Method, device, equipment and medium for predicting operating parameters of power unit combination

Abstract

The embodiment of the present application discloses a method, device, equipment, and medium for predicting the operating parameters of a power unit combination, which belongs to the technical field of power systems, including: in the current iteration round, using a machine learning model to perform feature prediction on the unit operating characteristic data of a unit subset in all time periods, and obtain preliminary operating parameters of the unit subset in the current power unit submodel in each time period; determining the unit operating characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operating parameters of the unit subset in the current power unit submodel in each time period; performing feature prediction based on the unit operating characteristic data of the next power unit submodel in all time periods; using the preliminary operating parameters of the unit subset in all power unit submodels of the last iteration round in each time period as the target operating parameters of the power unit combination in each time period, thereby improving the prediction efficiency of the operating parameters of the power unit combination.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a method, device, equipment, and medium for predicting operating parameters of a power unit combination. Background Art

[0002] The Unit Commitment Problem (UCP) is a core issue in the optimal dispatch of power systems. In related technologies, this problem is usually modeled as a mixed integer programming problem, and an operations optimization algorithm is used to solve the mixed integer programming problem to predict the operating parameters of each power unit in the power system in each time period, such as operating status, unit output, etc. In order to cope with the growing demand for electricity, the scale of the power system has gradually expanded, causing the scale of the mixed integer programming problem to increase dramatically. At this time, if an operations optimization algorithm is used to solve the unit commitment problem, the prediction efficiency of the power unit combination operating parameters will decrease. Summary of the invention

[0003] The main purpose of the embodiments of the present application is to propose a method, device, equipment, and medium for predicting the operating parameters of a power unit combination, aiming to improve the prediction efficiency of the operating parameters of the power unit combination.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application provides a method for predicting operating parameters of a power unit combination, the method comprising:

[0005] Acquire an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints;

[0006] Relaxing the operating constraints of the unit, and determining a target power unit model according to the relaxed operating constraints of the unit and the operating cost function;

[0007] Dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset;

[0008] In the current iteration round, each power unit sub-model is traversed in turn to obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, and the unit operation characteristic data of the unit subset in all time periods are predicted by the machine learning model to obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period;

[0009] Determine the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and perform characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in the last power unit submodel in each time period are obtained;

[0010] In the next iteration round, the preliminary operating parameters of the unit subsets in all power unit submodels in the current iteration round in each time period are updated until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels in the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

[0011] In some embodiments, the unit operation constraints include load balance constraints and line transmission safety constraints, and the step of relaxing the unit operation constraints and determining the target power unit model according to the relaxed unit operation constraints and the operation cost function includes:

[0012] The load balancing constraint condition is relaxed respectively by using the Lagrange multiplier and the first penalty coefficient;

[0013] Performing constraint relaxation on the line transmission safety constraint condition by using a first non-negative continuous variable and a second non-negative continuous variable;

[0014] Update the operation cost function according to the relaxed load balancing constraint, the first non-negative continuous variable, the second non-negative continuous variable and the second penalty coefficient;

[0015] The updated operation cost function and the relaxed line transmission safety constraint condition are used as the target power unit model.

[0016] In some embodiments, the unit operation characteristic data includes the unit operation status and unit output of the unit subset in the initial period, and the unit operation characteristic data of the unit subset in all periods are predicted by the machine learning model to obtain the preliminary operation parameters of the unit subset in each period in the current power unit submodel, including:

[0017] Characteristic encoding is performed on the unit operation characteristic data of the unit subset in all time periods to obtain forward unit operation characteristics and backward unit operation characteristics;

[0018] The forward unit operation characteristics, the backward unit operation characteristics, the unit operation status of the unit subset in the initial period and the unit output are feature decoded to obtain the preliminary operation parameters of the unit subset in each period in the current power unit submodel.

[0019] In some embodiments, the forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period are feature decoded to obtain the preliminary operation parameters of the unit subset in each period in the current power unit submodel, including:

[0020] Performing attention transformation on the forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period to obtain the attention characteristics of the current period;

[0021] Perform gated feature decoding on the attention feature of the current period to obtain the first reference operating parameter of the unit subset in the current power unit sub-model in the current period;

[0022] Performing a feasibility test on the first reference operating parameter of the current period, and determining a second reference operating parameter of the current period according to a test result of the feasibility test and the first reference operating parameter of the current period;

[0023] Performing attention transformation on the second reference operating parameter of the current period, the forward unit operating characteristic, the backward unit operating characteristic, the unit operating state and the unit output of the unit subset in the initial period to obtain the attention characteristic of the next period;

[0024] The attention features of the next period are gated feature decoded until the second reference operating parameters of the unit subset in the last period are obtained, and the second reference operating parameters of each period are used as the preliminary operating parameters of each period.

[0025] In some embodiments, the power unit sub-model includes a sub-operation constraint, and performing a feasibility test on the first reference operating parameter of the current period, and determining the second reference operating parameter of the current period according to the test result of the feasibility test and the first reference operating parameter of the current period, comprises:

[0026] Performing a preliminary feasibility test on the first reference operation parameter of the current period according to the sub-operation constraints;

[0027] If the test result of the preliminary feasibility test is that the preliminary feasibility test is passed, the unit optimization constraint conditions of the current power unit submodel are obtained;

[0028] Re-checking the feasibility of the first reference operating parameter of the current period according to the unit optimization constraint condition;

[0029] The second reference operating parameters of the current period are determined according to the detection result of the re-feasibility detection and the first reference operating parameters of the current period.

[0030] In some embodiments, determining the second reference operating parameter of the current period according to the detection result of the re-feasibility detection and the first reference operating parameter of the current period includes:

[0031] If the result of the re-feasibility test is that the re-feasibility test is passed, the first reference operating parameter of the current period is used as the second reference operating parameter of the current period;

[0032] If the result of the re-feasibility test is that the re-feasibility test fails, the first reference operating parameters of the current period are optimized by a branch cutting algorithm to obtain the second reference operating parameters of the current period.

