Spatial clustering unit commitment problem solving method and system based on power distribution factor

Abstract

The invention discloses a spatial clustering unit commitment problem solving method and system based on a power distribution factor, and belongs to the technical field of power operation, and the method comprises the steps: constructing a network constraint unit commitment model, setting a target function to minimize the total cost of a system, and setting a constraint condition; performing spatial clustering unit preliminary clustering on the physical parameters by applying a spatial clustering method, improving constraint conditions according to a preliminary clustering result, and introducing connection constraints; the contribution degrees of all thermal power generating units to load nodes are evaluated by calculating power distribution factors, and units of the remaining units in the same cluster with the power distribution factors larger than a deviation threshold value are removed. The method solves the problem that clustering is carried out only according to unit physical parameters in the prior art, but the operation condition of the unit is not considered. Example tests show that the algorithm can effectively improve the calculation efficiency of the network constraint unit commitment problem, and a high-quality result is obtained.

Description

Technical Field

[0001] This invention relates to the field of power operation technology, specifically to a method and system for solving the spatial clustering unit combination problem based on power distribution factor. Background Technology

[0002] The new power system revolution presents new requirements and challenges for power system planning research. The increasing number of new participants, such as renewable energy, demand response, and FACTS devices, will bring greater computational complexity to power system operation. Therefore, it is necessary to consider the specific details of power system operation during the planning stage. In particular, the increasing penetration rate of renewable energy in the power system brings significant uncertainty and variability, thereby increasing the demand for operational flexibility. To achieve high renewable energy penetration, detailed constraints related to operational flexibility need to be considered in power system planning models, such as capacity reserve requirements, unit ramping requirements, and transmission capacity limitations. Ignoring these constraints during the planning process may lead to infeasible or uneconomical planning schemes, resulting in a lack of operational flexibility. The Network Constrained Unit Composition (NCUC) model is widely used in short-term power system operation problems and is considered capable of capturing all operational details of the power system. However, NCUC models typically involve a large number of binary variables representing unit start-up and shutdown states, thus imposing a heavy computational burden when modeling medium- and long-term power system operation in planning models. Therefore, many studies have focused on developing simplified and accurate NCUC models.

[0003] As a mixed-integer programming problem, NCUC's main computational burden stems from integer variables. Most methods aimed at reducing computational complexity focus on reducing the number of binary variables; this technique is known as spatial clustering. For example, similar units are grouped into the same category, allowing integer variables to represent the state of all units in the cluster. To account for the impact of virtual nodes on network topology after unit clustering, some researchers have proposed a method that preserves the original topology while spatially clustering unit combinations. Furthermore, for a more refined clustering mechanism, some researchers have proposed power flow-based clustering methods to improve clustering performance under specific load conditions. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is to propose an NCUC model based on power distribution factor for unit clustering, which meets the requirements of unit cluster schemes that may vary in different time periods. Based on the power distribution factor under different power flow sections in each time period, the unit cluster is further divided, achieving a good balance between computational speed and solution quality.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for solving the unit combination problem based on spatial clustering of power distribution factor, which includes the following steps:

[0007] Construct a network-constrained unit combination model, set an objective function to minimize the total system cost, and define the constraints.

[0008] Spatial clustering method is applied to perform spatial clustering of physical parameters to initially cluster the units. Based on the preliminary clustering results, the constraints are improved and connection constraints are introduced.

[0009] By calculating the power distribution factor, the contribution of all thermal power units to the load nodes is evaluated, and units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold are removed.

[0010] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the network-constrained unit combination model includes setting an objective function to minimize the total system cost, including the operating cost of thermal power units, start-up and shutdown costs, penalty costs for wind and solar curtailment, and penalty costs for load shedding.

[0011] minCR+CS+CW+CE

[0012] Wherein, CR represents the operating cost of the thermal power unit, CS represents the start-up and shutdown cost of the thermal power unit, CW represents the penalty cost of curtailing wind and solar power, and CD represents the penalty cost of load shedding.

[0013] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the operating cost of the thermal power unit includes: the operating cost of the thermal power unit adopts a piecewise linear method to linearize the nonlinear objective, divides the operating cost into h segments, and transforms the total cost of the thermal power unit:

[0014]

[0015] Where T and G represent the total time period and the number of generating units, respectively; g and t represent the index markers of the generating units and time periods, respectively; i represents the index marker of the output interval after piecewise linearization; and x represents the index marker of the output interval after piecewise linearization. g,t This indicates the start-stop status of the thermal power unit g during time period t, represented by 0 or 1; Z represents the operating cost incurred by thermal power unit g when it operates at minimum output. g,i Let χ be the slope of the i-th segment after piecewise linearization of the operating cost of thermal power unit g. g,t,i Let g be the power output level of thermal power unit g in the i-th power output interval during time period t;

[0016] The formulas for calculating the piecewise linear parameters and variables mentioned above are as follows:

[0017]

[0018] in, and Let α represent the maximum and minimum output levels of the thermal power unit g, respectively. g ,β g and γ g These are the operating cost coefficients for thermal power unit g, and P. g,t The output level of the thermal power unit g during time period t;

[0019] The startup and shutdown costs of a thermal power unit are a linear function of the unit's startup and shutdown variables;

[0020] The penalty cost for curtailing wind and solar power is a linear function of the amount of wind and solar power curtailed.

[0021] The penalty cost for load shedding is a linear function of the amount of load shedding.

[0022] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the set constraint conditions include:

[0023] Actual output and load shedding constraints;

[0024] System constraints are applied by calculating the power flow between lines using DC power flow calculations.

[0025] The start-stop state constraints of the unit are determined by the start-stop state of the previous period and the start-up or shutdown operation of the current period, and the start-stop state constraints of the unit do not need to be met in the first period.

[0026] To establish ramp constraints, the output level of thermal power units shall not vary within a unit time period beyond the current physical limit of the unit. Furthermore, if a thermal power unit needs to start up in the next moment, its maximum output in the next moment shall be subject to additional restrictions. At the same time, if a thermal power unit needs to shut down in the next moment, its maximum output in the current moment shall be subject to additional restrictions.

[0027] Establish a minimum start-stop time limit.

