Unit combination determination method and device of power grid system and computer equipment

Abstract

The invention relates to a unit commitment determination method and device of a power grid system and computer equipment. According to the multi-source data of the power grid system, the scheduling scene type of the multi-source data and each preset recommendation algorithm, determining an initial recommendation algorithm corresponding to the scheduling scene type, and according to the initial recommendation algorithm and the attribute information of each unit combination, determining an initial unit combination with the minimum target value of the to-be-optimized target from each unit combination, and determining a target unit combination according to the initial unit combination and a preset iteration stop condition. In the embodiment of the invention, the initial recommendation algorithm corresponding to the scheduling scene type is determined from the preset recommendation algorithms, so that the target unit combination is determined based on the initial recommendation algorithm corresponding to the scheduling scene type, and the determined target unit combination can flexibly cope with different scheduling scenes; and the determination accuracy of the target unit combination is improved.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a method, device and computer equipment for determining a unit combination of a power grid system. Background Art

[0002] With the access of renewable energy and the increase in the complexity of the power grid system, traditional optimization methods are limited in response speed and uncertainty handling, making it difficult to meet dynamic scheduling needs. At present, the unit combination optimization technology of renewable energy is gradually developing towards the intelligent direction of "traditional + data-driven" and "pure data-driven" methods, but there is still a key problem of how to balance the adaptability of the algorithm and the computing speed.

[0003] Currently, fixed algorithms are mostly used to determine unit combinations, but this method is difficult to flexibly respond to the needs of different scheduling scenarios. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device and computer equipment for determining the unit combination of a power grid system that can cope with the requirements of different scheduling scenarios in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for determining a unit combination of a power grid system, comprising:

[0006] Determining an initial recommendation algorithm corresponding to the scheduling scenario type according to the multi-source data of the power grid system, the scheduling scenario type of the multi-source data, and each preset recommendation algorithm;

[0007] According to the initial recommendation algorithm and the attribute information of each unit combination, determining an initial unit combination with the minimum target value of the target to be optimized from each unit combination;

[0008] A target unit combination is determined according to the initial unit combination and a preset iteration stop condition.

[0009] In one embodiment, determining the initial recommendation algorithm corresponding to the scheduling scenario type according to the multi-source data of the power grid system, the scheduling scenario type of the multi-source data and each preset recommendation algorithm includes:

[0010] For each of the preset recommendation algorithms, inputting the multi-source data into the preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the cross-section limit value of the power grid system;

[0011] Determine a first weight coefficient corresponding to the calculation time according to the scheduling scenario type of the multi-source data, and determine a second weight coefficient corresponding to the cross-limit value of the section;

[0012] The initial recommendation algorithm is determined according to the calculation time of each of the preset recommendation algorithms, the cross-section limit value of the power grid system, the first weight coefficient and the second weight coefficient.

[0013] In one embodiment, the method further comprises:

[0014] According to a preset scenario classifier and the multi-source data, a scheduling scenario type corresponding to the multi-source data is determined.

[0015] In one of the embodiments, determining the target unit combination according to the initial unit combination and a preset iteration stop condition includes:

[0016] When the initial unit combination does not satisfy the preset iteration stop condition, optimizing the first weight coefficient and the second weight coefficient according to the operation information of the initial unit combination and the scheduling scenario type to obtain a new first weight coefficient and a new second weight coefficient, and obtaining a target recommendation algorithm based on the new first weight coefficient and the new second weight coefficient;

[0017] According to the target recommendation algorithm and the attribute information of each unit combination, a new initial unit combination with the minimum target value of the target to be optimized is determined from each unit combination;

[0018] In the case that the new initial unit combination does not satisfy the preset iteration stop condition, return to execute the step of optimizing the first weight coefficient and the second weight coefficient according to the operation information of the initial unit combination and the scheduling scenario type to obtain a new initial unit combination, until the new initial unit combination satisfies the preset iteration stop condition, and use the new initial unit combination that satisfies the preset iteration stop condition as the target unit combination.