[0033] In some embodiments, after performing a preliminary feasibility check on the first reference operating parameter of the current period according to the sub-operation constraints, the method further includes:

[0034] If the test result of the preliminary feasibility test is that the preliminary feasibility test fails, the first reference operating parameters of the current period are optimized by a branch cutting algorithm to obtain the second reference operating parameters of the current period.

[0035] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides an operating parameter prediction device for a power unit combination, the device comprising:

[0036] An acquisition module, used for acquiring an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints;

[0037] A relaxation module, used for relaxing the operating constraints of the unit, and determining a target power unit model according to the relaxed operating constraints of the unit and the operating cost function;

[0038] A decomposition module, used for dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset;

[0039] The first prediction module is used to traverse each power unit sub-model in turn in the current iteration round, obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, perform feature prediction on the unit operation characteristic data of the unit subset in all time periods through the machine learning model, and obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period;

[0040] The second prediction module is used to determine the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and perform characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in the last power unit submodel in each time period are obtained;

[0041] The updating module is used to update the preliminary operating parameters of the unit subsets in all power unit submodels of the current iteration round in each time period in the next iteration round until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels of the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

[0042] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the operating parameter prediction method of the power unit combination described in the first aspect above.

[0043] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the operating parameter prediction method of the power unit combination described in the first aspect above.

[0044] The operating parameter prediction method of the power unit combination, the operating parameter prediction device of the power unit combination, the electronic device and the computer-readable storage medium proposed in the embodiment of the present application obtain the initial power unit model of the power unit combination to use the initial power unit model to predict the operating parameters of the power unit combination. In order to reduce the model scale of the initial power unit model and improve the prediction efficiency of the operating parameters, the initial power unit model is decomposed to divide the unit combination problem into multiple sub-problems. In the decomposition process of the unit combination problem, the unit operation constraints are relaxed, and the target power unit model is determined according to the relaxed unit operation constraints and the operating cost function, so as to determine the relaxed unit combination problem according to the target power unit model. In order to improve the efficiency of solving the unit combination problem, the power unit combination is divided into multiple unit subsets, and the target power unit model is decomposed into multiple power unit sub-models according to each unit subset. The power unit sub-model will involve multiple variables and complex constraints. In order to further improve the efficiency of solving the unit combination problem, a machine learning model is introduced to solve the power unit sub-model. In the current iteration round, the machine learning model is used to predict the unit operation characteristic data of the unit subset in all time periods, and the preliminary operation parameters of the unit subset in the current power unit submodel in each time period are obtained. In the power system, the collaborative work between different unit subsets is usually required to ensure the stability of the system. In order to improve the accuracy of the operation parameter prediction, the correlation between the preliminary operation parameters of different power unit submodels is considered. The unit operation characteristic data of the unit subset in each time period in the next power unit submodel is determined according to the preliminary operation parameters of the unit subset in each time period in the current power unit submodel, and the characteristic prediction is performed according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in each time period in the last power unit submodel are obtained. In the next iteration round, the preliminary operation parameters of the unit subset in each time period in all power unit submodels of the current iteration round are updated, and the preliminary operation parameters are gradually optimized through multiple iterations to improve the accuracy of the prediction of the power unit operation parameters. The power unit combination includes multiple unit subsets. The operating parameters of the power unit combination are determined by the operating parameters of each unit subset until the last iteration round is reached. The preliminary operating parameters of the unit subsets in all power unit sub-models in the last iteration round are used as the target operating parameters of the power unit combination in each period. By leveraging the powerful feature learning ability of the machine learning model, each power unit sub-model can be quickly solved, thereby improving the prediction efficiency of the power unit combination operating parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of a method for predicting operating parameters of a power unit combination provided in an embodiment of the present application;

[0046] Figure 2 yes Figure 1 Flow chart of step S120 in FIG.

[0047] Figure 3 yes Figure 1 Flow chart of step S140 in FIG.

[0048] Figure 4 yes Figure 3 Flow chart of step S320 in FIG.

[0049] Figure 5 yes Figure 4 Flow chart of step S430 in FIG.

[0050] Figure 6 yes Figure 5 Flow chart of step S540 in FIG.

[0051] Figure 7 is a schematic diagram of the structure of a vehicle control device provided in an embodiment of the present application;

[0052] Figure 8 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0054] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0056] The unit combination problem is a core issue in the optimal dispatching of power systems. In related technologies, this problem is usually modeled as a mixed integer programming problem, and an operations optimization algorithm is used to solve the mixed integer programming problem to predict the operating parameters of each power unit in the power system in each period, such as operating status, unit output, etc. In order to cope with the growing demand for electricity, the scale of the power system has gradually expanded, causing the scale of the mixed integer programming problem to increase dramatically. At this time, if an operations optimization algorithm is used to solve the unit combination problem, the prediction efficiency of the power unit combination operating parameters will decrease.

[0057] Based on this, the embodiments of the present application provide a method for predicting operating parameters of a power unit combination, a device for predicting operating parameters of a power unit combination, an electronic device and a computer-readable storage medium, aiming to improve the prediction efficiency of the operating parameters of the power unit combination.

[0058] The operating parameter prediction method of the power unit combination, the operating parameter prediction device of the power unit combination, the electronic device and the computer-readable storage medium provided in the embodiments of the present application are specifically explained through the following embodiments. First, the operating parameter prediction method of the power unit combination in the embodiments of the present application is described.

[0059] The operating parameter prediction method of the power unit combination provided in the embodiment of the present application relates to the technical field of power system. The operating parameter prediction method of the power unit combination provided in the embodiment of the present application can be applied to the terminal, can also be applied to the server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or it can be configured as a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the operating parameter prediction method of the power unit combination, etc., but is not limited to the above forms.

[0060] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0061] The unit commitment problem is a key issue in power system dispatching. This problem is usually modeled as a mixed integer programming problem, and considers how to reasonably arrange the start and stop and output of generators in hours to optimize the load demand and reduce operating costs. By solving the unit commitment problem, the operating efficiency of the power system can be optimized to ensure that the system operates in an efficient and economical manner.