[0028] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the improved constraint conditions include: in the combination of spatially clustered units, the index of the unit cluster replaces the index of a single unit, and the variable characterizing the unit start-up and shutdown state is transformed from a binary variable into an integer variable. The unit start-up and shutdown state constraint is improved as follows:

[0029]

[0030] Where c is the index flag of the generator cluster, t is the index flag of the time period, and x c,t , and These represent the state of unit cluster c during time period t, and the number of units started and shut down, respectively. This indicates the number of units contained in unit cluster c;

[0031] In the objective function and constraints, variables and parameters indexed by g are replaced with variables and parameters indexed by c. The output of the unit cluster is represented as follows:

[0032]

[0033] in, Let P be the set of all units contained in unit cluster c. g,t P represents the output level of unit g during time period t. c,t The output level of unit cluster c in time period t;

[0034] Since the start-up and shutdown status of individual units within the same type of generating unit cannot be determined, the following constraint is added to the actual output and load shedding constraints:

[0035]

[0036] in, This represents the maximum permissible output range of unit g;

[0037] The minimum downtime constraint for generating units needs to be modified because the variable changes from unit state to unit cluster state. Specifically, the number of units in cluster c that are in a downtime state during time period t should be greater than or equal to the sum of the number of units shut down during the previous minimum downtime periods in cluster c, as shown below:

[0038]

[0039] Among them, TU c and TD c These represent the minimum allowed continuous start-up and shutdown times for unit cluster c, respectively.

[0040] The introduced connection constraints describe the relationship between the state of a single unit and the start / stop state of its cluster, as shown below:

[0041]

[0042] Where g is the index flag of the unit, x g,t , and These represent the state of unit g during time period t, as well as the start-up and shutdown operations.

[0043] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the calculation of the power distribution factor includes,

[0044] Define the topology of the power network, including the connection relationships of nodes and tie branches, determine the distribution of generating units, record the active power output of the generating units, and convert the load into an equivalent impedance in the network, and collect the load data of each node.

[0045] Calculate the node voltage in the power grid, record the voltage distribution of each unit at its connection node, determine the power output of each unit in the network, calculate the power transmission of the unit at different nodes, and distinguish the contribution of each unit.

[0046] Determine the power transfer between branches, including the power flow from one node to another, and calculate the total power in the branch and the power contributed by each unit;

[0047] The power distribution factor of the computer group includes the power distribution factor of the unit on the tie branch and the power distribution factor of the unit on the node load;

[0048] Based on the power distribution factor of the unit, calculate the power allocated to each load, trace the source of the power provided by the unit, and verify whether the power allocated by the unit to each node load conforms to the overall power distribution relationship according to the principle of non-destructive testing.

[0049] Calculate the optimal DC power flow for the entire time period, obtain initial point data, and use AC power flow calculation to adjust the unit output to meet power flow section constraints.

[0050] As a preferred embodiment of the spatial clustering unit combination problem solution method based on power distribution factor described in this invention, the calculation of the power distribution factor further includes:

[0051] Calculate the optimal DC power flow throughout the entire time period and set the deviation threshold for the power distribution factor;

[0052] Based on the physical parameters of the units, all units are clustered into category C;

[0053] Using the active power output and load information of the units in time period t obtained from the DC optimal power flow, calculate the AC power flow section and the power distribution factor of the units in time period t.

[0054] Select the power distribution factor of the c-type unit cluster and calculate whether the maximum intra-class distance of the power distribution factor matrix of the current unit cluster exceeds the set deviation threshold.

[0055] If the maximum intra-cluster distance exceeds the set deviation threshold, split clustering in hierarchical clustering is used to continuously remove units that are far apart and generate a new cluster until the maximum intra-cluster distance is less than or equal to the set deviation threshold; otherwise, the original clustering result is maintained.

[0056] If c = C, then the corrected clustering results for time period t are obtained, and all units are clustered into C. t If the current cluster is of type c, then select the power distribution factor of the c-th type of unit cluster and calculate the maximum intra-class distance of the matrix of the power distribution factors of the current unit cluster.

[0057] If t = T, output the clustering results of all the units in the last time period; otherwise, reuse the active power output and load information of the units in time period t obtained from the DC optimal power flow to calculate the AC power flow section and the power distribution factor of the units in time period t, where T is the total number of time periods.

[0058] Another objective of this invention is to provide a system for solving spatial clustering unit combination problems based on power distribution factors.

[0059] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a spatial clustering unit combination problem solving system based on power distribution factor, comprising: a network-constrained unit combination construction module, a spatial clustering module, and a power distribution factor-based calculation module.

[0060] The network-constrained unit combination construction module builds a network-constrained unit combination model, sets an objective function to minimize the total system cost, and sets constraints.

[0061] The spatial clustering module applies spatial clustering methods to perform preliminary spatial clustering of physical parameters into unit clusters, improves constraints based on preliminary clustering results, and introduces connection constraints.

[0062] Based on the power distribution factor calculation module, the contribution of all thermal power units to the load nodes is evaluated by calculating the power distribution factor, and units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold are removed.

[0063] The present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method for solving the spatial clustering unit combination problem based on power distribution factor.

[0064] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method for solving the spatial clustering unit combination problem based on power distribution factor.

[0065] The beneficial effects of this invention are as follows: Power systems contain numerous thermal power units of similar models and performance. Clustering them together for optimization can improve the computational efficiency of the model, but it can also introduce errors due to misjudging the state of units within the cluster. Based on this, this patent proposes a spatial clustering algorithm that considers power distribution factors. This algorithm overcomes the problem of traditional clustering based solely on unit physical parameters, neglecting to consider unit operating conditions. Numerical examples demonstrate that this algorithm can effectively improve the computational efficiency of network-constrained unit combination problems and obtain high-quality results. Attached Figure Description

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating the overall process of a spatial clustering method for solving unit combination problems based on power distribution factor, provided in one embodiment of the present invention.

[0068] Figure 2 This invention provides a piecewise linearization of the operating cost of thermal power units using a spatial clustering method for solving unit combination problems based on power distribution factor, as an embodiment of the present invention.

[0069] Figure 3 A three-node system topology diagram for a spatial clustering unit combination problem solution method based on power distribution factor provided in an embodiment of the present invention.

[0070] Figure 4 The output level results obtained by different methods on a three-node system using a spatial clustering unit combination problem-solving method based on power distribution factor, as provided in an embodiment of the present invention.

[0071] Figure 5 A schematic diagram of current distribution for a spatial clustering unit combination problem solution method based on power distribution factor provided in an embodiment of the present invention.

[0072] Figure 6 The output level result obtained by the spatial clustering method based on power distribution factor for solving the unit combination problem provided in an embodiment of the present invention on a three-node system.