[0019] In one embodiment, the method further comprises:

[0020] Determining a first evaluation value of the recommended algorithm to be added according to the sample multi-source data of the power grid system and the recommended algorithm to be added;

[0021] Determine a second evaluation value of each of the preset recommendation algorithms according to the sample multi-source data and each of the preset recommendation algorithms;

[0022] Determining a target evaluation value according to each of the second evaluation values;

[0023] When the first evaluation value is less than the target evaluation value, the recommendation algorithm to be added is used as the preset recommendation algorithm.

[0024] In one embodiment, the method further comprises:

[0025] The preset recommendation algorithms corresponding to the second evaluation value that is not less than the preset evaluation value are eliminated.

[0026] In a second aspect, the present application further provides a device for determining a unit combination of a power grid system, comprising:

[0027] A first determination module, configured to determine an initial recommendation algorithm corresponding to the scheduling scenario type according to multi-source data of the power grid system, the scheduling scenario type of the multi-source data, and each preset recommendation algorithm;

[0028] A second determination module is used to determine, from the unit combinations, an initial unit combination with the minimum target value of the target to be optimized;

[0029] The third determination module is used to determine the target unit combination according to the initial unit combination and a preset iteration stop condition.

[0030] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the method steps provided in the first aspect when executing the computer program.

[0031] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method steps provided in the first aspect.

[0032] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the method steps provided in the first aspect when executed by a processor.

[0033] The above-mentioned method, device and computer equipment for determining the unit combination of the power grid system determine the initial recommendation algorithm corresponding to the scheduling scenario type based on the multi-source data of the power grid system, the scheduling scenario type of the multi-source data and each preset recommendation algorithm, determine the initial unit combination with the minimum target value of the target to be optimized from each unit combination based on the initial recommendation algorithm and the attribute information of each unit combination, and determine the target unit combination based on the initial unit combination and the preset iteration stop condition. In the embodiment of the present application, the initial recommendation algorithm corresponding to the scheduling scenario type is determined from each preset recommendation algorithm, and the target unit combination is determined based on the initial recommendation algorithm corresponding to the scheduling scenario type, so that the determined target unit combination can flexibly respond to different scheduling scenarios, thereby improving the accuracy of determining the target unit combination. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments of the present application or the related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 An application environment diagram of a method for determining a unit combination of a power grid system in one embodiment;

[0036] Figure 2 A schematic flow chart of a method for determining a unit combination of a power grid system in one embodiment;

[0037] Figure 3 A schematic diagram of a flow chart of a method for determining an initial recommendation algorithm in one embodiment;

[0038] Figure 4 A schematic flow chart of a method for determining a unit combination of a power grid system in another embodiment;

[0039] Figure 5 A schematic diagram of a flow chart of a method for determining a preset recommendation algorithm in one embodiment;

[0040] Figure 6 The present invention is a structural block diagram of a device for determining a unit combination of a power grid system in one embodiment. DETAILED DESCRIPTION

[0041] 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.

[0042] The method for determining the unit combination of the power grid system provided in the embodiment of the present application can be applied to the following examples: Figure 1 The application environment includes a computer device, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 1As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. 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, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data for determining the unit combination of the power grid system. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for determining the unit combination of the power grid system is implemented. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0043] Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0044] In an exemplary embodiment, Figure 2 As shown, a method for determining the unit combination of a power grid system is provided, and the method is applied to Figure 1 The computer device in the example is used to illustrate, including the following S201 to S203. Among them:

[0045] S201, determining an initial recommendation algorithm corresponding to a scheduling scenario type according to multi-source data of a power grid system, a scheduling scenario type of the multi-source data, and various preset recommendation algorithms.

[0046] Optionally, the multi-source data of the power grid system may include: load data, real-time status data of the power grid system, dispatching data and environmental constraint data, etc.; the real-time status data may include power generation, load demand, and section power limitation; the dispatching data may include dispatching cycle; and the environmental constraint data may include equipment operation restrictions, shutdown plans, etc.

[0047] Optionally, the scheduling scenario types may include peak load periods, valley load periods, and emergency scheduling periods, etc.