[0062] With the increasing share of renewable energy such as wind and solar energy, the growing uncertainty and load volatility have brought great challenges to the safe and stable operation of the power system. In order to enhance the flexibility and reliability of the system to cope with the higher frequency of load and renewable energy supply fluctuations in the future. Sub-hourly unit combination is becoming the main dispatching decision model in the new power system, and it is also a very important part of the energy system with new energy as the main body. Sub-hourly unit combination usually makes dispatching decisions on unit combination on a time scale of 15 minutes or 30 minutes, which makes the decision frequency of unit combination problem higher, the time interval shorter, and the number of variables and constraints involved is more, resulting in a sharp increase in the scale of the problem. In addition, the relative reduction of the climbing ability of the unit in each time period leads to the enhancement of the dependency between the coupled time periods, which further increases the complexity of the problem.

[0063] For the unit combination problem, the traditional method uses operations research optimization algorithms to solve the operating parameters of each generator set at different times. This type of method has a relatively complete theoretical basis and proof reasoning, which can ensure the convergence of the algorithm, so that the solution of the problem can obtain the optimal solution or the approximate optimal solution. However, due to the characteristics of combinatorial optimization problems, that is, the complexity of the problem increases exponentially with the increase of the problem scale, the use of traditional operations research methods to solve large-scale sub-hourly unit combination problems will face problems such as long solution time and slow algorithm convergence, which cannot meet the actual operation and control requirements of the power system. The rapid development of artificial intelligence technology, especially machine learning algorithms, has brought new possibilities for the rapid solution of this problem. After completing offline training, machine learning algorithms can provide prediction results in seconds or even milliseconds for online problems. However, existing machine learning algorithms are usually used to deal with unconstrained classification or regression problems, and do not have the ability to solve problems with constraints. When machine learning algorithms are used to predict the results of problems with constraints, they often produce prediction values ​​that violate the constraints of the original problem, causing the problem to fail to be solved, and it is necessary to use traditional operations research optimization algorithms or heuristic methods such as intelligent optimization to solve it again.

[0064] There are three main types of operations research optimization algorithms: heuristic, branch-and-cut, and decomposition-coordination-based methods. Heuristic methods are generally used for small-scale problems or as auxiliary methods for the other two methods. Branch-and-cut methods are widely used in commercial solvers such as CPLEX and Gurobi. However, when faced with large-scale problems or unit combination problems with complex structures, such as the sub-hourly unit combination problem mentioned above, the performance of this method depends on time-consuming branch-and-bound and heuristic methods to search for approximate optimal solutions. In the face of large-scale complex unit combination problems, the decomposition-coordination method achieves exponential degradation of the complexity of the problem through decomposition. Among the many decomposition-coordination methods, the Lagrangian relaxation method is one of the main methods for solving unit combination problems in the early days. However, it has the disadvantages of slow convergence speed and high coordination requirements. Since then, this type of method has been continuously improved, and methods such as the proxy Lagrangian relaxation algorithm and the proxy absolute value Lagrangian relaxation algorithm have appeared successively. Although this type of method has made great breakthroughs in solution efficiency and algorithm convergence speed, its performance is not stable enough when faced with large-scale sub-hourly unit combination problems. In addition, when faced with problems with a high degree of coupling, the solution time of this method may still not be able to meet the near real-time response requirements of actual power system scheduling.

[0065] Artificial intelligence methods are mainly divided into two types of strategies: direct and indirect. Direct strategies refer to the use of various machine learning models to directly learn the decision-making solutions for unit commitment problems. Existing technologies have tried to use classic and advanced machine learning models such as K-nearest neighbor, support vector machine, random forest, graph neural network, gate loop control and reinforcement learning. These direct strategies can usually obtain fast and relatively accurate unit commitment decisions on small-scale problems. However, when the scale of the problem increases, it is difficult to provide high-quality feasible predictions stably. Indirect strategies refer to the use of machine learning technology to simplify problem modeling or to assist in accelerating traditional optimization algorithms. For example, contrastive learning is used to simulate the local branching process in large neighborhood search, or classification machine learning models such as K-nearest neighbor are used to identify stable integer variables (such as long-term unchanged unit switch states) or inactive constraints (such as line transmission safety constraints that will not be violated under maximum power generation scenarios). These indirect strategies can simplify the original problem model or accelerate the solution performance of traditional algorithms to a certain extent, but when faced with unit commitment problems in unknown scenarios or complex situations, their effectiveness is not significant and is highly dependent on the performance of traditional algorithms.

[0066] In summary, when relying on existing methods to deal with sub-hourly unit commitment problems, traditional operations optimization algorithms and artificial intelligence methods cannot directly meet the flexibility and reliability requirements of the power system for solving problems. Therefore, the embodiment of the present application comprehensively improves the efficiency and convergence speed of solving unit commitment problems by customizing the decomposition and coordination optimization strategy and integrating machine learning technology.

[0067] Figure 1 is an optional flow chart of the method for predicting operating parameters of a power unit combination provided in an embodiment of the present application, Figure 1 The method may include but is not limited to steps S110 to S160.

