Unit combination system and method based on time and space clustering

Abstract

The invention relates to a unit commitment system and method based on time and space clustering, and the method comprises the steps: obtaining first data of a target power system, and building a first unit commitment model according to the first data; updating the first unit commitment model based on the first clustering strategy and the second clustering strategy to obtain a second unit commitment model; and solving the second unit commitment model according to the first improved algorithm. Through the time and space clustering technology, the complexity of the unit commitment model is effectively reduced, and the calculation efficiency is improved. The uncertainty and variability of renewable energy sources are considered, and the operation flexibility of the power system is enhanced. By introducing the improved algorithm, the solving process of the unit commitment is optimized, and the feasibility and economical efficiency of the planning scheme are ensured. The unit combination system and method can adapt to continuously changing new participants and operation conditions in the power system, and powerful support is provided for efficient operation of the power system.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system unit combination, and in particular to a unit combination system and method based on time and space clustering. Background Art

[0002] The new power system revolution has brought new requirements and challenges to the planning research of power systems. The increasing number of new participants, such as renewable energy, demand response, FACTS devices, etc., will bring greater computational complexity to the operation of the power system. Therefore, it is necessary to consider the specific details of the power system operation in the planning stage. In particular, the increasing penetration of renewable energy in the power system has brought huge uncertainty and variability to the power system, thereby increasing the demand for operational flexibility. In order to achieve high renewable energy penetration, detailed operational flexibility-related constraints, such as capacity reserve requirements, unit ramping requirements, and transmission capacity limitations, need to be considered in the power system planning model. Ignoring these constraints in the planning process may lead to infeasible or uneconomical planning schemes, resulting in a lack of operational flexibility. The network constrained unit commitment (NCUC) model is widely used in short-term power system operation problems and is considered to be able to capture all the operational details of the power system. However, the NCUC model usually involves a large number of binary variables representing the start and stop states of the units, so the computational burden is heavy when modeling the medium- and long-term power system operation in the planning model. Therefore, many studies are devoted to developing simplified and accurate NCUC models.

[0003] As a mixed integer programming problem, NCUC's main computational burden is caused by integer variables. Most methods aimed at reducing computational complexity are committed to reducing the number of binary variables, a technique known as spatial clustering. For example, similar units are classified into the same category so that integer variables can represent the status of all units in the cluster. In order to consider the impact of virtual nodes on the network topology after unit clustering, some scholars have proposed a method of combining units while maintaining the original topology and spatially clustering them. Furthermore, in order to cluster units more finely, some scholars have proposed a clustering method based on power flow to improve clustering performance under specific load conditions.

[0004] Another commonly used method to reduce the computational burden of NCUC is spatial clustering. Spatial clustering reduces the computational burden by aggregating representative periods from the study period. Common spatial clustering is to cluster the load and renewable energy curves using traditional algorithms such as k-means or hierarchical clustering. There is currently no unit grouping method based on time and space clustering. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a unit combination system and method based on time and space clustering. Through the time and space clustering technology, the complexity of the unit combination model is effectively reduced and the calculation efficiency is improved. The uncertainty and variability of renewable energy are taken into account, and the operational flexibility of the power system is enhanced. By introducing an improved algorithm, the solution process of the unit combination is optimized, and the feasibility and economy of the planning scheme are ensured. The unit combination method and system of the present invention can adapt to the ever-changing new participants and operating conditions in the power system, and provide strong support for the efficient operation of the power system. By reducing the number of binary variables, the computational burden of the NCUC model is reduced, making the modeling of medium- and long-term power system operation more efficient. The unit combination method and system of the present invention not only ensure the operational flexibility of the power system, but also take network constraints into account, ensuring the stability and reliability of the power system.

[0006] The present invention solves the technical problem by adopting the following technical solutions:

[0007] A unit combination system based on time and space clustering includes a model building module, a model updating module and a solving module, wherein the model building module, the model updating module and the solving module are connected in sequence, the model building module is used to obtain first data of a target power system, and to establish a first unit combination model according to the first data; the model updating module is used to update the first unit combination model based on a first clustering strategy and a second clustering strategy to obtain a second unit combination model; the solving module is used to solve the second unit combination model according to a first improved algorithm.

[0008] A combination method of a unit combination system based on time and space clustering comprises the following steps:

[0009] Step 1: The model building module obtains first data of the target power system and builds a first unit combination model according to the first data;

[0010] Step 2: The model updating module updates the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain a second unit combination model;

[0011] Step 3: The solving module solves the second unit combination model according to the first improved algorithm to obtain a unit combination result.