[0073] Figure 7 The output level result obtained by the spatial clustering method based on power distribution factor for solving the unit combination problem provided in an embodiment of the present invention on a three-node system.

[0074] Figure 8 This invention provides a solution method for the unit combination problem based on spatial clustering of power distribution factor in an embodiment of the present invention, which examines the line power flow error of the network-constrained unit combination problem under different spatial clustering methods in an IEEE-RTS-24 node region. Detailed Implementation

[0075] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0076] Example 1, referring to Figures 1-6 This is the first embodiment of the present invention, which provides a method for solving the spatial clustering basis set combinatorial problem based on power distribution factor, including:

[0077] S1: Construct a network-constrained unit combination model, set an objective function to minimize the total system cost, and set constraints.

[0078] As a crucial decision-making tool for ensuring the safe and stable operation of the power system, network-constrained unit combination, based on load forecasts and renewable energy output forecasts, plans the start-up and shutdown decisions and output levels of generating units under the objective of minimizing operating costs. The time resolution of the operation simulation is set to one hour. The model is expressed as a mixed-integer programming problem, with unit start-up and shutdown states as integer variables and unit output levels as continuous variables, using DC power flow. Constraints include minimum start-up and shutdown time constraints, line thermal limit constraints, and unit output constraints. Considering that the unit combination model after large-scale renewable energy grid integration differs somewhat from traditional unit combination models, and that the clustering technique used later will further influence the objective function and constraints, the unit combination model after renewable energy integration will be described in detail below.

[0079] S101: Define the objective function

[0080] The objective function for unit combination is generally to minimize the total system cost, including the operating cost of thermal power units, start-up and shutdown costs, penalty costs for curtailment of wind and solar power, and penalty costs for load shedding.

[0081] minCR+CS+CW+CE(1)

[0082] Wherein, CR represents the operating cost of the thermal power unit, CS represents the start-up and shutdown cost of the thermal power unit, CW represents the penalty cost of curtailing wind and solar power, and CD represents the penalty cost of load shedding.

[0083] The operating cost of a thermal power unit is a quadratic function of its output, which can be expressed as:

[0084]

[0085] Where T and G represent the total number of time periods and the number of generating units, respectively; g and t represent the index markers for generating units and time periods, respectively; and P represents the index markers for generating units and time periods, respectively. g,t Let α be the output level of the thermal power unit g during time period t. g ,β g and γ g denoted as the operating cost coefficient for thermal power unit g.

[0086] Since the operating cost of a thermal power unit is a quadratic function, it affects the computational efficiency of the final model. Therefore, in this invention, the operating cost of the thermal power unit is linearized using a piecewise linear method to linearize the nonlinear objective and improve computational efficiency. Figure 2 As shown,

[0087] Dividing the operating cost into h segments, the total cost of the thermal power unit is transformed from equation (2) to equation (3):

[0088]

[0089] Where T and G represent the total time period and the number of generating units, respectively; g and t represent the index markers of the generating units and time periods, respectively; i represents the index marker of the output interval after piecewise linearization; and x represents the index marker of the output interval after piecewise linearization. g,t This indicates the start-stop status of the thermal power unit g during time period t, represented by 0 or 1; Z represents the operating cost incurred by thermal power unit g when it operates at minimum output. g,i Let χ be the slope of the i-th segment after piecewise linearization of the operating cost of thermal power unit g. g,t,i Let g be the power output level of thermal power unit g in the i-th power output interval during time period t;

[0090] The formulas for calculating the piecewise linear parameters and variables mentioned above are as follows:

[0091]

[0092] in, and Let α represent the maximum and minimum output levels of the thermal power unit g, respectively. g ,β g and γ g These are the operating cost coefficients for thermal power unit g, and P. g,tThe output level of the thermal power unit g during time period t;

[0093] The start-up and shutdown costs of a thermal power unit are a linear function of the unit's start-up and shutdown variables, which can be expressed as:

[0094]

[0095] in, and These represent the start-up or shutdown operation of the thermal power unit g during time period t, indicated by 0 or 1. and denoted as the start-up and shutdown cost coefficient for thermal power unit g.

[0096] The penalty cost of curtailing wind and solar power is a linear function of the amount of wind and solar power curtailed, which can be expressed as:

[0097]

[0098] Where W represents the total number of wind farms and photovoltaic power stations, w is the index of a wind farm or photovoltaic power station, and P w,t This represents the actual power output level of a wind farm or photovoltaic power station w during time period t. C represents the predicted power output level of a wind farm or photovoltaic power station w during time period t. w , where w is the penalty cost coefficient for wind farms or photovoltaic power plants.

[0099] The penalty cost for load shedding is a linear function of the load shedding amount, which can be expressed as:

[0100]

[0101] Where E is the total number of nodes, e is the index of a node, and D... e,t The actual load supply level of node e in time period t. C represents the predicted load demand level for node e in time period t. e Let be the penalty cost coefficient for node e.

[0102] S102: Set constraints:

[0103] ① Actual output and load shedding constraints

[0104] Each thermal power unit has its own inherent physical limitations. During startup, it can only operate stably if its output power is within a certain range. Curtailment of wind and solar power and load shedding also only have physical significance within practical limits. These constraints can be expressed as:

[0105]

[0106] Where g, t, w, and e are the index markers for thermal power units, time periods, wind farms, and load nodes, respectively, and x g,t This indicates the start-up and shutdown status of the thermal power unit g during time period t. and These represent the maximum and minimum output levels of thermal power unit g, respectively. This represents the predicted power output level of a wind farm or photovoltaic power station w during time period t. P represents the predicted load demand level for node e in time period t. g,t Let g be the output level of the thermal power unit during time period t. This represents the predicted power output level of a wind farm or photovoltaic power station w during time period t. Let be the predicted load demand level for node e in time period t.

[0107] ②System constraints

[0108] System scheduling needs to meet the system's load demands while ensuring safe operation. Therefore, it is necessary to guarantee power balance across all nodes and ensure that the power flow on each transmission line does not exceed its maximum allowable transmission capacity. Using DC power flow calculations to determine the power flow between lines, these constraints can be expressed as:

[0109]

[0110] Where l is the index flag of the line, and These represent the collection of thermal power units, wind farms, and photovoltaic power stations connected to node e. and Let F be the set of routes that start and end at node e. l,t F represents the power flowing through line l during time period t. l max and B l These represent the transmission capacity limit and susceptance of line l, respectively, and θ e,t Let be the phase angle of node e in time period t.