[0068] Step S110, obtaining an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints;

[0069] Step S120, relaxing the unit operation constraints, and determining the target power unit model according to the relaxed unit operation constraints and the operation cost function;

[0070] Step S130, dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset;

[0071] Step S140, in the current iteration round, traverse each power unit sub-model in turn, obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, perform feature prediction on the unit operation characteristic data of the unit subset in all time periods through the machine learning model, and obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period;

[0072] Step S150, determining the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and performing characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods, until obtaining the preliminary operation parameters of the unit subset in the last power unit submodel in each time period;

[0073] Step S160, in the next iteration round, the preliminary operating parameters of the unit subsets in all power unit submodels in the current iteration round in each time period are updated until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels in the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

[0074] In step S110 of some embodiments, an initial power unit model of the power unit combination in the power system is obtained, and the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints. The operation cost function is the objective function of the unit combination problem. The power unit combination includes multiple power units (generator sets), and one power unit includes multiple power modules. The operation cost function is used to minimize the total operation cost of all power units. The specific formula is as follows:

[0075] ,

[0076] Wherein, i represents the i-th power unit; I represents the number of power units in the power system; t represents the tth time period; T represents the number of time periods; represents the initial startup cost of the ith power unit in the tth period; represents the startup status of the i-th power unit in the t-th period; represents the no-load cost of the i-th power unit in the t-th period; represents the operating status of the i-th power unit in the t-th period; b represents the b-th power module, and B represents the number of power modules in the i-th power unit; represents the power generation cost of the bth power module of the i-th power unit in the t-th period; It represents the module output of the bth power module of the i-th power unit in the t-th time period.

[0077] It should be noted that no-load cost refers to the additional operating cost of the unit idling and not participating in the relevant power generation process. The unit startup status is a 0-1 variable. The unit startup status of 1 indicates that the power unit performs the startup action and the power unit starts to change from the offline shutdown state to the operating state; the unit startup status of 0 indicates that the power unit is stably in the offline shutdown state or the online operating state. The unit operating status is a 0-1 variable. The unit operating status of 1 indicates that the power unit is in the operating state. The unit operating status of 0 indicates that the power unit is in the off state, that is, the offline shutdown state. The unit output of the i-th power unit in the t-th time period is the sum of the module outputs of all power modules in the i-th power unit in the t-th time period, which can be expressed as:

[0078] ,

[0079] in, It represents the output of the i-th power unit in the t-th time period.

[0080] Unit operation constraints refer to the operating restrictions that power units must follow during operation to ensure that the power units can operate safely and stably. Unit operation constraints can be constraints at the power system level or at the unit level. Power system-level constraints include load balance constraints and line transmission safety constraints. Unit-level constraints can be appropriately added according to actual application scenarios, such as output power constraints, ramp rate constraints, start and stop time constraints, etc. In the load balance constraint, the sum of the unit power generation levels is equal to the sum of the node power demands, expressed as:

[0081] ,

[0082] in, Indicates the power consumption node n In the period t The power demand; N represents the number of power consuming nodes.

[0083] The line transmission safety constraint condition is used to constrain the power transmission line to be within a certain safe transmission capacity in each period. The power transmission line is a high-voltage transmission line at the power system transmission network level, and the safe transmission capacity is the maximum power that the power transmission line can withstand. The line transmission safety constraint condition is expressed as:

[0084] ,

[0085] ,

[0086] in, Indicates power transmission line DC flow in time period t; Represents the transfer factor, which is used to measure the power transmission line of the power consumption node n l The influence degree of DC flow; Indicates power transmission line l The set of power units that supply power to the power node n; and Indicates power transmission line l The transmission capacity limit Indicates the minimum transmission capacity limit, Indicates the maximum transmission capacity limit.

[0087] In order to reduce the problem scale of unit problems and improve the efficiency of solving the operating parameters of unit combination problems, the embodiment of the present application first uses the proxy absolute value Lagrangian relaxation algorithm to decompose the unit combination problem into multiple sub-problems, and then uses a machine learning model that combines the gated recurrent unit (GRU) and the attention mechanism to solve the sub-problems. It aims to use the traditional decomposition and coordination framework to ensure convergence and optimality of prediction results, while integrating machine learning methods to maximize the speed of solving sub-problems based on the data-physics dual-driven approach, thereby achieving stable, fast and efficient optimization solutions for unit combination problems such as sub-hourly unit combination problems to support the safe and stable operation of the power system.

[0088] See also Figure 2 In some embodiments, step S120 may include but is not limited to steps S210 to S240:

[0089] Step S210, relaxing the load balancing constraint condition by using the Lagrange multiplier and the first penalty coefficient respectively;

[0090] Step S220, relaxing the line transmission safety constraint condition by using the first non-negative continuous variable and the second non-negative continuous variable;

[0091] Step S230, updating the operation cost function according to the relaxed load balance constraint, the first non-negative continuous variable, the second non-negative continuous variable and the second penalty coefficient;

[0092] Step S240: Using the updated operation cost function and the relaxed line transmission safety constraint as the target power unit model.

[0093] In step S210 of some embodiments, in order to achieve the decomposition of the unit combination problem, the unit operation constraints are relaxed so that the unit operation constraints are changed from hard constraints to soft constraints, that is, the constraints are allowed to be violated, but with very high violation penalties. The load balance constraints are relaxed using Lagrange multipliers, and the violation of the load balance constraints by the power unit combination is penalized using a first penalty coefficient. The first penalty coefficient uses an adjustable penalty coefficient. The relaxed load balance constraints are expressed as:

[0094] ,

[0095] in, is the Lagrange multiplier of the tth period; is the first penalty coefficient.

[0096] In step S220 of some embodiments, two non-negative continuous variables, namely, a first non-negative continuous variable and a second non-negative continuous variable, are introduced to relax the line transmission safety constraint condition. The relaxed line transmission safety constraint condition is expressed as:

[0097] ,

[0098] ,

[0099] in, represents the first non-negative continuous variable; Represents the second non-negative continuous variable.

[0100] In step S230 of some embodiments, the second penalty coefficient is used to penalize the violation of the line transmission safety constraint condition by the power unit combination, and the second penalty coefficient adopts a fixed penalty coefficient. The operation cost function is updated according to the relaxed load balance constraint condition, the first non-negative continuous variable, the second non-negative continuous variable and the second penalty coefficient, and the updated operation cost function is expressed as:

[0101] ,

[0102] ,

[0103] Where L represents the total operating cost of the unit combination problem after relaxation; Represents the second penalty coefficient.

[0104] In step S240 of some embodiments, the load balance constraint is deleted from the relaxed constraints because it has been considered in the new objective function, and the other constraints remain unchanged. The updated operation cost function and the relaxed line transmission safety constraint are used as the target power unit model.