[0012] Moreover, the first unit combination model in step 1 includes a first objective function and a first constraint condition, the first objective function includes unit startup and shutdown costs, operating costs, wind power reduction penalties and load reduction; the first constraint condition includes output level and load reduction constraints, safety constraints, decision constraints and time constraints;

[0013] Among them, the combination model of the first unit is:

[0014]

[0015]

[0016] Among them, C Cost is the total system cost, g, w, n, t are index items of unit, wind farm, load and time period respectively; P g,t ,P w,t and D n,t are continuous variables, corresponding to the power generation of the generator, the dispatching amount of the wind farm and the load reduction; x g，t , and are integer variables, representing the start / stop state and start / stop state of the unit respectively; C w and C l is the operating cost coefficient; and is the hourly forecast of wind power generation and load; G, W, N and T are the total number of units, wind farms, loads and time periods respectively; w t is the weight of the time period after clustering; and is the maximum / minimum output limit of the unit; F l,t ,, and θ n,t They represent the power flowing through the line, the line susceptance and the phase angle of the node respectively; F l max Indicates the maximum power allowed to flow through the line; and They are the collection of the unit, wind farm, and the beginning and end of the line; RU g,t ,RD g,t ,SU g,t and SD g,t They are the up and down ramp limits and the minimum start and shutdown time of the unit; TU g and TD g They are the minimum power on / off time limit; RP 1 ,and are the subsets of the first and smallest on / off periods for each representative day, respectively; and The time periods are two consecutive representative days.

[0017] Furthermore, the step 2 comprises the following steps:

[0018] Step 2.1, performing a first update operation on the first unit combination model according to the first clustering strategy;

[0019] Step 2.2, selecting the first target factor after the first update operation;

[0020] Step 2.3: Perform a second update operation using a second clustering strategy based on the first objective factor to obtain a second unit commitment model.

[0021] Moreover, the specific implementation method of the first update operation in step 2.1 is:

[0022] Step 2.1.1, updating the first constraint condition in the first unit combination model;

[0023] Step 2.1.2, performing a first clustering solution on the first unit commitment model updated by the first updating operation according to the first improved algorithm;

[0024] Step 2.1.3: Select the first target factor after the first update operation according to the first clustering solution result.

[0025] Moreover, the clustering target of the first clustering strategy is the load, renewable energy forecast curve and the ramp condition of the unit; the clustering target of the second clustering strategy is the physical parameters of the unit, and the first target factor is the power distribution factor.

[0026] Moreover, the specific implementation method of the second updating operation in step 2.3 is: updating again the first constraint condition in the first unit commitment model after the first clustering strategy is updated.

[0027] The advantages and positive effects of the present invention are:

[0028] The present invention proposes a unit combination method and system based on time and space clustering, obtains first data of a target power system, and establishes a first unit combination model according to the first data; updates the first unit combination model based on a first clustering strategy and a second clustering strategy to obtain a second unit combination model; and solves the second unit combination model according to a first improved algorithm. Through the time and space clustering technology, the complexity of the unit combination model is effectively reduced and the calculation efficiency is improved. The uncertainty and variability of renewable energy are taken into account, and the operation flexibility of the power system is enhanced. By introducing the improved algorithm, the solution process of the unit combination is optimized, and the feasibility and economy of the planning scheme are ensured. The unit combination method and system of the present invention can adapt to the changing new participants and operating conditions in the power system, and provide strong support for the efficient operation of the power system. By reducing the number of binary variables, the calculation burden of the NCUC model is reduced, making the modeling of the medium and long-term power system operation more efficient. The unit combination method and system of the present invention not only ensure the operation flexibility of the power system, but also take network constraints into account, ensuring the stability and reliability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A method flow chart of a unit combination method and system based on time and space clustering provided by an embodiment of the present invention;

[0030] Figure 2 A flowchart of a hierarchical K-means clustering method improved by Trust-Tech technology of a unit commitment method and system based on time and space clustering provided by an embodiment of the present invention;

[0031] Figure 3 A flow chart of a time and space clustering method for a unit commitment method and system for network-constrained unit commitment problems based on time and space clustering provided by an embodiment of the present invention;

[0032] Figure 4 A schematic diagram of the structure of a modified 72-node system of a unit commitment method and system based on time and space clustering provided by an embodiment of the present invention;

[0033] Figure 5 A schematic diagram of cost error and calculation time of a unit commitment method based on time and space clustering and a system for network-constrained unit commitment problems under different time clustering methods provided by an embodiment of the present invention;

[0034] Figure 6 A schematic diagram of line power flow error of a network-constrained unit commitment problem under different spatial clustering methods in an area of ​​an IEEE-RTS-24 node, provided by a unit commitment method and system based on time and space clustering according to an embodiment of the present invention;

[0035] Figure 7 An internal structural diagram of a computer device of a unit combination method and system based on time and space clustering provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The present invention is further described in detail below with reference to the accompanying drawings.

[0037] Reference Figure 1-Figure 7 , which is the first embodiment of the present invention, and provides a unit combination method and system based on time and space clustering, including:

[0038] There are some problems in the existing related technologies. For example, the traditional unit combination method often finds it difficult to balance the computational efficiency and the accuracy of the results when dealing with large-scale power grid systems. In addition, due to the complexity of the power grid system, traditional optimization algorithms often find it difficult to respond quickly and effectively when faced with variable load demands and the uncertainty of renewable energy. These problems lead to instability and economic problems in power grid operation, and increase the risks and costs of power grid operation.