[0111] ③ Unit start-up and shutdown status constraints

[0112] The start-up and shutdown status of the generating unit is a binary decision variable. The start-up and shutdown status of each time period is jointly determined by the start-up and shutdown status of the previous time period and the start-up or shutdown operation of the current time period. These constraints can be expressed as:

[0113]

[0114] in, and These represent the start-up or shutdown operation of the thermal power unit g during time period t, indicated by 0 or 1.

[0115] It should be noted that the start-stop state of the unit in time period 1 is constrained based on the initial start-stop state of the unit; where T1 represents the first time period. This paper assumes that the initial start-stop state of the unit is unknown, so the start-stop state constraint of the unit does not need to be satisfied in the first time period.

[0116] ④ Climbing constraints

[0117] The output level of a thermal power unit cannot vary beyond its physical limitations within a given time period. Furthermore, if the unit needs to start up in the next moment, its maximum output at that moment is subject to additional limitations; similarly, if the unit needs to shut down in the next moment, its maximum output at that moment is subject to additional limitations. These constraints can be expressed as:

[0118]

[0119] Among them, RU g RD g SU g and SD g These represent the maximum upward / downward climbing ability of thermal power unit g, the maximum output level during the first period after startup, and the maximum output level during the last period before shutdown, respectively.

[0120] It should be noted that the ramp-up constraint of the unit in time period 1 is calculated based on the initial output level of the unit.

[0121] ⑤ Minimum start-stop time constraint

[0122] Thermal power units are powered by boilers, and boilers have start-up and shutdown time limitations. Therefore, thermal power units also have minimum start-up and shutdown time limitations. These constraints can be expressed as:

[0123]

[0124]

[0125] Among them, TU g and TD g These represent the minimum number of consecutive start-up and shutdown periods allowed for thermal power unit g, respectively.

[0126] S2: Apply spatial clustering method to perform spatial clustering of physical parameters to initially cluster the units, improve the constraints based on the initial clustering results and introduce connection constraints.

[0127] S201: Traditional Spatial Clustering Algorithm

[0128] Spatial clustering aims to group units with similar physical parameters and functions together, thereby greatly reducing the number of integer variables representing the start-up and shutdown states of the units and thus improving the solution speed of unit combination problems.

[0129] Generally, after spatial clustering, since units of the same type are considered the same, the physical parameters of units of the same type, such as maximum / minimum output level, ramp capability, minimum start-stop time limit, etc., are set to the average value of parameters of the same type, as shown below:

[0130]

[0131]

[0132] Where 'c' is the index flag of the generator cluster. This refers to the set of all units included in class C. α represents the total number of units included in class C. g ,β g and γ g Let g be the operating cost coefficient for thermal power unit g. and Let g be the start-up and shutdown cost coefficient for thermal power unit g. and RU represents the maximum and minimum output levels of thermal power unit g, respectively. g RD g SU g and SD g These represent the maximum upward / downward climbing ability of thermal power unit g, the maximum output level during the first period after startup, and the last period before shutdown, respectively. TU g and TD g These represent the minimum number of consecutive start-up and shutdown periods allowed for thermal power unit g, respectively.

[0133] In spatial clustering of unit combinations, the index of the unit cluster replaces the index of a single unit, and the variables representing the start-up and shutdown states of the units are transformed from binary variables to integer variables. Therefore, the constraints on the start-up and shutdown states of the units are modified from equations (15) and (16) to equations (36) and (37).

[0134]

[0135] Where c is the index flag of the generator cluster, t is the index flag of the time period, and x c,t , and These represent the state of unit cluster c during time period t, and the number of units started and shut down, respectively. This indicates the number of units contained in unit cluster c.

[0136] Similarly, variables and parameters indexed by g in the objective function and constraints need to be replaced with variables and parameters indexed by c. The output of the unit cluster can be expressed as:

[0137]

[0138] in, Let P be the set of all units contained in unit cluster c. g,t P represents the output level of unit g during time period t. c,t This represents the output level of unit cluster c during time period t.

[0139] Since the start-up and shutdown status of each unit within the same type of generating unit cannot be determined, the following constraints are required in terms of actual output and load shedding:

[0140]

[0141] in, This represents the maximum permissible output range of unit g.

[0142] The minimum downtime constraint for generating units needs to be modified because the variable changes from unit state to generating cluster state. Specifically, the number of generating units in cluster c during time period t that are in a downtime state should be greater than or equal to the sum of the number of generating units shut down during the previous minimum downtime periods in cluster c, as shown below:

[0143]

[0144] Among them, TU c and TD c These represent the minimum allowed continuous start-up and shutdown times for unit cluster c, respectively.

[0145] In addition, connection constraints need to be introduced to describe the relationship between the state of a single unit and the start / stop state of its cluster, as shown below:

[0146]

[0147] Where g is the index flag of the unit, x g,t , and These represent the state of unit g during time period t, as well as the start-up and shutdown operations.

[0148] S3: By calculating the power distribution factor, assess the contribution of all thermal power units to the load nodes, and remove units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold.

[0149] It should be noted that the number of units in each category and the similarity of units within each category are two crucial factors affecting the computational speed and approximation of results in spatial clustering of unit combinations. Fewer categories and more units within each category mean less computation time, but may lead to inaccurate unit combination and scheduling schemes. Traditional spatial clustering methods only focus on the similarity of unit physical parameters, ignoring the influence of system topology, load distribution, and operating conditions on unit clustering. This can lead to significant errors in the unit combination results after spatial clustering. For example, if two units with the same physical parameters but far electrical distances and significantly different load distributions and variations are grouped into the same cluster, the final unit combination result may differ significantly from the result of not grouping them together. This is because the two units have considerably different impacts on different nodes and lines in the system.

[0150] Consider a three-node system with the following topology: Figure 3 As shown in Table 1, the physical parameters of the two thermal power units are the same, the load demand is shown in Table 2, and the line parameters are shown in Table 3.

[0151] Table 1 Unit parameters of the three-node system

[0152] <![CDATA[P min ]]> <![CDATA[P max ]]> <![CDATA[P0]]> RU / RD SU / SD TU / TD 200MW 350MW 350MW 50MW 250MW 1h

[0153] Table 2 Load requirements of the three-node system

[0154]

[0155] Table 3 Line parameters of the three-node system

[0156] line r x b 1-2 0.01pu 0.085pu 0p.u. 2-3 0.01pu 0.085pu 0p.u.