[0048] In the embodiment of the present application, for A={A1, A2, ..., An}, the multi-source data can be input into the preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the section over-limit value of the power grid system, and the first weight coefficient corresponding to the calculation time is determined according to the scheduling scenario type of the multi-source data, as well as the second weight coefficient corresponding to the section over-limit value. Based on the first weight coefficient and the second weight coefficient, the calculation time and the section over-limit value are weightedly summed to obtain the initial recommendation algorithm corresponding to the scheduling scenario type.

[0049] In a possible implementation, before inputting the multi-source data into a preset recommendation algorithm, the multi-source data may be standardized, and the multi-source data may be converted into the same format through standardization and preprocessing to ensure the consistency of the input data.

[0050]

[0051] Among them, min(x) is the minimum value of the same type of data in multi-source data, max(x) is the maximum value of the same type of data in multi-source data, x is each data of the same type in multi-source data, and x' is the standardized data of the same type in multi-source data.

[0052] Optionally, each preset recommendation algorithm can be divided into three categories: traditional algorithms, improved heuristic algorithms, and data-driven methods. Among them, traditional algorithms may include linear programming, dynamic programming algorithms, etc., which are suitable for simple scheduling scenarios. Improved heuristic algorithms may include genetic algorithms, particle swarm optimization algorithms, etc., which are suitable for medium-complexity scheduling scenarios. Data-driven methods may include deep reinforcement learning, which is suitable for complex scheduling scenarios with uncertainty and nonlinear constraints. For example, in load peak scenarios, algorithms with strong section over-limit control capabilities (such as particle swarm optimization) are preferred, while linear programming algorithms with faster calculation speeds are considered during non-peak periods.

[0053] Optionally, the initial recommendation algorithm may be any one of the above-mentioned preset recommendation algorithms, or a combination of multiple preset recommendation algorithms.

[0054] S202: Determine, from among the unit combinations, an initial unit combination with a minimum target value of the target to be optimized, according to the initial recommendation algorithm and the attribute information of each unit combination.

[0055] Optionally, the target value of the objective to be optimized may be the total power generation cost.

[0056] In the embodiment of the present application, it is assumed that the initial recommendation algorithm is a linear programming or dynamic programming method, which can be based on Minimize Determine the minimum target value among the target values ​​of each unit combination.

[0057] Among them, C(x) represents the total cost of each unit combination, that is, the target value, x i Indicates the status of each unit in the unit combination, c i is the unit power generation cost of each unit in the unit combination.

[0058] In a possible implementation, it is assumed that the initial recommendation algorithm is a hybrid algorithm. The unit combination 1 is generated based on the genetic algorithm and the attribute information of each unit combination, and the unit combination 1 is optimized by using a deep neural network or deep reinforcement learning to output the initial unit combination.

[0059] Optionally, deep reinforcement learning can adopt an Actor-Critic structure.

[0060] S203, determining a target unit combination according to the initial unit combination and a preset iteration stop condition.

[0061] In an embodiment of the present application, when the initial unit combination does not satisfy the preset iteration stop condition, the target recommendation algorithm is re-determined based on the operating information and scheduling scenario type of the initial unit combination, and a new initial unit combination with the smallest target value of the target to be optimized is determined from each unit combination based on the target recommendation algorithm and the attribute information of each unit combination. When the new initial unit combination does not satisfy the preset iteration stop condition, the step of determining a new initial unit combination based on the operating information and scheduling scenario type of the initial unit combination is returned until the new initial unit combination satisfies the preset iteration stop condition, and the new initial unit combination that satisfies the preset iteration stop condition is used as the target unit combination.

[0062] When the initial unit combination meets the preset iteration stop condition, the initial unit combination is used as the target unit combination.

[0063] Optionally, the preset iteration stop condition may be the number of iterations, the iteration time, or after multiple iterations, when the difference between the target value corresponding to the initial unit combination obtained currently and the target value corresponding to the initial unit combination obtained last time is less than a preset value.

[0064] In a possible implementation, when determining a new initial unit combination, each unit combination may be screened first to determine multiple candidate unit combinations with the highest correlation with the initial unit combination, and then, based on the target recommendation algorithm and the attribute information of each candidate unit combination, a new initial unit combination with the smallest target value of the target to be optimized may be determined from among the candidate unit combinations.