[0105] In the above steps S210 to S240, the unit combination problem is relaxed through a customized decomposition and coordination optimization strategy to relax the constraints and objective function of the unit combination problem, thereby simplifying the unit combination problem, making it easier to solve, and improving the efficiency of solving the unit combination problem.

[0106] In step S130 of some embodiments, J relatively uniform regions are divided according to the location distribution of the power units, and for each region, the unit subset of the region is determined according to all the power units in the region. The target power unit model is decomposed into J power unit sub-models according to the J unit subsets, and each power unit sub-model corresponds to a unit subset. Thus, the relaxed unit combination problem can be divided into J sub-problems, and the unit set of each sub-problem is all the units in the region j. The power unit sub-model includes a sub-operation cost function and a sub-operation constraint. After decomposition, the sub-problem is iteratively solved, and the objective function of the sub-problem j of the kth iteration, that is, the sub-operation cost function of the power unit sub-model, is expressed as:

[0107] ,

[0108] ,

[0109] Wherein, k represents the kth iteration; j represents the jth power unit sub-model; represents the subset of units belonging to the jth power unit submodel (subproblem j); Represents other unit subsets that do not belong to the jth power unit sub-model.

[0110] Power flow in relaxed line transmission security constraints It is expressed as:

[0111] ,

[0112] in, for and The intersection of l The set of units connected to the power consumption node n; According to the area where the power units are located, According to the power transmission line l The connection position division; for and The intersection of l The set of units connected to the power node n.

[0113] Substituting the above power flow representation into the relaxed line transmission safety constraints can obtain the sub-operation constraints.

[0114] It should be noted that in each sub-problem, only and The output level of the units in the intersection of is controllable, and the units that do not belong to this intersection, that is, and The intersection of will use the values ​​obtained from solving other subproblems in the previous iteration to calculate the value of the power flow.

[0115] In the current iteration round, each power unit submodel is traversed in turn to perform rolling solutions on the power unit submodel, and the unit operation characteristic data of the unit subset in the current power unit submodel in each period is obtained. The unit operation characteristic data is an input data sequence, which contains all the necessary elements required to solve the current power unit submodel. For subproblem j, the unit operation characteristic data of the unit subset in the kth iteration in each period is expressed as:

[0116] ,

[0117] Among them, the unit operation characteristic data of time period t includes the unit operation characteristic and the given initial unit states of all power units (unit subsets) belonging to subproblem j , It represents the unit operation characteristics of the unit subset associated with the current power unit sub-model j solution in time period t. The unit operation characteristics include Lagrange multiplier, first penalty coefficient, second penalty coefficient, sum of unit outputs belonging to other sub-problems in the previous iteration round k-1, power transmission line l The sum of the transfer factors of all power nodes n and the sum of the unit outputs of the set of units that belong to other sub-problems and are connected to the power node n in the previous iteration round k-1. That is:

[0118] ,

[0119] in, ; represents the Lagrange multiplier; represents the first penalty coefficient; represents the second penalty coefficient; represents the sum of the unit outputs belonging to other sub-problems in the previous iteration round k-1; Indicates power transmission line l The sum of the transfer factors of all power consumption nodes n; It represents the sum of the unit outputs of the set of units that belong to other sub-problems and are connected to the power node n in the previous iteration round k-1.

[0120] The given initial unit state is the initial value of the system state given before solving the problem. It is a constant during the solution process. The given initial unit state is expressed as:

[0121] ,

[0122] ,

[0123] in, represents the i-th power unit in the unit subset belonging to the power unit sub-model j; represents the average unit power generation cost of power unit i; represents the initial operating state of the power unit i, that is, the operating state of the power unit i in the initial period 0; represents the initial power level of power unit i, that is, the unit output of power unit i in the initial period 0; Indicates the continuous startup time of power unit i; Indicates the continuous shutdown time of power unit i.

[0124] Different sub-problems contain different numbers of power units, so the length of a given initial unit state will be different. In this embodiment of the application, unused positions in the given initial unit state are filled with 0, so that The length of will be fixed to a sufficiently large value, and the input sequence can maintain a general structure without being affected by the number of specific sub-problem units, which improves the general ability of the model to solve different sub-problems.

[0125] The machine learning model is used to predict the characteristics of the unit operation characteristic data of the unit subset in all time periods, and the preliminary operating parameters of the unit subset in the current power unit submodel in each time period are obtained. By combining the customized decomposition and coordination optimization strategy with machine learning technology, it is possible to ensure the convergence of the unit combination problem while accelerating the efficiency of unit combination optimization through the solution of sub-problems. It can solve the unit combination problem more quickly and reliably, and provide the power system with a reliable and efficient approximate optimal solution to the sub-hourly unit combination problem, so as to better meet the increasingly complex and changeable energy needs of the power system, improve the reliability and flexibility of the system, and ensure the safe and reliable supply of electricity.

[0126] See also Figure 3 In some embodiments, step S140 may include but is not limited to steps S310 to S320:

[0127] Step S310, feature encoding the unit operation feature data of the unit subset in all time periods to obtain forward unit operation features and backward unit operation features;

[0128] Step S320, feature decoding is performed on the forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period, and the preliminary operation parameters of the unit subset in each period in the current power unit submodel are obtained.

[0129] In step S310 of some embodiments, the machine learning model includes an encoder and a decoder, and a three-layer recurrent neural network (three-layer bidirectional GRU) with a gated attention mechanism is used as the encoder, and the encoder is used to analyze the unit operation characteristic data of the unit subset of the current power unit submodel in all time periods. Forward and backward processing is performed in time, and the output of the current layer is fed back to the next layer to convert the input sequence into a hidden state vector inside the network ,Will As the forward unit operation characteristics, As a backward unit operating characteristics. is the internal state variable of the forward return period t, is the internal state variable of period t during the backward return.

