltaic output is determined as the operating region:

[0027] in, This represents the change in total photovoltaic power output; This is the proportionality coefficient between the predicted output values ​​of each photovoltaic power source; For marginal units i Maximum output; The matrix representing the output variation of marginal units; This is the matrix of node voltage changes for nodes with no voltage upper limit. For marginal units i Efforts made at the present moment; Nodes with no voltage upper limit i Slack variables; for In the matrix and the marginal unit i Related elements, representing the first i Output distribution coefficient of each marginal unit; for In the matrix, the node with voltage as the upper limit i Related elements; for Elements related to marginal units; for Elements related to nodes with no upper limit on voltage; To Find the partial derivative.

[0028] In step S4, the specific process of determining whether there is an uncovered range in the photovoltaic output fluctuation range between the current operating domain and other operating domains is as follows: Determine whether the upper boundary value of the first execution domain is less than the lower boundary value of the second execution domain: + < -

[0029] Furthermore, determine whether the lower boundary value of the first operating domain is greater than the upper boundary value of the third operating domain: - > +

[0030] If any of the above conditions are met, it is determined that there is an uncovered region; in, For the total predicted power output of the photovoltaic unit; This represents the maximum downward variation in photovoltaic output allowed in the first operating domain. This represents the maximum upward variation in photovoltaic output allowed in the first operating domain. This represents the maximum total output of the photovoltaic unit. This represents the minimum total output of the photovoltaic unit. The maximum downward variation in photovoltaic output allowed in operating domain 2; This represents the maximum upward variation in photovoltaic output allowed in the third operating domain.

[0031] Furthermore, in step S5, the marginal sensitivity coefficient is a linear distribution coefficient, and the incremental output of the marginal unit satisfies the following linear relationship with the change in total photovoltaic output:

[0032] in, This represents the change in total photovoltaic power output. For the first i The output increment of each marginal unit For the first i The output distribution coefficient of each marginal unit is determined by the corresponding element in the coefficient matrix of the linear equation system of marginal sensitivity.

[0033] As another important technical solution, the present invention also provides a real-time safe and economical dispatching system for DC distribution networks based on marginal sensitivity, comprising: The benchmark point and boundary scenario calculation module is used to obtain the load forecast and photovoltaic output forecast values ​​within a future set time period; based on the load forecast and photovoltaic output forecast values, it solves the DC distribution network optimization model with the goal of minimizing the distribution network operating cost, obtains the benchmark operating point of adjustable power supply output, and the scheduling results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limit constraints of adjustable power supply output; The marginal sensitivity linear modeling module is used to identify adjustable power sources whose output has not reached the limit as a set of marginal units, and to identify the constraints that the node voltage reaches the limit as a set of boundary constraints, based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario. Based on the identified set of marginal units and the set of boundary constraints, a set of marginal sensitivity linear equations characterizing the marginal unit output changes with the photovoltaic output is constructed. The single-point operating domain analysis module is used to solve the range of allowable fluctuations in photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged, based on the marginal sensitivity linear equation system, and to serve as the operating domain corresponding to each benchmark operating point; The global operating domain sequence construction module is used to determine whether there are uncovered intervals between the current operating domain and other operating domains in the photovoltaic power output fluctuation range. If so, the boundary point of the uncovered interval is used as the new operating point, and the marginal unit set and boundary constraint set corresponding to the new operating point are re-identified. The corresponding marginal sensitivity linear equation system is constructed, and the calculation is repeated until a continuous operating domain sequence covering the entire photovoltaic power output fluctuation range is obtained. The marginal sensitivity coefficient corresponding to each operating domain is extracted and stored. The real-time marginal scheduling decision module is used to obtain the actual photovoltaic output value of the DC distribution network during the real-time operation phase, determine the current operating domain of the photovoltaic power generation based on the actual photovoltaic output value, call the corresponding marginal sensitivity coefficient, calculate the output adjustment amount of each marginal unit, and execute the scheduling.

[0034] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity as described above.

[0035] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity, as described above.

[0036] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a continuous operating domain sequence covering the entire fluctuation range of photovoltaic power output offline and extracts the marginal sensitivity coefficient corresponding to each operating domain. In real-time scheduling, it is only necessary to locate the operating domain according to the actual photovoltaic power output and call the linear allocation coefficient to quickly generate an economical scheduling scheme that meets voltage safety constraints. This effectively solves the problem that existing methods cannot simultaneously guarantee safety and economy.