[0157] The above three-node system is solved using both the original unit combination model and the unit combination model after spatial clustering. The resulting unit output levels are as follows: Figure 4 As shown in (b), since the two units have the same physical parameters, traditional spatial clustering methods would cluster the two units into the same cluster.

[0158] Figure 4(a) is the original unit configuration result, which is correct. Both thermal power units need to be running in the first three time periods, and Unit 1 is shut down in the fourth time period. If there is no load reduction in the second time period, both units need to operate at their maximum output level (350MW) in the second time period. Meanwhile, for Unit 1 to be shut down in the fourth time period, its output level in the third time period needs to be lower than its maximum output level before shutdown (SD = 250MW). However, due to the limited ramp-up capability of the units (RD = 50MW / h), these two requirements cannot be met simultaneously. Therefore, if Unit 1 needs to be shut down in the fourth time period, its maximum output level in the second time period is 300MW, which explains why a 50MW load reduction is required in the second time period.

[0159] Figure 3 (b) shows the result of unit combination using the traditional spatial clustering method. The specific states of the two thermal power units cannot be directly obtained from the diagram, but based on the known parameters and constraints, it can be determined that both units are operational in the first three time periods, with only one unit operating in the fourth. In the second and third time periods, the output level of the unit cluster decreases at a rate of 100MW, which is feasible because both units have a ramp-up capability of 50MW / h. In the third and fourth time periods, the output level of the unit cluster is further reduced by 250MW, which is also feasible because the maximum allowed output before unit shutdown is 250MW, satisfying the reduction capability of the unit cluster output. This scheme does not superficially violate any constraints of spatial clustering unit combination and does not require load reduction, resulting in a better outcome than without spatial clustering. However, since both thermal power units operated at maximum output during the first and second periods, they would need to maintain an output of 300MW during the third period to meet load demand in order to satisfy the ramp-up constraints between the second and third periods. This output level exceeds the maximum allowable output level before the units were shut down, making this outcome practically infeasible.

[0160] This example precisely reveals a fundamental flaw in spatial clustering methods: the unit combination model after unit clustering lacks the output level of each individual unit, leading to errors compared to the original, non-clustered unit combination model. The unit clustering model only focuses on the total output level of the entire cluster. However, as shown in the three-node system example above, the flexibility provided by a unit group depends not only on the total output level but also on how that output level is distributed among the units within the cluster. Furthermore, due to constraints such as past decisions and future load demand (e.g., shutdown operations in the fourth period and unit ramp-up limitations forcing one unit to have an output level of 300MW in the second period), the output levels among units within the cluster cannot be freely allocated.

[0161] It is important to note that, in reality, thermal power units can completely disconnect from the grid, meaning their direct physical output level is zero. While such unplanned outages incur additional costs, they can still avoid the load reduction scenarios described above. In the spatially clustered unit combination model, because it is impossible to identify whether a unit is directly disconnected from the grid within the cluster, these additional costs cannot be fully represented. In this context, compared to the traditional unit combination model, while the spatially clustered unit combination model reduces the probability of generating infeasible solutions, it is more likely to produce suboptimal scheduling schemes because it does not fully consider the actual costs incurred in this situation.

[0162] To compensate for the shortcomings of spatial clustering methods, an intuitive idea is to avoid making overly "complex" allocations to the overall output level of unit clusters. This requires further improvement in the quality of the unit clusters obtained after spatial clustering. Traditional spatial clustering methods only consider the physical parameters of the units. However, as shown in the three-node system example above, this method may not be able to accurately represent the start-up and shutdown states of units within the same cluster when there are significant differences in load distribution and operating states. Following this line of thought, we can draw some insights: the spatial clustering results may differ across time periods. That is, if there are two time periods where the load distribution and operating states of the units within certain clusters are similar, these corresponding clusters in these two time periods should contain the same units; otherwise, these clusters in these time periods should contain different units.

[0163] It should be noted that, based on the above analysis, this invention proposes a spatial clustering method based on the power distribution factor. This method simultaneously considers the physical parameters and operating characteristics of the generating units to help the spatially clustered unit combination results approach the traditional un-clustered unit combination results. The core idea of ​​this method is to calculate the power distribution factor to evaluate the contribution of each thermal power unit to each load node, and to remove units whose power distribution factors significantly deviate from those of other units in the same cluster. This factor comprehensively considers factors such as load distribution, grid topology, and unit output level, thus better coordinating physical parameter characteristics and obtaining more accurate unit clustering results. The calculation method of the power distribution factor is as follows:

[0164] Suppose an electrical network has n nodes and q generating units. Given a fixed power flow profile, the generating units are described as current sources, denoted as q0. This represents the set of all generating units; loads are equivalent to impedances and connected to the network. The node voltage equations of the power grid are as follows:

[0165]

[0166] Among them, Z im These are elements in the impedance matrix Z; Let be the voltage of the i-th node; Let be the current of the m-th unit.

[0167] According to basic circuit theory, the power generated by the k-th (k∈G) unit in the network is:

[0168]

[0169] Since the port power of an ideal source is entirely output by itself and does not contain any power components from other units, therefore, the definition is... The power component in the form of a power component represents the power generated by unit k when unit m and unit k are operating simultaneously in the network.

[0170] In the connecting branch ij, the power flowing from node i to node j for:

[0171]

[0172] Where i,j=1,2,...,n,y ij Let be the impedance between lines ij.

[0173] The load power S of node i is:

[0174]

[0175] in, Let be the injected current at node i.

[0176] Therefore, the transmission power of the connecting branch ij The power provided by unit k for:

[0177]

[0178] in,

[0179] According to the formula I ij←ks Further, it can be expressed as in, Let be the distributed voltage of unit k at node i, then The physical meaning is the distributed current of unit k on branch ij.

[0180] available:

[0181]

[0182] Define β as the current distribution factor of the generating branch. ij←ks This represents the current distribution factor of unit k on branch ij.

[0183] Similarly, the load power at node i is the power provided by unit k. for:

[0184]

[0185] in, Similarly, by analyzing its expression, we can obtain... Let be the distributed current of unit k in the load at node i. Then we have

[0186]

[0187] Where, β i←ks Let be the current distribution factor of unit k at node i load. Therefore, we can obtain...

[0188]

[0189] in, These represent the distributed currents of unit k on branch ij and the node loads, respectively; β ij←ks =y ij (Z ik -Z jk ), β ij←ks =y i Z ik These are the current distribution factors of unit k on branch ij and node i load, respectively.