[0130] In step S320 of some embodiments, the encoded internal state variables are transmitted to the decoding stage to generate an output sequence. The decoding stage uses a decoder with an attention mechanism, and the decoder includes an attention mechanism, a gated recurrent unit, and a linear layer connected in sequence. The attention mechanism is implemented using a multi-layer perceptron with a single hidden layer. The attention mechanism enables the decoder to focus on related elements that show strong dependencies.

[0131] The unit operation characteristic data includes the unit operation status of each power unit in the unit subset during the initial period And unit output ,in , , The decoder is used to analyze the forward unit operation characteristics, the backward unit operation characteristics, and the unit operation status of the unit subset in the initial period. And unit output Perform feature decoding to obtain the unit operation status of the unit subset in the current power unit submodel at each time period And unit output , the unit operation status and unit output of time period t are taken as the preliminary operation parameters of time period t. It should be noted that if the unit operation status is 0, that is, the power unit is in an offline shutdown state, the unit output is 0.

[0132] In the above steps S310 to S320, a machine learning model is used to perform time series prediction on an input sequence of indefinite length, and the operating parameters of each sub-problem at different time periods are obtained, thereby improving the efficiency and accuracy of the operating parameter prediction.

[0133] See also Figure 4 In some embodiments, step S320 may include but is not limited to steps S410 to S450:

[0134] Step S410, performing attention transformation on the forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period to obtain the attention characteristics of the current period;

[0135] Step S420, performing gated feature decoding on the attention feature of the current period to obtain a first reference operating parameter of the unit subset in the current power unit submodel in the current period;

[0136] Step S430, performing a feasibility test on the first reference operating parameter of the current period, and determining a second reference operating parameter of the current period according to the test result of the feasibility test and the first reference operating parameter of the current period;

[0137] Step S440, performing attention transformation on the second reference operating parameter of the current period, the forward unit operating characteristics, the backward unit operating characteristics, the unit operating status and the unit output of the unit subset in the initial period, to obtain the attention characteristics of the next period;

[0138] Step S450, gated feature decoding is performed on the attention features of the next time period until the second reference operating parameters of the unit subset in the last time period are obtained, and the second reference operating parameters of each time period are used as preliminary operating parameters of each time period.

[0139] In step S410 of some embodiments, if the current period is the first period, the forward unit operation characteristics, the backward unit operation characteristics, and the unit operation status of the unit subset in the current power unit submodel in the initial period are analyzed by the attention mechanism. And unit output Perform attention transformation to obtain the attention features of the first period.

[0140] In step S420 of some embodiments, the attention feature of the first period is gated feature decoded by the gated recurrent unit to obtain the first reference operating parameter of the unit subset in the current power unit submodel in the first period. The first reference operating parameter is the unit operating state of the power unit And unit output .

[0141] In step S430 of some embodiments, a feasibility test is performed on the first reference operating parameter of the first period, and a second reference operating parameter of the first period is determined based on the test result of the feasibility test and the first reference operating parameter of the first period. The second reference operating parameter is the unit operating state and unit output of the power unit.

[0142] In step S440 of some embodiments, attention transformation is performed on the forward unit operation characteristics, the backward unit operation characteristics, the second reference operation parameters of the first time period, the unit operation status and the unit output of the unit subset in the current power unit submodel in the initial time period to obtain the attention characteristics of the next time period, i.e., the second time period.

[0143] In step S450 of some embodiments, the attention features of the next time period are gated and decoded by a gated loop unit, and a feasibility check is performed to obtain the second reference operating parameters of the next time period, and steps S440 to S450 are repeated until the second reference operating parameters of the unit subset in the last time period are obtained, and the second reference operating parameters of the unit subset in each time period are used as the preliminary operating parameters of the unit subset in each time period.

[0144] The above steps S410 to S450 solve the sub-problems through the machine learning model, which can establish the connection between the operating parameters of each time period, and improve the feasibility of the predicted value through feasibility verification, thereby ensuring the accuracy of the operating parameters of each time period.

[0145] See also Figure 5 In some embodiments, step S430 may also include but is not limited to steps S510 to S540:

[0146] Step S510, performing a preliminary feasibility test on the first reference operating parameter of the current period according to the sub-operation constraints;

[0147] Step S520, if the result of the preliminary feasibility test is that the preliminary feasibility test is passed, then obtaining the unit optimization constraint conditions of the current power unit submodel;

[0148] Step S530, re-checking the feasibility of the first reference operating parameter of the current period according to the unit optimization constraint condition;

[0149] Step S540: determining the second reference operating parameters of the current period according to the result of the feasibility test again and the first reference operating parameters of the current period.

[0150] In step S510 of some embodiments, in order to improve the feasibility of the predicted solution of the sub-problem, a preliminary feasibility test is performed on the first reference operating parameter of the current period through the sub-operation constraint. If the first reference operating parameter satisfies the sub-operation constraint, the test result of the preliminary feasibility test is that the preliminary feasibility test passes; if the first reference operating parameter does not satisfy the sub-operation constraint, the test result of the preliminary feasibility test is that the preliminary feasibility test fails.

[0151] In step S520 of some embodiments, the coupling between sub-problems is achieved through Lagrange multipliers, which can represent the power supply and demand of the power system, and the Lagrange multipliers converge only when the supply and demand relationship reaches a balance. For a sub-problem j in the kth iteration process, the goal is to obtain a sufficiently good feasible solution to meet the proxy optimal condition, that is, the unit optimization constraint condition. If the test result of the preliminary feasibility test is that the preliminary feasibility test passes, the unit optimization constraint condition of the current power unit submodel is obtained, and the unit optimization constraint condition is defined as follows:

[0152] ,

[0153] Where L is the sub-operation cost function of the power unit sub-model; the left-hand side of the inequality represents the set collected at the kth iteration The value of the variable L; the right side of the inequality represents the values ​​of the set collected at the kth iteration The value of the variable L.

[0154] If the test result of the preliminary feasibility test is that the preliminary feasibility test fails, it means that the predicted solution of the sub-problem is not a feasible solution. Then the first reference operating parameters of the current period are used as the initial solution, and the initial solution is searched for the optimal solution through the branch and cut algorithm to obtain the second reference operating parameters of the current period.