[0037] This invention transfers a large number of complex optimal power flow calculations to the offline stage. The online stage only performs running domain matching and linear multiplication-addition operations, which takes very little time and can meet the real-time scheduling requirements at the second or minute level. It overcomes the shortcomings of traditional deterministic optimization methods, which are time-consuming to solve, and stochastic optimization and robust optimization methods, which have excessive computational burden.

[0038] This invention introduces the concept of operating domain, predetermines the effective range of the marginal sensitivity coefficient during the offline phase, and achieves seamless connection between different operating domains. This avoids the sensitivity failure problem caused by the benchmark operating point crossing the safety boundary in the traditional marginal sensitivity method, and ensures the effectiveness and accuracy of the scheduling strategy within the entire photovoltaic power output fluctuation range.

[0039] This invention employs a DC distribution network optimization model that considers line losses in the solution of the baseline operating point, and accurately accounts for the impact of network losses through the virtual load iteration method. While ensuring the accuracy of the model, it achieves the transformation from a high-precision model to an efficient real-time scheduling model by combining the operating domain sequence and marginal sensitivity, thus balancing the accuracy and speed of scheduling decisions. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the process in the implementation of the present invention.

[0041] Figure 2 This is a schematic diagram of the effective safe and economical operating domain and the most economical operating trajectory within the maximum range of photovoltaic fluctuations provided in the implementation of this invention.

[0042] Figure 3 This is a node DC distribution network topology diagram provided in the implementation of this invention.

[0043] Figure 4 These are the load curves and photovoltaic output curves provided in the implementation of this invention.

[0044] Figure 5 This is a comparison diagram of the node voltages obtained by the virtual load iteration method and the second-order cone relaxation method provided in this invention.

[0045] Figure 6 This is a comparison chart of line losses obtained by the virtual load iteration method and the second-order cone relaxation method provided in this invention.

[0046] Figure 7 This is a comparison chart of the operating costs of three scheduling methods in real-time operation provided in the implementation of this invention.

[0047] Figure 8 This is a comparison diagram of node voltages during real-time operation of marginal scheduling and robust scheduling provided in the implementation of this invention. Detailed Implementation

[0048] The present invention will be further described in detail below through specific embodiments, but it should not be construed as limiting the scope of the subject matter of the present invention to the following embodiments. All technologies implemented based on the above content of the present invention fall within the scope of the present invention.

[0049] In some implementations, such as Figure 1 As shown, this invention provides a real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity, comprising the following steps: S1: Obtain the load forecast and photovoltaic output forecast for a future set time period from the dispatch and control center of the distribution network; based on the load forecast and photovoltaic output forecast, solve the DC distribution network optimization model with the goal of minimizing the operating cost of the distribution network, obtain the benchmark operating point of the adjustable power supply output, and the dispatch results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits constraints of adjustable power supply output; Solving the DC distribution network optimization model with the objective of minimizing distribution network operating costs employs the virtual load iteration method. The specific process is as follows: Initialize parameters by setting the total distribution network loss, effective transmission factor, and virtual load to preset initial values; solve the DC distribution network optimization model under the current parameters to obtain the current adjustable power output value; In some implementations, the total distribution network loss, effective transmission factor, and virtual load are updated based on the obtained current adjustable power output value; the DC distribution network optimization model is then re-solved based on the updated parameters to obtain a new adjustable power output value. The DC distribution network optimization model is repeatedly updated and its parameters are updated until the output value of the adjustable power source converges compared with the previous iteration result, at which point the solution ends.

[0050] The DC distribution network optimization model takes minimizing the distribution network operating cost as its objective function and includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits of adjustable power source output constraints, specifically:

[0051] st

[0052] ,

[0053] ,

[0054]

[0055]

[0056]

[0057] in, N This refers to the number of nodes in the distribution network system. c i For the first i The cost of adjustable power generation at nodes; For the first i Adjustable unit output at nodes; The number of photovoltaic units;DF i For the first i The effective transmission factor of injected power to each node; For the first i The load on the node; For the first j One photovoltaic power source output; P loss This represents the total losses in the distribution network. and The first i The minimum and maximum adjustable power output of the node; V min , V max These are the minimum and maximum values ​​of the node voltage, respectively. For transmission from the head node to the node i The collection of all routes through which the flow occurs; r k For the line k resistance; For nodes m To the line k The power transfer factor; For the first m Adjustable power output of the node; For the first m The load on the node; For the node m Virtual load; For the first Total losses in the distribution network at any given time; N l This refers to the number of distribution network lines. For the first i Node to Line k Power transfer factor; subscript i ( j ) indicates the first j The node where the photovoltaic power source is located i ; Subscript m ( j ) indicates the first j The node where the photovoltaic power source is located m ; For the first Time Node i Virtual load; For the first Time Node i Virtual load; For the first Time of the first i The effective transmission factor of injected power to each node; To connect at node iA collection of routes.