[0039] The present application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to explain in detail how to implement the unit combination method based on time and space clustering.

[0040] Figure 1 A method flow chart of a unit combination method and system based on time and space clustering is shown, including:

[0041] S101, acquiring first data of a target power system, and establishing a first unit combination model according to the first data;

[0042] It should be noted that the new power system revolution has put forward new requirements and challenges for the planning and research of power systems. The increasing number of new participants, such as renewable energy, demand response, FACTS equipment, etc., will bring greater computational complexity to the operation of the power system. Therefore, it is necessary to consider the specific details of the operation of the power system in the planning stage. In particular, the increase in the penetration rate of renewable energy in the power system has brought huge uncertainty and variability to the power system, thereby increasing the demand for operational flexibility.

[0043] In addition, in order to achieve high renewable energy penetration, detailed operational flexibility constraints need to be considered in the power system planning model, 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 that lack operational flexibility. The network constrained unit commitment (NCUC) model is widely used in short-term power system operation problems and is considered to be able to capture all the operational details of the power system.

[0044] In an optional embodiment, the NCUC model usually involves a large number of binary variables representing the start and stop states of the units, so the computational burden is heavy when modeling the medium- and long-term power system operation in the planning model. Therefore, many studies are devoted to developing simplified and accurate NCUC models.

[0045] In an optional embodiment, the target power system may be a comprehensive power network including multiple generator sets and renewable energy. The system optimizes the combination of generator sets by using a temporal and spatial clustering algorithm to adapt to the intermittency and uncertainty of renewable energy.

[0046] In an optional embodiment, the system is able to classify the operating status of the generator sets based on historical data and forecast information, and minimize operating costs and environmental impact while meeting power demand.

[0047] In an optional embodiment, the system is also capable of adjusting the operating plan of the generator sets in real time in response to changes in grid load and fluctuations in renewable energy output.

[0048] It should be noted that through this approach, the power system can improve operational flexibility and ensure stable operation under high renewable energy penetration.

[0049] In an optional embodiment, the first data of the target power system includes at least the power generation of the generator, the dispatching amount of the wind farm and the load reduction amount, the start / stop state of the unit and the start / shutdown state amount, the hourly predicted wind power generation and load forecast amount, and the hourly predicted wind power generation and load forecast amount and other parameter data;

[0050] In the embodiment of the present application, the first unit combination model includes a first objective function and a first constraint condition;

[0051] The first objective function includes at least unit startup and shutdown costs, operating costs, wind power curtailment penalties, and load curtailment;

[0052] The first constraint condition includes at least output level and load reduction constraint, safety constraint, decision constraint and time constraint.

[0053] In the embodiment of the present application, the NCUC model (i.e., the first unit combination model) under time clustering is established as follows:

[0054]

[0055]

[0056] Among them, C Cost is the total system cost. g, w, n, t are index items of unit, wind farm, load and time period respectively; P g,t ,P w,t and D n,t are continuous variables, corresponding to the power generation of the generator, the dispatching amount of the wind farm and the load reduction; x g,t , and are integer variables, representing the start / stop state and start / stop state of the unit respectively; C w and C l is the operating cost coefficient; and is the hourly forecast of wind power generation and load; G, W, N and T are the total number of units, wind farms, loads and time periods respectively; w t is the weight of the time period after clustering; and is the maximum / minimum output limit of the unit; F l,t ,, and θ n,t They represent the power flowing through the line, the line susceptance and the phase angle of the node respectively; F l max Indicates the maximum power allowed to flow through the line; and They are the collection of the unit, wind farm, and the beginning and end of the line; RU g,t ,RD g,t ,SU g,t and SD g,t They are the up and down ramp limits and the minimum start and shutdown time of the unit; TU g and TD g They are the minimum power on / off time limit; RP 1 ,,and are the subsets of the first and smallest on / off periods for each representative day, respectively; and The time periods are two consecutive representative days.

[0057] In an embodiment of the present application, objective function (1) aims to minimize the total system cost, including unit startup and shutdown costs, operating costs, wind power reduction penalties, and load reduction. Constraints (2)-(4) limit the output level of power generation units and wind farms and the degree of load reduction. Constraints (5)-(8) ensure the security of the power transmission network. Constraints (9)-(10) define the relationship between the start and stop status of the unit and the start / shutdown decision. Constraints (11)-(14) simulate the climbing ability and start / shutdown time of each unit, while constraint (15) associates consecutive representative days.

[0058] In an optional embodiment, the first objective function may also take environmental impact into consideration, and optimize the power generation mix by introducing carbon emission costs to achieve the purpose of reducing greenhouse gas emissions.

[0059] In an optional embodiment, it can be further expanded to include forecasting and dispatching of renewable energy to improve the sustainability and efficiency of the entire power system. Through this comprehensive consideration, a balance of economic, environmental and social benefits of the power system can be achieved.