[0190] Consider a generalized node that includes all connecting branches in the network. Since the directed graph of current distribution does not contain directed loops, it can be assumed that the injected current from the unit flows out through all loads via the network. Taking unit k as an example, its injected current... Defined as the sum of the current components distributed to each load, i.e.

[0191]

[0192] Among them, i ks→i for A component represents the current component allocated by unit k to the load at node i; D is the set of all loads. From the above definitions, we can obtain the following... Figure 4 The diagram shows the current distribution. (See diagram:) The injected current for unit k; Let be the current of the load at node i; D is the set of all loads.

[0193] like Figure 5 As shown: Unit current It contains the current component β distributed to each load. i←ks (k∈G); Current of the load at node i It contains the distributed current I of each unit. i←ks (k∈G).

[0194] Due to the non-destructive nature of current, the distributed current of unit k at node i load can be represented. Tracing back to unit k, we can conclude that the current component distributed by unit k to the load at node i is numerically equal to the distributed current I obtained by the load at node i from unit k, i.e.

[0195]

[0196] in, The current component allocated from unit k to the load at node i; Let be the distributed current of unit k in the load at node i.

[0197] Therefore, the injected power of unit k is

[0198]

[0199] definition As can be seen from Section 2.1, for One component represents the current component distributed from unit k to the load at node i at the unit end, from which the following can be derived: for One component represents the power component distributed from unit k to the load at node i. This component should be equal to the losses incurred during the power transfer from unit k to the load at node i, plus the power drawn from unit k at the load at node i. From equation (14), we can obtain

[0200]

[0201] It can also be expressed as

[0202]

[0203] in, Let be the power distribution factor of unit k and the load of node i.

[0204] Therefore, given a fixed power flow section, the power allocated from unit k to node i can be determined by the power distribution factor. It can be obtained directly from equation (57).

[0205] It is worth noting that the calculation of the power distribution factor depends on the power flow section. Therefore, it is necessary to calculate the optimal DC power flow for the entire time period before the calculation, and to use the active power output and load of the units as the initial points for calculating the AC power flow. The specific steps of the spatial clustering algorithm based on the power distribution factor proposed in this paper are as follows.

[0206] Step 1: Calculate the optimal DC power flow for the entire time period and set the deviation threshold ε for the power distribution factor.

[0207] Step 2: Based on the physical parameters of the units, all units are clustered into class C with indices c = 1, ..., C, and t = 1.

[0208] Step 3: Using the active power output and load information of the units for time period t obtained from the optimal DC power flow, calculate the AC power flow section and the power distribution factor β of the units for time period t. t The size of the matrix is ​​G×E, and let c=1.

[0209] Step 4: Select the power distribution factor for the Class C unit cluster. Its size is Calculate the maximum intra-class distance of this matrix. Does it exceed the set threshold ε? If Then, split clustering, a type of hierarchical clustering, is used to continuously remove groups that are far apart and generate a new group, until... Otherwise, retain the original clustering results. Let c = c + 1.

[0210] Step 5: If c = C, obtain the corrected clustering results for time period t. All units are clustered into C. t 'Class, index c' t =1,...,C t ', let t = t + 1. Otherwise, return to step 4.

[0211] Step 6: If t = T, output the clustering results of all time periods. Otherwise, return to step 3.

[0212] Among them, the maximum intra-class distance The calculation formula is as follows:

[0213]

[0214] To verify the effectiveness and implementation steps of the aforementioned spatial clustering method based on power distribution factor, we applied it to the three-node system described above for testing. Specifically, we first set the minimum output P of the two units... min=0MW, calculate the optimal DC power flow of the three-node system throughout the entire time period, and calculate the AC power flow section using the active power output and load information of the units in each time period. Next, calculate the power distribution factor for each time period, as shown in Table 4. It can be found that the power distribution factors of the two units differ significantly from time period 1 to time period 3, indicating that their operating conditions are quite different and they should not be clustered into one unit cluster. However, in time period 4, the power distribution factors of the two units are similar, indicating that their operating conditions are very similar and they can be clustered into one unit cluster. Using this spatial clustering result to calculate the unit grouping, the resulting unit output level is as follows: Figure 6 As shown.

[0215] Table 4 Power distribution factors of the three-node system at different time periods

[0216]

[0217]

[0218] It can be observed that the unit combination results obtained by the spatial clustering method based on power distribution factors are consistent with the original unit combination results. In the second time period, this method resulted in a 50MW load reduction. In the first and third time periods, the two units were not aggregated into a single cluster; Unit 2 maintained a consistent output level of 350MW, while Unit 1 continuously reduced its output level with a ramp-up capability of 50MW / h. This part of the result is consistent with the original unit combination results. In the fourth time period, the two units were aggregated into a single cluster, and only the total cluster output level of 350MW is visible, but the minimum combined output of the units is P. min =200MW, and the output of Unit 1 in the third period was 250MW. It can be inferred that only Unit 2 in the unit cluster was turned on, and Unit 1 was turned off in the fourth period.

[0219] The example of the three-node system above illustrates the effectiveness of the spatial clustering method based on power distribution factors proposed in this paper. However, constructing a unit combination model using this clustering result requires some modifications compared to traditional spatial clustering unit combination methods. One modification involves replacing the retrieval marker 'c' of the unit cluster with 'c', which varies over time. t In addition, ramp-up constraints, unit start-up and shutdown constraints, and connection constraints all need to be adjusted because the unit group may change over time.

[0220] As shown in equations (60)-(65), when the clustering results of the units change between consecutive time periods, the ramping constraints must consider the different unit clusters before and after. In the worst case with extremely low probability, the load and renewable energy fluctuate greatly in each time period, so the unit clustering results in each time period may be different, which may generate twice the number of ramping constraints. In general, the unit clusters do not change frequently, and a large number of useless ramping constraints may be generated in the model. Methods such as effective constraint identification can be used to filter them to further improve the calculation speed. At the same time, in order to ensure that the minimum start-up and shutdown time limit is not affected by the unit clustering results that change over time, the unit start-up and shutdown state constraints and connection constraints must be applied to all unit clusters in each time period, as shown in equations (66)-(71).

[0221]

[0222]

[0223] Where g, c t ' and t are the indexes for the machine group, cluster, and time period, respectively. For unit cluster c t 'At the output level in time period t, and These are the unit clusters c t 'The number of units started and shut down during time period t.' and These represent the generator cluster c. t Maximum uphill / downhill climbing ability, maximum output level during the first period after startup and the last period before shutdown. Indicates unit cluster c t The minimum output level of 'x' g,t , and These represent the start-up and shutdown states of unit g during time period t, and the start-up and shutdown operations, respectively. and These are the unit clusters c t 'The included units and their total number.'