[0058] S2: Based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario, the adjustable power sources whose output has not reached the limit are identified as the marginal unit set, and the constraints that the node voltage reaches the limit are identified as the boundary constraint set; based on the identified marginal unit set and boundary constraint set, a set of linear equations characterizing the marginal sensitivity of the marginal unit output to the change in photovoltaic output is constructed. In some implementations, based on the scheduling results of the scenarios with maximum and minimum photovoltaic output, a set of marginal sensitivity linear equations is constructed for each benchmark operating point. This set of marginal sensitivity linear equations is a linear equation system transformed from the constraints acting at the optimal solution, used to determine the set of marginal generating units, the set of boundary constraints, and the changes in marginal generating unit output and total photovoltaic output at the current benchmark operating point. The analytical relationship between them.

[0059] The marginal unit set includes adjustable power sources whose output at the current optimal solution has not reached its upper or lower limit, and the boundary constraint set includes node voltage constraints or line transmission constraints that have reached the boundary at the current optimal solution. The linear relationship of the marginal sensitivity linear equation system is used to describe the marginal sensitivity of the unit output to changes in photovoltaic output, specifically: In some implementations, the power balance equality constraints and nodal voltage inequality constraints are rewritten as matrix equations with relaxation variables:

[0060] in, express of DF vector; = This indicates that the voltage-not-over-limit node is related to the non-marginalized unit. matrix; = This indicates that the voltage exceeding the upper limit node is related to non-marginalized units. matrix; This indicates the maximum allowable voltage variation at a node (where the voltage is not exceeded). Vector, element value V max - V 0; This indicates the maximum allowable voltage variation at the node (voltage limit exceeded). Vector, element value V max - V 0; express of DF vector; This indicates that the node whose voltage has not exceeded the upper limit is related to all nodes. matrix; This indicates that the node whose voltage exceeds the upper limit is related to all nodes. matrix; express of DF vector; This indicates that the voltage has no upper limit for the node related to the photovoltaic unit. Matrix; This indicates that the voltage exceeding the upper limit node is related to the marginal unit. Matrix; express of DF vector; This indicates that the voltage exceeding the upper limit node is related to the marginal unit. Matrix; This indicates that the voltage has not exceeded the upper limit node for marginal units. of matrix; 1 represents the identity matrix; 0 represents the zero matrix; 1 represents the identity matrix; N express A vector of 1.

[0061] The matrix equation can be simplified to the following form:

[0062] Multiply both sides of the equation by A -1 Solving the matrix equation yields the marginal variables. MP G , NV The expression for this is the system of linear equations for marginal sensitivity:

[0063] In the formula, , , , , , ; in, A The order is The square formation, The number of marginal units; The number of nodes with voltage as the upper limit; MP G For the output vector of the marginal unit, NVThis is the relaxation vector for node voltages that have not exceeded their limits. NP G For non-marginal unit output vectors, CV This is the relaxation vector for node voltage exceeding the limit. P D For node load vectors, For photovoltaic power generation matrix, This is a virtual load matrix; for ; for ; for ; for ; for ; for The elements of each preset coefficient matrix are determined by the network parameters and the current baseline operating point.

[0064] S3: Based on the linear equations of marginal sensitivity, solve for the range of photovoltaic output fluctuations that are allowed when the current set of marginal units and the set of boundary constraints remain unchanged. This range serves as the effective safe and economical operating domain for each benchmark operating point. The effective safe and economical operating domain refers to the continuous range in which photovoltaic output is allowed to be adjusted while ensuring system voltage safety and economical operation, provided that the current set of marginal units and the set of boundary constraints remain unchanged.