[0155] In step S530 of some embodiments, in order to obtain the optimal operating parameters, the first reference operating parameters of the current period are re-tested for feasibility according to the unit optimization constraints. If the first reference operating parameters meet the unit optimization constraints, the result of the re-test is that the re-test is passed. If the first reference operating parameters do not meet the unit optimization constraints, the result of the re-test is that the re-test is not passed.

[0156] In step S540 of some embodiments, if the result of the re-feasibility test is that the re-feasibility test is passed, the first reference operating parameter of the current period is used as the second reference operating parameter of the current period. If the result of the re-feasibility test is that the re-feasibility test is not passed, the first reference operating parameter of the current period is optimized by a branch cutting algorithm to obtain the second reference operating parameter of the current period.

[0157] In the above steps S510 to S540, the predicted solution is tested through feasibility testing, so that better operating parameters can be obtained, thereby ensuring the accuracy of the prediction of the operating parameters.

[0158] See also Figure 6 In some embodiments, step S540 may include but is not limited to step S610 or step S620:

[0159] Step S610, if the result of the re-feasibility test is that the re-feasibility test is passed, the first reference operating parameter of the current period is used as the second reference operating parameter of the current period;

[0160] Step S620: If the result of the re-feasibility test is that the re-feasibility test fails, the first reference operating parameters of the current period are optimized by a branch cutting algorithm to obtain the second reference operating parameters of the current period.

[0161] In step S610 of some embodiments, if the result of the re-feasibility test is that the re-feasibility test is passed, it means that the first reference operating parameter is the optimal solution to the current sub-problem, and the first reference operating parameter of the current period is used as the second reference operating parameter of the current period.

[0162] In step S620 of some embodiments, if the result of the re-feasibility test is that the re-feasibility test fails, it means that the first reference operating parameters are not the optimal solution to the current sub-problem. In this case, the first reference operating parameters of the current time period are used as the initial solution, and the solution space is searched for the initial solution through the branch cutting algorithm to obtain the second reference operating parameters of the current time period.

[0163] Through the above steps S610 to S620, better operating parameters with lower objective function values ​​can be obtained, which ensures the accuracy of operating parameter prediction and enables each power unit in the unit subset to operate stably.

[0164] In step S150 of some embodiments, when solving the next power unit sub-model, the unit operation characteristic data of the unit subset in the next power unit sub-model in each time period are determined according to the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period, and the unit operation characteristic data of the next power unit sub-model in all time periods are feature predicted through the machine learning model until the preliminary operation parameters of the unit subset in the last power unit sub-model in each time period are obtained.

[0165] In step S160 of some embodiments, the Lagrange multiplier and the first penalty coefficient are updated according to a preset update step size, and sufficiently good operating parameters are obtained by updating the multiplier and the penalty coefficient based on the alternative subgradient. In the next iteration round, the preliminary operating parameters of the unit subsets in all power unit submodels of the current iteration round in each time period are updated with reference to steps S140 to S150 until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels of the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period. The last iteration round refers to the iteration round that meets the termination condition, and the termination condition can be that the solution time is greater than the preset solution time, the value of the sub-operation cost function is lower than the preset threshold, or the number of iterations reaches the preset number of iterations. If the termination condition is not met, continue to the next iteration round.

[0166] It should be noted that the target operating parameters are the solutions of the dual space. In order to obtain the solutions of the original space, the greedy algorithm and the random method can be used to search the neighborhood of the target operating parameters. The target operating parameters of the power unit combination in each period can be expressed as , the constraint analysis of the target operation parameters is obtained .in, , distribute the unit output of power unit i to the power modules in the power unit, and the module output of each power module is equal to its maximum module output. That is:

[0167] .

[0168] See also Figure 7 The embodiment of the present application further provides an operating parameter prediction device for a power unit combination, which can implement the operating parameter prediction method for the power unit combination, and the device includes:

[0169] An acquisition module 710 is used to acquire an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints;

[0170] A relaxation module 720, used to relax the unit operation constraints and determine the target power unit model according to the relaxed unit operation constraints and the operation cost function;

[0171] A decomposition module 730, for dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset;

[0172] The first prediction module 740 is used to traverse each power unit sub-model in turn in the current iteration round, obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, perform feature prediction on the unit operation characteristic data of the unit subset in all time periods through the machine learning model, and obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period;

[0173] The second prediction module 750 is used to determine the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and perform characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in the last power unit submodel in each time period are obtained;

[0174] The updating module 760 is used to update the preliminary operating parameters of the unit subsets in all the power unit submodels in the current iteration round in each time period in the next iteration round until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all the power unit submodels in the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

[0175] The specific implementation of the operating parameter prediction device of the power unit combination is basically the same as the specific implementation of the operating parameter prediction method of the power unit combination mentioned above, and will not be repeated here.

[0176] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned method for predicting the operating parameters of the power unit combination when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0177] See also Figure 8 , Figure 8 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0178] The processor 810 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0179] The memory 820 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 820 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 820, and the processor 810 calls and executes the operating parameter prediction method of the power unit combination in the embodiment of this application;

[0180] Input / output interface 830, used to implement information input and output;

[0181] Communication interface 840, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.);

[0182] A bus 850 that transmits information between various components of the device (e.g., the processor 810, the memory 820, the input / output interface 830, and the communication interface 840);

[0183] The processor 810 , the memory 820 , the input / output interface 830 and the communication interface 840 are connected to each other in communication within the device via a bus 850 .

[0184] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the operating parameter prediction method of the power unit combination is implemented.