[0060] In an optional embodiment, the first constraint condition may also include a restriction on a maintenance and repair time window of the power generation unit to ensure long-term stable operation of the power generation equipment.

[0061] In an optional embodiment, constraints on electricity market trading rules may also be added, such as price fluctuations, trading capacity restrictions, etc., to meet the needs of market-oriented operations.

[0062] In an optional embodiment, consideration of power demand side management may also be added to optimize the user's power consumption behavior through intelligent scheduling and demand response mechanisms, thereby reducing peak loads and improving the overall operating efficiency of the power system.

[0063] In an optional embodiment, the safety constraints of the power system may also be considered to ensure that the stability and reliability of the power system are not affected under various operating conditions.

[0064] However, in the embodiment of the present application, only the above-mentioned formulas (1)-(15) are considered, and other constraints and costs are not considered. There is no limitation itself. Relevant technical personnel can design the corresponding first unit combination model, that is, the first objective function and the first constraint condition according to actual needs to meet specific power system operation goals.

[0065] It should be noted that obtaining the first data of the target power system and establishing the first unit combination model based on the first data improves the prediction accuracy of the model, because the first data can reflect the actual operating state of the power system under specific time and space conditions. By establishing the first unit combination model, it is possible to respond more flexibly to changes in the power market and quickly adjust the power generation plan to adapt to demand fluctuations. Since the safety constraints of the power system are taken into account, this method can ensure the stability and reliability of the power system under various operating conditions, thereby ensuring the continuity of power supply.

[0066] S102, updating the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain a second unit combination model;

[0067] In an optional embodiment, the first clustering strategy and the second clustering strategy may be clustering based on time or space, or a combination of the two.

[0068] In an optional embodiment, the first clustering strategy may focus on temporal clustering based on periodic changes in power demand;

[0069] In an optional embodiment, the second clustering strategy may focus on spatial clustering according to geographic location or grid structure.

[0070] It should be noted that through this combination, various factors affecting the operation of the power system can be captured more comprehensively, thereby improving the adaptability and accuracy of the unit combination model. In addition, the update process can include analysis of historical data to identify and predict patterns in power demand, thereby optimizing the combination and dispatch strategy of generator sets.

[0071] In the embodiment of the present application, the first unit combination model is updated based on the first clustering strategy and the second clustering strategy to obtain the second unit combination model including:

[0072] performing a first updating operation on the first unit combination model according to the first clustering strategy;

[0073] Selecting a first target factor after the first update operation;

[0074] The second clustering strategy performs a second updating operation based on the first target factor to obtain a second unit combination model.

[0075] In an optional embodiment, the first objective factor may be a power distribution factor, a power generation efficiency factor, or a cost factor. By selecting these key factors, the second clustering strategy can more accurately classify and optimize the generator sets, thereby minimizing energy consumption and maximizing economic benefits while meeting the load requirements of the power grid. For example, if the first objective factor is the power distribution factor, the second clustering strategy will perform spatial clustering based on the power output characteristics of each unit in different time periods to ensure that units with strong power output capacity can be dispatched during high-load periods, and units with higher efficiency and lower costs can be dispatched during low-load periods. Such a strategy not only improves the operating efficiency of the power system, but also enhances its adaptability to power grid fluctuations.

[0076] In the embodiment of the present application, the first clustering strategy and the second clustering strategy also include:

[0077] The clustering targets of the first clustering strategy are load, renewable energy forecast curve and the ramping condition of the unit;

[0078] The clustering target of the second clustering strategy is the physical parameters of the unit.

[0079] In the embodiment of the present application, the first update operation includes:

[0080] updating a first constraint condition in the first unit combination model;

[0081] performing a first clustering solution on the first unit commitment model updated by the first updating operation according to the first improved algorithm;

[0082] The first target factor after the first update operation is selected according to the first clustering solution result.

[0083] In the embodiment of the present application, the second update operation includes:

[0084] The second updating operation is used to update again the first constraint condition in the first unit commitment model after the first clustering strategy is updated.

[0085] Exemplarily, the first clustering strategy uses a time clustering method, and the selection of different representative days affects the results of time clustering unit commitment. This application introduces an enhanced time clustering method, which is different from the existing methods in two key aspects.

[0086] First, the first constraint in the first unit combination model is updated. The proposed method combines the load curve, renewable energy, and unit ramping capability. Assume that all units are in the startup state and calculate the DC optimal power flow (DCOPF). Specifically, most of the time clustering methods use load curves and renewable energy as clustering objects, as shown in (16), while the proposed method combines the ramping capability obtained from DCOPF as the clustering object to better reflect the operating characteristics, as shown in (17):

[0087]

[0088] in, w=1,...,W,y=1,...,365,h=1,...,24,g=1,...,G, and r g,y,h They are the load forecast of node n at the hth hour, the wind power output forecast of wind farm w and the ramp capacity of generator g, the load at the hth hour on the yth day, the wind power output of the wind farm and the ramp capacity of unit g. y and ψ y is the vector of cluster object data for day y.