[0065] The process of determining the allowable fluctuation range of photovoltaic output while keeping the current set of marginal units and the set of boundary constraints constant is as follows: The total change in photovoltaic output in the distribution network is allocated to each photovoltaic power source according to a proportional coefficient:

[0066] The sensitivity relationship between the change in photovoltaic power output and the change in marginal variables is as follows:

[0067]

[0068] The sensitivity of marginal variables to changes in photovoltaic power output is:

[0069]

[0070] In some implementations, the feasible domain for adjusting photovoltaic output is determined by constraints such as ensuring that the output of marginal units does not exceed limits and that the voltage of nodes that have not exceeded limits does not exceed limits. This domain serves as the effective, safe, and economical operating domain.

[0071] in, This represents the change in total photovoltaic power output; This is the proportionality coefficient between the predicted output values ​​of each photovoltaic power source; For marginal units i Maximum output; The matrix representing the output variation of marginal units; This is the matrix of node voltage changes for nodes with no voltage upper limit. For marginal units i Efforts made at the present moment; Nodes with no voltage upper limit i Slack variables; for In the matrix and the marginal unit i Related elements; for In the matrix, the node with voltage as the upper limit i Related elements. for Elements related to marginal units; for Elements related to nodes with no upper limit on voltage; To Find the partial derivative.

[0072] S4: Determine whether there is an uncovered interval between the current operating domain and other operating domains in the photovoltaic power output fluctuation range; if so, take the boundary point of the uncovered interval as the new operating point, re-identify the marginal unit set and boundary constraint set corresponding to the new operating point, construct the corresponding marginal sensitivity linear equation system, and repeat steps S3 to S4 until a continuous operating domain sequence covering the entire photovoltaic power output fluctuation range is obtained; extract and store the marginal sensitivity coefficient corresponding to each operating domain; The specific process for determining whether there is an uncovered range in the photovoltaic output fluctuation range between the current effective safe and economical operating domain and other effective safe and economical operating domains is as follows: Determine whether the upper boundary value of the first effective safe and economical operating domain is less than the lower boundary value of the second effective safe and economical operating domain: + < -

[0073] Determine whether the lower boundary value of the first effective safe and economical operating domain is greater than the upper boundary value of the third effective safe and economical operating domain: - > +

[0074] If any of the above conditions are met, it is determined that there is an uncovered region; in, For the total predicted power output of the photovoltaic unit; The maximum downward variation in photovoltaic output allowed in the first effective, safe, and economical operating domain; The maximum upward change in photovoltaic output allowed in the first effective, safe, and economical operating domain; This represents the maximum total output of the photovoltaic unit. This represents the minimum total output of the photovoltaic unit. The maximum downward variation in photovoltaic output allowed in Domain 2 for effective, safe, and economical operation; This represents the maximum upward variation in photovoltaic output allowed in the third effective, safe, and economical operating domain.

[0075] S5: During the real-time operation phase, the total power and total load of the distribution network are compared in real time to obtain the power imbalance. Based on the power imbalance, it is determined whether the power fluctuation exceeds the effective safe and economical operating range.

[0076] The actual photovoltaic output value of the DC distribution network is obtained, and its current effective safe and economical operating range is determined based on the actual photovoltaic output value. The real-time photovoltaic output value is compared with the predicted value to obtain the change in photovoltaic output and to determine whether the current power fluctuation exceeds the effective safe and economical operating range.

[0077] In some implementations, the marginal sensitivity coefficient is a linear distribution coefficient, and the incremental output of the marginal unit satisfies the following linear relationship with the change in total photovoltaic output:

[0078] in, For the first i The output increment of each marginal unit For the first i The output distribution coefficient of each marginal unit is determined by the corresponding element in the coefficient matrix of the linear equation system of marginal sensitivity.

[0079] In some implementations, if the current 5-minute scheduling cycle has not yet ended (for example, there are still unfinished sampling moments within the cycle), the process returns to step S5 to continue obtaining the actual photovoltaic output value at the next moment, repeating the domain matching, marginal sensitivity coefficient call and unit output adjustment until the end of the cycle. If the current 5-minute scheduling cycle has ended, exit the real-time scheduling loop and enter the pre-scheduling stage of the next scheduling cycle, that is, return to step S1, re-obtain the load forecast value and photovoltaic output forecast value for the future time period, and execute the entire process of offline operation domain construction and online marginal scheduling again.