[0065] In the above-mentioned method for determining the unit combination of the power grid system, the initial recommendation algorithm corresponding to the scheduling scenario type is determined based on the multi-source data of the power grid system, the scheduling scenario type of the multi-source data and each preset recommendation algorithm, and the initial unit combination with the minimum target value of the target to be optimized is determined from each unit combination based on the initial recommendation algorithm and the attribute information of each unit combination, and the target unit combination is determined based on the initial unit combination and the preset iteration stop condition. In the embodiment of the present application, the initial recommendation algorithm corresponding to the scheduling scenario type is determined from each preset recommendation algorithm, and the target unit combination is determined based on the initial recommendation algorithm corresponding to the scheduling scenario type, so that the determined target unit combination can flexibly respond to different scheduling scenarios, thereby improving the accuracy of determining the target unit combination.

[0066] Figure 3 FIG. 1 is a flow chart of a method for determining an initial recommendation algorithm in an embodiment. Figure 3 As shown, the embodiment of the present application relates to a possible implementation method of determining an initial recommendation algorithm corresponding to a scheduling scenario type according to multi-source data of a power grid system, a scheduling scenario type of the multi-source data, and each preset recommendation algorithm, including the following steps:

[0067] S301 , for each preset recommendation algorithm, input multi-source data into the preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the cross-section over-limit value of the power grid system.

[0068] Optionally, the cross-section limit value may include the cross-section limit times and the cross-section limit degree.

[0069] In the embodiment of the present application, multi-source data are respectively input into each preset recommendation algorithm to obtain the calculation time of each preset recommendation algorithm and the cross-section limit value of the power grid system.

[0070] S302: Determine a first weight coefficient corresponding to the calculation time according to the scheduling scenario type of the multi-source data, and determine a second weight coefficient corresponding to the cross-limit value of the section.

[0071] In the embodiment of the present application, in each scheduling scenario, the ratio of the first weight coefficient to the second weight coefficient can be adjusted based on a weight adjustment model (such as an adaptive weight update algorithm based on gradient descent optimization). For example, in a peak load scheduling scenario, the impact of the cross-section limit may be greater, and the second weight coefficient is increased; while in normal time scheduling, the first weight coefficient is appropriately increased to speed up the scheduling.

[0072] For example, the first weight coefficient and the second weight coefficient can be dynamically determined by an adaptive gradient descent method (such as AdaGrad, Adam) to cope with changes in scheduling scenarios. Example: In an emergency scheduling scenario, in order to speed up the calculation, the Adam algorithm is used to continuously sum the weights and increase the first weight coefficient of the calculation time to ensure that the initial recommendation algorithm selected based on the first weight coefficient and the second weight coefficient algorithm is more biased towards a time-sensitive preset recommendation method.

[0073] S303, determining an initial recommendation algorithm according to the calculation time of each preset recommendation algorithm, the cross-limit value of the power grid system, the first weight coefficient and the second weight coefficient.

[0074] In the embodiment of the present application, for each preset recommendation algorithm, a weighted sum is performed based on the first weight coefficient and the second weight coefficient and the calculation time and the cross-section limit value to obtain the weighted value of each preset recommendation algorithm. The preset recommendation algorithm corresponding to the minimum weighted value is used as the initial recommendation algorithm. Specifically, it can be expressed as:

[0075] S(A i )=w t ·T(A i )+w l ·L(A i )

[0076] Among them, T(A i ) is the preset recommendation algorithm A i The calculation time of L(A i ) is the cross-section limit value of the power grid system, w t is the first weight coefficient, w l is the second weight coefficient, S(A i ) is the preset recommendation algorithm A i The weighted value of .