[0089] In an optional embodiment, the proposed method utilizes a hierarchical K-means clustering method improved by Trust-Tech technology (i.e., the first improved algorithm). Compared with the existing method using hierarchical clustering or K-means clustering, the method can achieve faster clustering and better clustering results without presetting the final number of clusters.

[0090] Specifically, assuming For the initial data, the specific hierarchical K-means clustering method improved by Trust-Tech technology is as follows:

[0091]

[0092]

[0093] It should be noted that θ is used as an indicator to evaluate the similarity between members and centroids, and its calculation formula is as follows:

[0094]

[0095] Where n is the size of the class, L q is the qth element in the class, and c is the centroid of the class. Similarity can measure the closeness between an element and the centroid. The smaller the value, the closer the class is and the better the clustering effect. Therefore, for each class, if its similarity index can meet the threshold set by the user, this application uses its centroid to represent the class; otherwise, this application divides this "bad" class into several subclasses and uses their centroids to represent each subclass.

[0096] In an optional embodiment, based on the advantages of the hierarchical K-means algorithm in terms of fast calculation speed and good clustering effect, the algorithm improved by using Trust-Tech technology can automatically determine the appropriate number of clusters by setting a similarity threshold and obtain better clustering results. The flowchart of the algorithm is as follows Figure 2 shown.

[0097] In the embodiment of the present application, the first target factor after the first update operation is selected according to the first clustering solution result, and the first target factor is a power distribution factor.

[0098] In the embodiment of the present application, the second clustering strategy uses spatial clustering, and the first constraint in the first unit combination model updated by the first clustering strategy is updated again by the second clustering strategy. The present application will specifically introduce the classic spatial clustering network constraint unit combination model. The physical parameters of the units in the same cluster (such as maximum and minimum power generation limits, minimum start and stop time, etc.) are set to their average values ​​because these units are considered to be the same. Using the unit cluster index c instead of the single unit index g, the variable x c,t , Take integer values ​​instead of 0 / 1 variables. Therefore, the logic of unit commitment and start / shutdown is constrained by equations (19)-(20) instead of (9)-(10).

[0099]

[0100] In an optional embodiment, for other constraints, except for formula (5), all constraints need to be expressed in P c,t To replace P g,t, to represent the total power generation of the units in the cluster, as shown in formula (21). Due to the uncertainty of the state of a single unit in the cluster, the output constraint must be modified according to formula (22) instead of formula (2). In addition, a link constraint is introduced to describe the relationship between a single unit and its cluster, as shown in formulas (23)-(25).

[0101]

[0102] In an optional embodiment, the computational efficiency and solution quality of the spatial clustering method for network-constrained unit commitment are significantly affected by the number of unit clusters and the degree of similarity of units within each cluster.

[0103] In an optional embodiment, aggregating more units can reduce the computation time. However, the increased differences between units of the same type may lead to higher computation errors. Classical spatial clustering methods for network-constrained unit combinations usually rely only on the physical parameters of the units to calculate their similarity, which may not be sufficient. For example, combining two units with the same physical parameters but with significantly different impacts on different nodes and lines may lead to significant errors when their states are represented by an integer variable.

[0104] In the embodiments of the present application, in order to solve these problems, the present application proposes a spatial clustering method based on power distribution factors, which takes into account both the physical characteristics and operating characteristics of the units. The core idea is to calculate the power distribution factors of each unit at different time periods to evaluate the contribution of each unit to each load point, and then eliminate the units with significantly different power distribution factors in the same category.

[0105] In the embodiment of the present application, the following is designed: Figure 3 The flowchart of the time and space clustering method for the network-constrained unit commitment problem shown clearly shows the specific steps mentioned above.

[0106] In the embodiment of the present application, the detailed calculation steps are as follows:

[0107]

[0108] It should be noted that the power distribution factor is related to the operating state of the system. This relationship between the spatial clustering method and the temporal clustering method is reflected in the power distribution factor; the representative time periods obtained from the temporal clustering can help identify potential errors in the spatial clustering, thereby guiding the adjustment of the spatial clustering scheme for these specific time periods.

[0109] In the embodiment of the present application, index c and c t Consistent with the meaning, each cluster c tThe number of online units, startup times, and shutdown times are limited by the total number of units N in the cluster.

[0110]

[0111] In an alternative embodiment, ramp limits, commitment constraints, and link constraints require significant adjustments because the fleet clusters may vary at different time periods.

[0112] In an optional embodiment, when the clustering results change between consecutive time periods, the ramp limit must take into account different unit clusters, as shown in (27)-(30).

[0113] In an optional embodiment, in the worst case, if the load and renewable energy fluctuate greatly in each time period, the clustering results of the units in each time period may be different. However, this situation is extremely rare. In order to ensure that the minimum start and stop time limit is not affected by the change of clustering results, the commitment and link constraints must be applied to the clustering configuration of all units in each time period, as shown in (31)-(35),

[0114]

[0115] in, Is the same type of unit c t A collection of .