[0080] Through this periodic determination mechanism, the present invention achieves multi-period rolling optimization scheduling, ensuring that the running domain sequence can be updated based on the latest prediction information at the beginning of each scheduling cycle, while relying only on pre-calculated linear coefficients for rapid response within the cycle, thus balancing global economy and real-time speed.

[0081] Example 1 The verification is carried out using a 33-node DC distribution network system as an example. The topology of the 33-node DC distribution network is as follows: Figure 3 As shown.

[0082] The topology includes: Node 1 at the top end is connected to the upstream grid via a converter station; adjustable power sources are connected to Node 18 (CHP unit, capacity 4MW) and Node 32 (CHP unit, capacity 2MW); photovoltaic power is connected to Node 33 (capacity 3MW); the converter station has a capacity of 4MW. The electricity purchase price is 1 yuan / kWh, and the power generation costs for Nodes 18 and 32 are 0.5 yuan / kWh and 1.5 yuan / kWh, respectively.

[0083] The scheduling cycle is 24 hours, with a step size of 1 hour. The system's daily total load curve and photovoltaic output prediction curve are as follows: Figure 4 As shown, from Figure 4 It can be seen that the peak load period is concentrated between 11:00 and 14:00, which basically coincides with the peak output period of photovoltaic power. This provides a typical scenario for verifying the scheduling effect of the method of the present invention during peak hours.

[0084] The virtual load iteration method is used to solve the problem, and the accuracy of the linear optimal power flow model considering loss and voltage constraints adopted in this invention is verified by comparison with the second-order cone relaxation method, which is recognized as having higher accuracy.

[0085] like Figure 5~Figure 6 As shown, the node voltages and line losses obtained by the virtual load iteration method are in high agreement with the results of the second-order cone relaxation method, with a very small maximum error; Figure 5 The voltage curves of nodes 18#, 28#, and 33# were compared under the two methods. The maximum difference in voltage between nodes 18# and 33#, which are located at the end of the line, during the peak load and photovoltaic output period (11:00-14:00) was only 0.17%. Figure 6 The comparison of line loss curves for branches #2, #5, and #30 under the two methods is shown, with the maximum difference being 2% of the total loss. This verifies that the benchmark optimization model used in this invention is accurate and can provide a reliable benchmark for the subsequent construction of an effective, safe, and economical operating domain.

[0086] Taking a specific 5-minute scheduling cycle as an example, the construction process of the effective, safe, and economical operation domain is explained in detail, and the construction result is as follows: Figure 2As shown: the horizontal axis represents the change in photovoltaic power output, and the vertical axis represents the marginal unit output; point A is the baseline operating point obtained from the predicted values, and points B and C are the extreme baseline operating points under the scenarios of maximum and minimum photovoltaic power output, respectively; the intervals γ1, γ2...γ7 are the constructed continuous effective safe and economical operating domain sequence, and the line segments A-A' to A'-A'' are the most economical operating trajectories of the system within each effective safe and economical operating domain. Figure 2 It can be seen that each effective safe and economical operation domain is continuously covered on the photovoltaic power output change axis, and the operation trajectory within each domain is a straight line, indicating that the marginal unit output and the photovoltaic power output change are linearly related within this interval.

[0087] Based on the current 5-minute load and photovoltaic forecast, the optimization model is solved to obtain the adjustable power supply output baseline value (corresponding to...). Figure 2 Point A in the equation. Simultaneously, the maximum and minimum photovoltaic scenarios are solved to obtain points B and C.

[0088] Taking point A as an example, analyze the optimization results at this point to identify marginal units (such as node 18, CHP) and effective constraints (such as a node voltage reaching its upper limit). Establish a system of linear equations for the marginal sensitivity at this baseline operating point.

[0089] Solving for the effective safe and economical operating domain γ1: Based on the marginal equation, the range of allowable adjustment of photovoltaic output is derived under the premise that the marginal unit set and the operating constraints remain unchanged. Figure 2 The interval in [ , Within this interval, the system's trajectory is a straight line A-A'.

[0090] Calculation of Boundary and Extension: Calculating the boundary value of γ1 reveals that when the photovoltaic output exceeds... When new constraints come into effect, the original marginal relationships are changed; at this time, the marginal units are re-identified with boundary point A' as the new base point, a new marginal equation is established, and the next effective safe and economical operating domain γ2 is solved.