[0224] Example 2, refer to Figure 7 and Figure 8 This is the second embodiment of the present invention, which provides a method for solving the spatial clustering unit combination problem based on power distribution factor. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0225] 1. Test system configuration

[0226] The performance of the proposed method was evaluated using a modified 72-node system. This system comprises three IEEE-RTS-24-node systems interconnected by twelve 500MW transmission lines. The modified system includes 96 thermal units, three wind farms located at nodes 9, 33, and 57, 126 transmission lines, and 51 loads, such as... Figure 7 As shown. In addition, to alleviate the symmetry problem, a random +1% noise was introduced into the unit's operating cost during the unit combination solution process.

[0227] 2. Comparison Indicators

[0228] To quantify the effectiveness of the proposed method, we calculated several effectiveness metrics to evaluate its performance in solving the unit combination problem. The results of the original network-constrained unit combination model are denoted by the symbol "^".

[0229] ① Total system cost deviation Δ Cost The percentage difference of the objective function values ​​is calculated as follows:

[0230]

[0231] Among them, C Cost This represents the total cost of the system.

[0232] ② Unit status error, the percentage difference Δ between planned start / stop status for each unit US The calculation is as follows:

[0233]

[0234] Where, x g,t Let G be the start / stop status of unit g during time period t, G be the total number of units, and T be the total number of time periods.

[0235] ③ Power flow error in the transmission system; percentage error Δ in power flow for each transmission line. PF The calculation is as follows:

[0236]

[0237] Among them, F l,t Let be the power flow magnitude of line l during time period t.

[0238] ④ Unit Combination Calculation Time: The unit combination calculation time is set as the running time of the solver (the GUROBI solver is used in this test) to solve the unit combination problem.

[0239] ⑤ Total computation time, which includes the unit combination computation time, as well as the total running time of other operations such as preprocessing and clustering.

[0240] 3. Research on the effectiveness of spatial clustering methods

[0241] Three different spatial clustering methods were tested on a 7-day dataset to solve the network-constrained unit combination problem. The first method, labeled "M1," clusters units solely based on their physical parameters. The second method, labeled "M2," further subdivides the clustering results based on the power flow results of the unit combination under specific load conditions, building upon M1. The third method, proposed in this patent, labeled "M3," further subdivides the clustering results based on the power distribution factor of each unit, with a power distribution factor deviation threshold set to ε = 0.15. Specific results are shown in Table 1.

[0242] Table 5 Comparison of results for the network-constrained unit combinatorial problem under different spatial clustering methods.

[0243]

[0244] These results indicate that: (1) M1 clusters all units into 9 categories, achieving the maximum reduction in integer variables and the fastest computation speed. However, it also has relatively large errors in the objective function, unit status, and line power flow. (2) M2 further refines the clustering by clustering units into 33 categories and verifies it through line power flow under specific load conditions. This method produces smaller errors compared to M4. Although it takes slightly longer, the preprocessing time is longer because additional calculations of network-constrained unit combination problems are required for each load condition. (3) The proposed M3 further clusters units according to the power distribution factor at different time periods, resulting in four different unit clustering results. Although it has the longest computation time due to the additional constraints added by managing unit clustering that varies at different time periods, it also provides the most accurate solution.

[0245] Different spatial clustering methods can lead to different levels of error in power flow calculation. Figure 8 The results show the percentage deviation of three spatial clustering methods from the line power flow based on the original network-constrained unit combination model within a specific time period in an IEEE-RTS-24 node region. Among these methods, M1 exhibits the largest power flow deviation. Although M2 generally produces smaller errors than M1, it may produce larger errors on certain lines under specific load conditions because the specific load conditions selected by M2 do not fully cover the test cases. In contrast, M3 adjusts the unit clustering results according to the power distribution factor, thus producing the smallest power flow error. These results clearly demonstrate that the results of spatial clustering should vary with system operating conditions.

[0246] Example 3 is the third embodiment of the present invention, which differs from the previous two embodiments in that:

[0247] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0248] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0249] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0250] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0251] Example 4 is the fourth embodiment of the present invention. This embodiment provides a system for solving the spatial clustering unit combination problem based on power distribution factor, including: a network-constrained unit combination construction module, a spatial clustering module, and a power distribution factor-based calculation module.

[0252] The network-constrained unit combination construction module builds a network-constrained unit combination model, sets an objective function to minimize the total system cost, and sets constraints.

[0253] The spatial clustering module applies spatial clustering methods to perform preliminary spatial clustering of physical parameters into unit clusters, improves constraints based on preliminary clustering results, and introduces connection constraints.

[0254] Based on the power distribution factor calculation module, the contribution of all thermal power units to the load nodes is evaluated by calculating the power distribution factor, and units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold are removed.

[0255] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for solving the unit combination problem based on spatial clustering of power distribution factor, characterized in that: include, Construct a network-constrained unit combination model, set an objective function to minimize the total system cost, and define the constraints. Spatial clustering method is applied to perform spatial clustering of physical parameters to initially cluster the units. Based on the preliminary clustering results, the constraints are improved and connection constraints are introduced. By calculating the power distribution factor, the contribution of all thermal power units to the load nodes is evaluated, and units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold are removed.

2. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 1, characterized in that: The network-constrained unit combination model includes setting an objective function to minimize the total system cost, including the operating cost of thermal power units, start-up and shutdown costs, penalty costs for curtailment of wind and solar power, and penalty costs for load shedding. minCR+CS+CW+CE Wherein, CR represents the operating cost of the thermal power unit, CS represents the start-up and shutdown cost of the thermal power unit, CW represents the penalty cost of curtailing wind and solar power, and CD represents the penalty cost of load shedding.

3. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 2, characterized in that: The operating cost of the thermal power unit includes the following: the operating cost of the thermal power unit is linearized using a piecewise linear method, which divides the nonlinear objective into h segments, and transforms the total cost of the thermal power unit: Where T and G represent the total time period and the number of generating units, respectively; g and t represent the index markers of the generating units and time periods, respectively; i represents the index marker of the output interval after piecewise linearization; and x represents the index marker of the output interval after piecewise linearization. g,t This indicates the start-stop status of the thermal power unit g during time period t, represented by 0 or 1; Z represents the operating cost incurred by thermal power unit g when it operates at minimum output. g,i Let χ be the slope of the i-th segment after piecewise linearization of the operating cost of thermal power unit g. g,t,i Let g be the power output level of thermal power unit g in the i-th power output interval during time period t; The formulas for calculating the piecewise linear parameters and variables mentioned above are as follows: in, and Let α represent the maximum and minimum output levels of the thermal power unit g, respectively. g ,β g and γ g These are the operating cost coefficients for thermal power unit g, and P. g,t The output level of the thermal power unit g during time period t; The startup and shutdown costs of a thermal power unit are a linear function of the unit's startup and shutdown variables; The penalty cost for curtailing wind and solar power is a linear function of the amount of wind and solar power curtailed. The penalty cost for load shedding is a linear function of the amount of load shedding.

4. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 3, characterized in that: The set constraints include, Actual output and load shedding constraints; System constraints are applied by calculating the power flow between lines using DC power flow calculations. The start-stop state constraints of the unit are determined by the start-stop state of the previous period and the start-up or shutdown operation of the current period, and the start-stop state constraints of the unit do not need to be met in the first period. To establish ramp constraints, the output level of thermal power units shall not vary within a unit time period beyond the current physical limit of the unit. Furthermore, if a thermal power unit needs to start up in the next moment, its maximum output in the next moment shall be subject to additional restrictions. At the same time, if a thermal power unit needs to shut down in the next moment, its maximum output in the current moment shall be subject to additional restrictions. Establish a minimum start-stop time limit.

5. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 4, characterized in that: The improved constraints include, in the combination of spatially clustered units, replacing the index of a single unit with the index of the unit cluster, and transforming the variables characterizing the start-up and shutdown states of the units from binary variables to integer variables. The improved constraints on the start-up and shutdown states of the units are as follows: Where c is the index flag of the generator cluster, t is the index flag of the time period, and x c,t , and These represent the state of unit cluster c during time period t, and the number of units started and shut down, respectively. This indicates the number of units contained in unit cluster c; In the objective function and constraints, variables and parameters indexed by g are replaced with variables and parameters indexed by c. The output of the unit cluster is represented as follows: in, Let P be the set of all units contained in unit cluster c. g,t P represents the output level of unit g during time period t. c,t The output level of unit cluster c in time period t; Since the start-up and shutdown status of individual units within the same type of generating unit cannot be determined, the following constraint is added to the actual output and load shedding constraints: in, This represents the maximum permissible output range of unit g; The minimum downtime constraint for generating units needs to be modified because the variable changes from unit state to unit cluster state. Specifically, the number of units in cluster c that are in a downtime state during time period t should be greater than or equal to the sum of the number of units shut down during the previous minimum downtime periods in cluster c, as shown below: Among them, TU c and TD c These represent the minimum allowed continuous start-up and shutdown times for unit cluster c, respectively; The introduced connection constraints include those describing the relationship between the state of a single unit and the start / stop state of its cluster, as shown below: Where g is the index flag of the unit, x g,t , and These represent the state of unit g during time period t, as well as the start-up and shutdown operations.

6. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 4, characterized in that: The calculated power distribution factor includes, Define the topology of the power network, including the connection relationships of nodes and tie branches, determine the distribution of generating units, record the active power output of the generating units, and convert the load into an equivalent impedance in the network, and collect the load data of each node. Calculate the node voltage in the power grid, record the voltage distribution of each unit at its connection node, determine the power output of each unit in the network, calculate the power transmission of the unit at different nodes, and distinguish the contribution of each unit. Determine the power transfer between branches, including the power flow from one node to another, and calculate the total power in the branch and the power contributed by each unit; The power distribution factor of the computer group includes the power distribution factor of the unit on the tie branch and the power distribution factor of the unit on the node load; Based on the power distribution factor of the unit, calculate the power allocated to each load, trace the source of the power provided by the unit, and verify whether the power allocated by the unit to each node load conforms to the total power distribution relationship according to the principle of non-destructive testing. Calculate the optimal DC power flow for the entire time period, obtain initial point data, and use AC power flow calculation to adjust the unit output to meet power flow section constraints.

7. The method for solving the unit combination problem based on spatial clustering of power distribution factor as described in claim 4, characterized in that: The calculation of the power distribution factor also includes, Calculate the optimal DC power flow throughout the entire time period and set the deviation threshold for the power distribution factor; Based on the physical parameters of the units, all units are clustered into category C; Using the active power output and load information of the units in time period t obtained from the DC optimal power flow, calculate the AC power flow section and the power distribution factor of the units in time period t. Select the power distribution factor of the c-type unit cluster and calculate whether the maximum intra-class distance of the power distribution factor matrix of the current unit cluster exceeds the set deviation threshold. If the maximum intra-cluster distance exceeds the set deviation threshold, split clustering in hierarchical clustering is used to continuously remove units that are far apart and generate a new cluster until the maximum intra-cluster distance is less than or equal to the set deviation threshold; otherwise, the original clustering result is maintained. If c = C, then the corrected clustering results for time period t are obtained, and all units are clustered into C. t If the current cluster is of type c, then select the power distribution factor of the c-th type of unit cluster and calculate the maximum intra-class distance of the matrix of the power distribution factors of the current unit cluster. If t = T, output the clustering results of all the units in the last time period; otherwise, reuse the active power output and load information of the units in time period t obtained from the DC optimal power flow to calculate the AC power flow section and the power distribution factor of the units in time period t, where T is the total number of time periods.

8. A system for solving the spatial clustering unit combination problem based on power distribution factor, using the method for solving the spatial clustering unit combination problem based on power distribution factor as described in any one of claims 1 to 7, characterized in that, include: The module includes a network-constrained unit combination construction module, a spatial clustering module, and a power distribution factor-based calculation module. The network-constrained unit combination construction module builds a network-constrained unit combination model, sets an objective function to minimize the total system cost, and sets constraints. The spatial clustering module applies spatial clustering methods to perform preliminary spatial clustering of physical parameters into unit clusters, improves constraints based on preliminary clustering results, and introduces connection constraints. Based on the power distribution factor calculation module, the contribution of all thermal power units to the load nodes is evaluated by calculating the power distribution factor, and units of the remaining units in the same cluster whose power distribution factor is greater than the deviation threshold are removed.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for solving the spatial clustering unit combination problem based on power distribution factor 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, When the computer program is executed by the processor, it implements the steps of the method for solving the spatial clustering unit combination problem based on power distribution factor as described in any one of claims 1 to 7.

Images