[0116] It should be noted that the first unit combination model is updated based on the first clustering strategy and the second clustering strategy to obtain the second unit combination model, which improves the flexibility of the unit combination and can better adapt to changes in the grid load. Through the time clustering strategy, units with similar operating characteristics in a specific time period can be identified to optimize the scheduling plan. The spatial clustering strategy helps to discover the synergy between units with similar geographical locations and further improve the overall operating efficiency. The updated second unit combination model can more accurately predict future electricity demand and reduce the waste of spare capacity. It can also reduce the number of unit starts and stops, extend the service life of equipment, and reduce maintenance costs.

[0117] S103, solving the second unit combination model according to the first improved algorithm.

[0118] In an optional embodiment, the first improved algorithm can be an optimization method based on a genetic algorithm. This method iteratively solves the unit combination model by simulating natural selection and genetics principles in order to achieve a global optimal solution. In the genetic algorithm, each unit combination scheme is regarded as an individual, and new individuals are generated through operations such as selection, crossover and mutation, thereby continuously evolving a better scheduling scheme. In addition, a fitness function is introduced into the algorithm to evaluate the pros and cons of each individual to ensure that excellent unit combination schemes can be retained and passed on to the next generation. In this way, the first improved algorithm can effectively handle complex unit combination problems and improve solution efficiency and quality.

[0119] In an optional embodiment, the first improved algorithm can also be an optimization method based on simulated annealing. This method draws on the principle of the annealing process of solid materials in physics, and searches for the global optimal solution by gradually reducing the "temperature" of the system. In the simulated annealing algorithm, the system starts from an initial state, generates new states through random perturbations, and accepts or rejects these new states according to a certain probability. This probability is related to the "temperature" of the system and the energy difference between the new and old states, thereby allowing the system to jump out of the local optimum and increase the possibility of finding the global optimal solution. As the "temperature" gradually decreases, the system tends to accept states with lower energy and eventually converges to a stable global optimal solution. This method is particularly suitable for solving large-scale and complex unit combination problems, and can effectively avoid falling into local optimal solutions and improve the globality and stability of the solution.

[0120] In an optional embodiment, the first improved algorithm can also use the K-means clustering algorithm. The K-means clustering algorithm is an algorithm widely used in the field of data mining and pattern recognition. It iteratively distributes data points to K clusters to minimize the sum of squares of distances within the cluster. In the unit combination problem, the K-means algorithm can be used to classify the units and classify similar units into the same category, thereby simplifying the complexity of the problem. In this way, the combination optimization of the units can be performed more efficiently, the amount of calculation can be reduced, and the solution speed can be improved. In addition, the parallel processing capability of the K-means algorithm also makes it advantageous when processing large-scale data sets, which helps to improve the performance of the algorithm in practical applications.

[0121] In the embodiment of the present application, solving the second unit combination model according to the first improved algorithm includes:

[0122] Establishing a second unit commitment model based on the first target factor and the second clustering strategy;

[0123] solving the second unit combination model according to the first improved algorithm;

[0124] The first improved algorithm is an arbitrary clustering algorithm that operates by configuring a similarity threshold.

[0125] In the embodiment of the present application, the hierarchical K-means clustering method improved by the Trust-Tech technology is used as the first improved algorithm.

[0126] In summary, the present invention proposes a unit combination method based on time and space clustering to obtain the first data of the target power system, and establish a first unit combination model according to the first data; update the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain the second unit combination model; solve the second unit combination model according to the first improved algorithm. Through the time and space clustering technology, the complexity of the unit combination model is effectively reduced and the calculation efficiency is improved. The uncertainty and variability of renewable energy are taken into account, and the operational flexibility of the power system is enhanced. By introducing the improved algorithm, the solution process of the unit combination is optimized to ensure the feasibility and economy of the planning scheme. The unit combination method and system of the present invention can adapt to the changing new participants and operating conditions in the power system, and provide strong support for the efficient operation of the power system. By reducing the number of binary variables, the calculation burden of the NCUC model is reduced, making the modeling of the medium and long-term power system operation more efficient. The unit combination method and system of the present invention not only ensure the operational flexibility of the power system, but also take into account the network constraints, ensuring the stability and reliability of the power system.

[0127] Example 2

[0128] This embodiment also provides a unit combination system based on time and space clustering, including:

[0129] A model building module, used for acquiring first data of a target power system and building a first unit combination model according to the first data;

[0130] A model updating module, used for updating the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain a second unit combination model;

[0131] The solution module is used to solve the second unit combination model according to the first improved algorithm.

[0132] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0133] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 7As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a unit combination method based on time and space clustering is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse, etc.

[0134] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0135] Acquire first data of the target power system, and establish a first unit combination model according to the first data;

[0136] The first unit combination model is updated based on the first clustering strategy and the second clustering strategy to obtain a second unit combination model;

[0137] The second unit combination model is solved according to the first improved algorithm.