[0091] Repeat the above process until the constructed effective, safe, and economical operating domain covers the entire possible fluctuation range of photovoltaic output. , ], ultimately resulting in Figure 2 The sequence of continuous, effective, safe, and economical operating domains γ1, γ2, ..., γ7 is shown, along with the linear distribution coefficient matrix corresponding to each domain.

[0092] In the following 5-minute real-time operation phase, 60 output scenarios were randomly generated in 5-second increments within a fluctuation range of ±20% in photovoltaic output to simulate real-time fluctuations. At each sampling point: Marginal scheduling (method of this invention): Detect the current photovoltaic output value and determine if it falls within the range. Figure 2 Which effective safe and economical operation domain is selected? Directly call the pre-calculated allocation coefficient of that domain to calculate the output adjustment of each unit.

[0093] Comparison method: Two-stage robust scheduling and second-order cone relaxation scheduling (as the economically optimal benchmark) are compared.

[0094] like Figure 7 As shown, the cost curve of marginal scheduling is basically consistent with that of second-order cone scheduling, and is far superior to robust scheduling. Figure 7 In the figure, the horizontal axis represents the real-time running time (60 sampling points within 5 minutes), and the vertical axis represents the system operating cost. The marginal scheduling curve almost coincides with the second-order cone scheduling curve, while the robust scheduling curve is significantly higher. The method of this invention successfully reproduced the economically optimal trajectory of online optimization in real-time operation.

[0095] like Figure 8 As shown, under marginal scheduling, the node voltage can operate closer to the upper voltage limit, making full use of the voltage margin to improve economy, and no limit overruns have occurred. Figure 8 The diagram shows the voltage variation curves of node 18 (the highest voltage node) and node 25 (the lowest voltage node) under two scheduling schemes. Under the marginal scheduling scheme, the voltage of node 18 always operates close to the upper voltage limit, while under the robust scheduling scheme, the voltage of node 18 is still far from the upper voltage limit. This verifies that the method of the present invention, through the constraint of an effective, safe, and economical operating domain, strictly guarantees voltage safety while pursuing economy.

[0096] In another embodiment of the present invention, a real-time safe and economical dispatching system for DC distribution networks based on marginal sensitivity is proposed, comprising: The benchmark point and boundary scenario calculation module is used to obtain the load forecast and photovoltaic output forecast values ​​within a future set time period; based on the load forecast and photovoltaic output forecast values, it solves the DC distribution network optimization model with the goal of minimizing the distribution network operating cost, obtains the benchmark operating point of adjustable power supply output, and the scheduling results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limit constraints of adjustable power supply output; The marginal sensitivity linear modeling module is used to identify adjustable power sources whose output has not reached the limit as a set of marginal units, and to identify the constraints that the node voltage reaches the limit as a set of boundary constraints, based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario. Based on the identified set of marginal units and the set of boundary constraints, a set of marginal sensitivity linear equations characterizing the marginal unit output changes with the photovoltaic output is constructed. The single-point operating domain analysis module is used to solve the range of allowable fluctuations in photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged, based on the marginal sensitivity linear equation system, and to serve as the operating domain corresponding to each benchmark operating point; The global operating domain sequence construction module is used to determine whether there are uncovered intervals between the current operating domain and other operating domains in the photovoltaic power output fluctuation range. If so, the boundary point of the uncovered interval is used as the new operating point, and the marginal unit set and boundary constraint set corresponding to the new operating point are re-identified. The corresponding marginal sensitivity linear equation system is constructed, and the calculation is repeated until a continuous operating domain sequence covering the entire photovoltaic power output fluctuation range is obtained. The marginal sensitivity coefficient corresponding to each operating domain is extracted and stored. The real-time marginal scheduling decision module is used to obtain the actual photovoltaic output value of the DC distribution network during the real-time operation phase, determine the current operating domain of the photovoltaic power generation based on the actual photovoltaic output value, call the corresponding marginal sensitivity coefficient, calculate the output adjustment amount of each marginal unit, and execute the scheduling.

[0097] In another embodiment of the present invention, an electronic device is proposed, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity as described above.

[0098] In another embodiment of the present invention, a computer-readable storage medium is provided storing a computer program that causes a computer to execute a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity as described above.