[0185] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0186] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0187] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0188] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0189] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0190] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0191] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0192] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0193] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0195] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0196] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A method for predicting operating parameters of a power unit combination, characterized in that: The method comprises: Acquire an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints; Relaxing the operating constraints of the unit, and determining a target power unit model according to the relaxed operating constraints of the unit and the operating cost function; Dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset; In the current iteration round, each power unit sub-model is traversed in turn to obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, and the unit operation characteristic data of the unit subset in all time periods are predicted by the machine learning model to obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period; Determine the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and perform characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in the last power unit submodel in each time period are obtained; In the next iteration round, the preliminary operating parameters of the unit subsets in all power unit submodels in the current iteration round in each time period are updated until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels in the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

2. The method according to claim 1, characterized in that The unit operation constraint conditions include load balance constraint conditions and line transmission safety constraint conditions, and the constraint condition relaxation is performed on the unit operation constraint conditions, and the target power unit model is determined according to the relaxed unit operation constraint conditions and the operation cost function, including: The load balancing constraint condition is relaxed respectively by using the Lagrange multiplier and the first penalty coefficient; Performing constraint relaxation on the line transmission safety constraint condition by using a first non-negative continuous variable and a second non-negative continuous variable; Update the operation cost function according to the relaxed load balancing constraint, the first non-negative continuous variable, the second non-negative continuous variable and the second penalty coefficient; The updated operation cost function and the relaxed line transmission safety constraint condition are used as the target power unit model.

3. The method according to any one of claims 1 to 2, characterized in that: The unit operation characteristic data includes the unit operation status and unit output of the unit subset in the initial period. The machine learning model is used to perform feature prediction on the unit operation characteristic data of the unit subset in all periods to obtain preliminary operation parameters of the unit subset in each period in the current power unit submodel, including: Characteristic encoding is performed on the unit operation characteristic data of the unit subset in all time periods to obtain forward unit operation characteristics and backward unit operation characteristics; The forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period are feature decoded to obtain preliminary operation parameters of the unit subset in each period in the current power unit submodel.

4. The method according to claim 3, characterized in that The forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period are decoded to obtain the preliminary operation parameters of the unit subset in each period in the current power unit submodel, including: Performing attention transformation on the forward unit operation characteristics, the backward unit operation characteristics, the unit operation status and the unit output of the unit subset in the initial period to obtain the attention characteristics of the current period; Perform gated feature decoding on the attention feature of the current period to obtain the first reference operating parameter of the unit subset in the current power unit sub-model in the current period; Performing a feasibility test on the first reference operating parameter of the current period, and determining a second reference operating parameter of the current period according to a test result of the feasibility test and the first reference operating parameter of the current period; Performing attention transformation on the second reference operating parameter of the current period, the forward unit operating characteristic, the backward unit operating characteristic, the unit operating state and the unit output of the unit subset in the initial period to obtain the attention characteristic of the next period; The attention features of the next period are gated feature decoded until the second reference operating parameters of the unit subset in the last period are obtained, and the second reference operating parameters of each period are used as the preliminary operating parameters of each period.

5. The method according to claim 4, characterized in that The power unit sub-model includes a sub-operation constraint, wherein the feasibility detection is performed on the first reference operation parameter of the current period, and the second reference operation parameter of the current period is determined according to the detection result of the feasibility detection and the first reference operation parameter of the current period, including: Performing a preliminary feasibility test on the first reference operation parameter of the current period according to the sub-operation constraints; If the test result of the preliminary feasibility test is that the preliminary feasibility test is passed, the unit optimization constraint conditions of the current power unit submodel are obtained; Re-checking the feasibility of the first reference operating parameter of the current period according to the unit optimization constraint condition; The second reference operating parameters of the current period are determined according to the detection result of the re-feasibility detection and the first reference operating parameters of the current period.

6. The method according to claim 5, characterized in that The determining of the second reference operating parameter of the current period according to the detection result of the re-feasibility detection and the first reference operating parameter of the current period includes: If the result of the re-feasibility test is that the re-feasibility test is passed, the first reference operating parameter of the current period is used as the second reference operating parameter of the current period; If the result of the re-feasibility test is that the re-feasibility test fails, the first reference operating parameters of the current period are optimized by a branch cutting algorithm to obtain the second reference operating parameters of the current period.

7. The method according to claim 5, characterized in that After performing a preliminary feasibility check on the first reference operating parameter of the current period according to the sub-operation constraints, the method further includes: If the test result of the preliminary feasibility test is that the preliminary feasibility test fails, the first reference operating parameters of the current period are optimized by a branch cutting algorithm to obtain the second reference operating parameters of the current period.

8. An operating parameter prediction device for a power unit combination, characterized in that: The device comprises: An acquisition module, used for acquiring an initial power unit model of the power unit combination; the initial power unit model includes an operation cost function of the power unit combination and unit operation constraints; A relaxation module, used for relaxing the operating constraints of the unit, and determining a target power unit model according to the relaxed operating constraints of the unit and the operating cost function; A decomposition module, used for dividing the power unit combination into a plurality of unit subsets, and decomposing the target power unit model into a plurality of power unit sub-models according to each unit subset; The first prediction module is used to traverse each power unit sub-model in turn in the current iteration round, obtain the unit operation characteristic data of the unit subset in the current power unit sub-model in each time period, perform feature prediction on the unit operation characteristic data of the unit subset in all time periods through the machine learning model, and obtain the preliminary operation parameters of the unit subset in the current power unit sub-model in each time period; The second prediction module is used to determine the unit operation characteristic data of the unit subset in the next power unit submodel in each time period according to the preliminary operation parameters of the unit subset in the current power unit submodel in each time period, and perform characteristic prediction according to the unit operation characteristic data of the next power unit submodel in all time periods until the preliminary operation parameters of the unit subset in the last power unit submodel in each time period are obtained; The updating module is used to update the preliminary operating parameters of the unit subsets in all power unit submodels of the current iteration round in each time period in the next iteration round until the last iteration round is reached, and the preliminary operating parameters of the unit subsets in all power unit submodels of the last iteration round in each time period are used as the target operating parameters of the power unit combination in each time period.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the method for predicting operating parameters of a power unit combination as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for predicting operating parameters of a power unit combination according to any one of claims 1 to 7 is implemented.

Images