[0138] Example 3

[0139] Reference Figure 4-Figure 6 , which is an embodiment of the present invention, provides a unit combination method and system based on time and space clustering. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0140] In the present embodiment, a modified 72-node system is used to evaluate the performance of the proposed method. The system is composed of three IEEE-RTS-24 node systems and connected 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 4 In addition, in order to alleviate the symmetry problem, a random +1% noise is introduced into the operating cost of the unit during the unit commitment solution process.

[0141] In the embodiment of the present application, in order to quantify the effectiveness of the proposed method, the present application calculates multiple effectiveness indicators to evaluate the solution effect of the unit commitment problem. The result of the original network constraint unit commitment model is represented by the symbol "^", as follows:

[0142] 1. Total system cost deviation

[0143] The percentage difference in the objective function value is calculated as follows:

[0144]

[0145] 2. Unit status error

[0146] The percentage difference in the planned on / off status of each unit is calculated as follows:

[0147]

[0148] 3. Transmission power flow error

[0149] The power flow percentage error for each transmission line is calculated as follows:

[0150]

[0151] 4. Unit combination calculation time

[0152] The unit commitment calculation time is set to the running time of the solver (the GUROBI solver is used in this test) to solve the unit commitment problem.

[0153] 5. Total calculation time

[0154] The total computing time includes the unit combination computing time, as well as the total running time of other operations such as preprocessing and clustering.

[0155] In an embodiment of the present application, three different time clustering methods were tested on a data set containing 7 days to solve the computational effect of the network constrained unit combination problem, and no spatial clustering method was used. The first method is marked as "M1", which uses K-means to cluster the load and renewable energy sequences. The second method is marked as "M2", which adds the ramping condition of the unit to the clustering target based on M1. The third method, which is the method proposed in this application, is marked as "M3", which uses the hierarchical K-means clustering method improved by Trust-Tech technology to cluster the ramping conditions of loads, renewable energy and units.

[0156] In the examples of this application, the comparison results of the three methods are as follows Figure 5As shown. Fewer representative days will result in faster calculations, however, this situation usually increases the error of the objective function. In addition, for the same number of representative days, different clustering methods will result in different calculation times and objective function errors. For two and six representative days, M2 and M3 produce the same clustering results, while Ml and M2 also produce the same clustering results in five representative days. Therefore, in these cases, the objective function is the same, and the difference in total calculation time stems from the clustering method used. Although the K-means method introduces some randomness, M2, which takes into account the running state, usually improves the calculation accuracy over Ml. Among these methods, the hierarchical K-means clustering method improved by Trust-Tech technology achieves better clustering results, allowing M3 to achieve the lowest objective function error in most cases.

[0157] In an embodiment of the present application, three different spatial clustering methods are tested on the same data containing 7 days to solve the computational effect of the network constrained unit combination problem, and no time clustering method is used. The first method is marked as "M4", which clusters the units only according to the physical parameters of the units. The second method is marked as "M5", which is based on M4 and further subdivides the clustering results of the units according to the flow results of the unit combination under specific load conditions. The third method, which is the method proposed in the present application, is marked as "M6". It is based on M4 and subdivides the clustering results of the units based on the power distribution factors of each unit. The power distribution factor deviation threshold is set to ε=0.15. The specific results are shown in Table 1.

[0158] Table 1 Comparison of results of network-constrained unit commitment problem under different spatial clustering methods

[0159]

[0160]

[0161] It should be noted that these results show that: 1) M4 clusters all units into 9 categories, achieving the maximum reduction in integer variables and the fastest calculation speed. However, it also has relatively large errors in the objective function, unit status, and line flow. 2) M5 further refines the clustering by clustering the units into 33 categories and verifies it through line flow under specific load conditions. This method produces smaller errors than M4. Although it takes a little longer, the preprocessing time is longer due to the need to additionally calculate the network constraint unit combination problem under each load condition. 3) The proposed M6 further clusters the units according to the power distribution factors in different time periods, and obtains four different unit clustering results. Although the calculation time is the longest due to the addition of additional constraints due to the need to manage the unit clustering that changes in different time periods, it also provides the most accurate solution.

[0162] In the embodiments of the present application, different spatial clustering methods may result in different levels of power flow calculation errors. Figure 6 The percentage deviation of the line power flow of the three spatial clustering methods from the original network-constrained unit commitment model is shown for a specific period of time in an IEEE-RTS-24 node area. Among these methods, M4 shows the largest power flow deviation. Although M5 generally produces smaller errors than M4, it may produce larger errors on some lines under certain specific load conditions. This is because the specific load conditions selected by M5 do not fully cover the test cases. In contrast, M6 adjusts the results of unit clustering according to the power distribution factor, resulting in the smallest power flow error. These results clearly show that the results of spatial clustering should change with changes in system operating conditions.

[0163] In the embodiment of the present application, one year of data is selected to test the effectiveness of the spatiotemporal clustering method, and the clustering is set to 30 representative days. The proposed method, namely M3+M6, is compared with the existing method combining temporal and spatial clustering, namely M1+, and the results are shown in Table 2.