[0099] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent substitutions, and improvements made by those skilled in the art to the above embodiments without departing from the scope of the technical solution of the present invention, based on the technical essence of the present invention, shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity, characterized in that, Includes the following steps: S1: Obtain the load forecast and photovoltaic output forecast for a future set time period; based on the load forecast and photovoltaic output forecast, solve the DC distribution network optimization model with the goal of minimizing the distribution network operating cost, obtain the benchmark operating point of adjustable power supply output, and the scheduling results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits constraints of adjustable power supply output; S2: Based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario, the adjustable power sources whose output has not reached the limit are identified as the marginal unit set, and the constraints that the node voltage reaches the limit are identified as the boundary constraint set; based on the identified marginal unit set and boundary constraint set, a set of linear equations characterizing the marginal sensitivity of the marginal unit output to the change in photovoltaic output is constructed. S3: Based on the linear equations of marginal sensitivity, solve for the range of allowable fluctuations in photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged, and use it as the operating domain corresponding to each benchmark operating point; S4: Determine whether there are any uncovered intervals in the photovoltaic power output fluctuation range between the current operating domain and other operating domains; If it exists, the boundary point of the uncovered interval is taken as the new operating point. The set of marginal units and the set of boundary constraints corresponding to the new operating point are re-identified, and the corresponding set of marginal sensitivity linear equations is constructed. Steps S3 to S4 are repeated until a continuous operating domain sequence covering the entire fluctuation range of photovoltaic power output is obtained. The marginal sensitivity coefficient corresponding to each operating domain is extracted and stored. S5: During the real-time operation phase, obtain the actual photovoltaic output value of the DC distribution network, determine its current operating domain based on the actual photovoltaic output value, call the corresponding marginal sensitivity coefficient, calculate the output adjustment amount of each marginal unit, and execute the scheduling.

2. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 1, characterized in that: In step S1, the solution to the DC distribution network optimization model with the objective of minimizing the distribution network operating cost is achieved using the virtual load iteration method. The specific process is as follows: Initialize parameters by setting the total distribution network loss, effective transmission factor, and virtual load to preset initial values; solve the DC distribution network optimization model under the current parameters to obtain the current adjustable power output value; Update the total distribution network loss, effective transmission factor, and virtual load based on the obtained current adjustable power output value; re-solve the DC distribution network optimization model based on the updated parameters to obtain a new adjustable power output value; The DC distribution network optimization model is repeatedly updated and its parameters are updated until the output value of the adjustable power source converges compared with the previous iteration result, at which point the solution ends.

3. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 2, characterized in that: In step S1, the DC distribution network optimization model takes the minimum operating cost of the distribution network as the objective function, and includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits of adjustable power supply output constraints, specifically: s.t. ， , in, N This refers to the number of nodes in the distribution network system. c i for i The cost of adjustable power generation at nodes; for i Adjustable unit output at nodes; The number of photovoltaic units; DF i For the first i The effective transmission factor of injected power at each node, subscript i ( j ) indicates the first j The node where the photovoltaic power source is located i ; for i The load on the node; For the first j The output of a single photovoltaic power source, subscript m ( j ) indicates the first j The node where the photovoltaic power source is located m ; P loss This represents the total losses in the distribution network. and They are nodes i The minimum and maximum output values ​​of the adjustable power supply; V min , V max These are the minimum and maximum values ​​of the node voltage, respectively. For transmission from the head node to the node i The collection of all routes through which the flow occurs; r k For the line k resistance; For nodes m To the line k The power transfer factor; For nodes m Adjustable power output; for m The load on the node; For nodes m Virtual load.

4. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 1, characterized in that: In step S2, the construction of a system of linear equations characterizing the marginal sensitivity of the unit output to changes in photovoltaic output specifically involves: in, A Is the order of The square formation, The number of marginal units; The number of nodes with voltage as the upper limit; MP G For the output vector of the marginal unit, NV This is the relaxation vector for node voltages that have not exceeded their limits. NP G For non-marginal unit output vectors, CV This is the relaxation vector for node voltage exceeding the limit. P D For node load vectors, For photovoltaic power generation matrix, This is a virtual load matrix; , , , , and This is a preset coefficient matrix.

5. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 1, characterized in that: In step S3, the specific process of solving for the allowable fluctuation range of photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged is as follows: The total change in photovoltaic output in the distribution network is allocated to each photovoltaic power source according to a proportional coefficient: The sensitivity relationship between the change in photovoltaic power output and the change in marginal variables is as follows: With the constraints of ensuring that the output of marginal units does not exceed the limit and that the voltage of nodes that have not exceeded the limit does not exceed the limit, the feasible region for adjusting the photovoltaic output is determined as the operating region: in, This represents the change in total photovoltaic power output; This is the proportionality coefficient between the predicted output values ​​of each photovoltaic power source; For marginal units i Maximum output; The matrix representing the output variation of marginal units; This is the matrix of node voltage changes for nodes with no voltage upper limit. For marginal units i Efforts made at the present moment; Nodes with no voltage upper limit i Slack variables; for In the matrix and the marginal unit i Related elements, representing For the first i Output distribution coefficient of each marginal unit; for In the matrix, the node with voltage as the upper limit i Related elements.

6. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 1, characterized in that: In step S4, the specific process of determining whether there is an uncovered range in the photovoltaic output fluctuation range between the current operating domain and other operating domains is as follows: Determine whether the upper boundary value of the first execution domain is less than the lower boundary value of the second execution domain: + < - Determine whether the lower boundary value of the first execution domain is greater than the upper boundary value of the third execution domain: - > + If any of the above conditions are met, it is determined that there is an uncovered region; in, For the total predicted power output of the photovoltaic unit; This represents the maximum downward variation in photovoltaic output allowed in the first operating domain. This represents the maximum upward variation in photovoltaic output allowed in the first operating domain. This represents the maximum total output of the photovoltaic unit. This represents the minimum total output of the photovoltaic unit. The maximum downward variation in photovoltaic output allowed in operating domain 2; This represents the maximum upward variation in photovoltaic output allowed in the third operating domain.

7. The real-time safe and economical dispatch method for DC distribution networks based on marginal sensitivity according to claim 1, characterized in that: In step S5, the marginal sensitivity coefficient is a linear distribution coefficient, and the incremental output of the marginal unit satisfies the following linear relationship with the change in total photovoltaic output: in, This represents the change in total photovoltaic power output. For the first i The output increment of each marginal unit For the first i Output distribution coefficient of each marginal unit; This refers to the number of marginal units.

8. A real-time safe and economical dispatching system for DC distribution networks based on marginal sensitivity, characterized in that, include: The benchmark point and boundary scenario calculation module is used to obtain the load forecast and photovoltaic output forecast values ​​within a future set time period; Based on the load forecast and photovoltaic output forecast, a DC distribution network optimization model with the goal of minimizing the distribution network operating cost is solved to obtain the benchmark operating point of adjustable power source output, as well as the scheduling results under the scenarios of maximum and minimum photovoltaic output; wherein, the DC distribution network optimization model includes power balance constraints, line loss constraints, node voltage constraints, and upper and lower limits constraints of adjustable power source output. The marginal sensitivity linear modeling module is used to identify adjustable power sources whose output has not reached the limit as a set of marginal units, and to identify the constraints that the node voltage reaches the limit as a set of boundary constraints, based on the baseline operating point, the scheduling results of the maximum photovoltaic output scenario, and the scheduling results of the minimum photovoltaic output scenario. Based on the identified set of marginal units and the set of boundary constraints, a set of marginal sensitivity linear equations characterizing the marginal unit output changes with the photovoltaic output is constructed. The single-point operating domain analysis module is used to solve the range of allowable fluctuations in photovoltaic output when the current set of marginal units and the set of boundary constraints remain unchanged, based on the marginal sensitivity linear equation system, and to serve as the operating domain corresponding to each benchmark operating point; The global operating domain sequence construction module is used to determine whether there are uncovered intervals between the current operating domain and other operating domains in the photovoltaic power output fluctuation range. If so, the boundary point of the uncovered interval is used as the new operating point, and the marginal unit set and boundary constraint set corresponding to the new operating point are re-identified. The corresponding marginal sensitivity linear equation system is constructed, and the calculation is repeated until a continuous operating domain sequence covering the entire photovoltaic power output fluctuation range is obtained. The marginal sensitivity coefficient corresponding to each operating domain is extracted and stored. The real-time marginal scheduling decision module is used to obtain the actual photovoltaic output value of the DC distribution network during the real-time operation phase, determine the current operating domain of the photovoltaic power generation based on the actual photovoltaic output value, call the corresponding marginal sensitivity coefficient, calculate the output adjustment amount of each marginal unit, and execute the scheduling.

9. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program causes the computer to execute a real-time safe and economical dispatching method for DC distribution networks based on marginal sensitivity as described in any one of claims 1 to 7.

Images