[0164] Table 2 Comparison of results of different spatiotemporal clustering methods in network-constrained unit commitment problems

[0165]

[0166] It should be noted that these numerical results show that: 1) the hybrid time and space clustering method can significantly improve the computational speed of the network-constrained unit commitment problem. 2) although the proposed method, M3+M6, is slightly slower than the existing method, M1+M4, the proposed method is extremely accurate and greatly reduces the objective function error compared to the original network-constrained unit commitment model.

[0167] In summary, for the network-constrained unit combination problem, this application proposes a hybrid spatiotemporal clustering method, which has the following key features: 1) The time clustering method takes into account the load, renewable energy and unit ramping conditions to better capture the characteristics of different time periods, and uses the hierarchical K-means clustering algorithm improved by Trust-Tech technology to obtain better clustering results; 2) The spatial clustering method calculates the power distribution factors of different units in each time period based on the results of time clustering, so as to cluster the units more accurately, making up for the traditional clustering based only on the physical parameters of the units, but lacks consideration of the unit operation conditions. 3) The hybrid spatiotemporal clustering method can effectively improve the computational efficiency of the network-constrained unit combination problem and obtain a high-quality result.

[0168] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0169] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0170] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0171] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0172] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1A step that specifies a function in one or more boxes.

[0173] It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation manner. Any other implementation manners derived by those skilled in the art based on the technical solution of the present invention also fall within the scope of protection of the present invention.

Claims

1. A unit combination system based on time and space clustering, characterized in that: It includes a model building module, a model updating module and a solving module, wherein the model building module, the model updating module and the solving module are connected in sequence, the model building module is used to obtain the first data of the target power system, and establish a first unit combination model according to the first data; the model updating module is used to update the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain the second unit combination model; the solving module is used to solve the second unit combination model according to the first improved algorithm.

2. A method for combining unit combination systems based on time and space clustering as claimed in claim 1, characterized in that: The following steps are involved: Step 1: The model building module obtains first data of the target power system and builds a first unit combination model according to the first data; Step 2: The model updating module updates the first unit combination model based on the first clustering strategy and the second clustering strategy to obtain a second unit combination model; Step 3: The solving module solves the second unit combination model according to the first improved algorithm to obtain a unit combination result.

3. The method for combining unit combination systems based on time and space clustering according to claim 2, characterized in that: The first unit combination model in step 1 includes a first objective function and a first constraint condition, wherein the first objective function includes unit startup and shutdown costs, operating costs, wind power reduction penalties, and load reduction; the first constraint condition includes output level and load reduction constraints, safety constraints, decision constraints, and time constraints; Among them, the combination model of the first unit is: Among them, C Cost is the total system cost, g, w, n, t are index items of unit, wind farm, load and time period respectively; P g,t ,P w,t and D n,t are continuous variables, corresponding to the power generation of the generator, the dispatching amount of the wind farm and the load reduction; x g，t , and are integer variables, representing the start / stop state and start / stop state of the unit respectively; C w and C l is the operating cost coefficient; and is the hourly forecast of wind power generation and load; G, W, N and T are the total number of units, wind farms, loads and time periods respectively; w t is the weight of the time period after clustering; and is the maximum / minimum output limit of the unit; F l,t ,, and θ n,t They represent the power flowing through the line, the line susceptance and the phase angle of the node respectively; F l max Indicates the maximum power allowed to flow through the line; and They are the collection of the unit, wind farm, and the beginning and end of the line; RU g,t ,RD g,t ,SU g,t and SD g,t They are the up and down ramp limits and the minimum start and shutdown time of the unit; TU g and TD g They are the minimum power on / off time limit; RP 1 ,and are the subsets of the first and smallest on / off periods for each representative day, respectively; and The time periods are two consecutive representative days.

4. The method for combining unit combination systems based on time and space clustering according to claim 2, characterized in that: The step 2 comprises the following steps: Step 2.1, performing a first update operation on the first unit combination model according to the first clustering strategy; Step 2.2, selecting the first target factor after the first update operation; Step 2.3: Perform a second update operation using a second clustering strategy based on the first objective factor to obtain a second unit commitment model.

5. The method for combining unit combination systems based on time and space clustering according to claim 4, characterized in that: The specific implementation method of the first update operation in step 2.1 is: Step 2.1.1, updating the first constraint condition in the first unit combination model; Step 2.1.2, performing a first clustering solution on the first unit commitment model updated by the first updating operation according to the first improved algorithm; Step 2.1.3: Select the first target factor after the first update operation according to the first clustering solution result.

6. The method for combining unit combination systems based on time and space clustering according to claim 5, characterized in that: The clustering targets of the first clustering strategy are load, renewable energy forecast curve and unit ramping condition; the clustering targets of the second clustering strategy are unit physical parameters, and the first target factor is the power distribution factor.

7. The method for combining unit combination systems based on time and space clustering according to claim 4, characterized in that: The specific implementation method of the second updating operation in step 2.3 is: updating the first constraint condition in the first unit commitment model after the first clustering strategy is updated again.

Images