[0077] In the embodiment of the present application, for each preset recommendation algorithm, multi-source data is input into the preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the cross-section over-limit value of the power grid system, and the first weight coefficient corresponding to the calculation time is determined according to the scheduling scenario type of the multi-source data, and the second weight coefficient corresponding to the cross-section over-limit value is determined, and the initial recommendation algorithm is determined according to the calculation time of each preset recommendation algorithm, the cross-section over-limit value of the power grid system, the first weight coefficient and the second weight coefficient. When determining the initial recommendation algorithm, the embodiment of the present application fully considers the calculation time and the cross-section over-limit value, and can take into account both the calculation efficiency and the over-limit control, thereby improving the accuracy and efficiency of the target unit combination determined based on the initial recommendation algorithm.

[0078] In an exemplary embodiment, the scheduling scenario type corresponding to the multi-source data is determined according to a preset scenario classifier and the multi-source data.

[0079] Optionally, the preset scenario classifier may include a decision tree, a support vector machine or a lightweight deep learning model, a classifier constructed based on a tree-based ensemble method (such as XGBoost), etc. For example, the preset scenario classifier is a decision tree, and the sample source data (load data, environmental constraints, etc.) is input into the decision tree for real-time classification to obtain the scheduling scenario type.

[0080] When training the preset scenario classifier, the multi-source data of samples collected continuously during the actual operation, including load data, section power limit, generator status, etc., can be used to form a training set containing different types of dispatch scenarios. The preset scenario classifier integrated with the semi-supervised learning model is used to extract features and classify the collected training set data, so as to dynamically update the preset scenario classification model so that it can identify new types of dispatch scenarios. With the support of limited labeled data and a large amount of unlabeled data, the semi-supervised learning model can quickly learn and adapt to the changing needs of the power grid system.

[0081] After the preset scenario classifier is trained, it is applied to the scheduling execution process in real time. The recommended algorithm determined based on the predicted scheduling scenario type is regularly compared with the recommended algorithm under the historical scheduling scenario type, and the preset scenario classifier is updated according to the comparison results. For incorrectly classified scenario data, the system will re-label and retrain the model to improve the accuracy of the classifier and the scenario matching ability.

[0082] Semi-supervised learning models can include unlabeled data learning methods such as autoencoders and generative adversarial networks to enhance the scene classification capabilities of preset scene classifiers. Example: A large amount of unlabeled data collected during the scheduling process is input into the autoencoder, and after the semi-supervised learning model automatically extracts features, it is used to classify scheduling scenarios and identify special scenarios such as peak load and abnormal load.

[0083] In an exemplary embodiment, a target unit combination is determined based on an initial unit combination and a preset iteration stop condition, including: when the initial unit combination does not meet the preset iteration stop condition, the first weight coefficient and the second weight coefficient are optimized according to the operating information and the scheduling scenario type of the initial unit combination to obtain a new first weight coefficient and a new second weight coefficient, and a target recommendation algorithm is obtained based on the new first weight coefficient and the new second weight coefficient; according to the target recommendation algorithm and the attribute information of each unit combination, a new initial unit combination with the minimum target value of the target to be optimized is determined from each unit combination; when the new initial unit combination does not meet the preset iteration stop condition, return to execute the step of optimizing the first weight coefficient and the second weight coefficient according to the operating information and the scheduling scenario type of the initial unit combination to obtain a new initial unit combination, until the new initial unit combination meets the preset iteration stop condition, and the new initial unit combination that meets the preset iteration stop condition is used as the target unit combination.

[0084] In the embodiments of the present application, Figure 4 As shown, the data input module obtains multi-source data and performs standardization on the multi-source data. The intelligent algorithm selection module determines the initial recommendation algorithm corresponding to the scheduling scenario type based on the standardized multi-source data, the scheduling scenario type of the multi-source data and each preset recommendation algorithm. The scheduling optimization module determines the initial unit combination with the smallest target value of the target to be optimized from each unit combination according to the initial recommendation algorithm and the attribute information of each unit combination. The result feedback module determines whether the initial unit combination meets the preset iteration stop condition. If the initial unit combination does not meet the preset iteration stop condition, the operation information of the initial unit is fed back to the intelligent algorithm selection module.

[0085] The intelligent algorithm selection module optimizes the first weight coefficient and the second weight coefficient based on the operation information and the scheduling scenario type, and determines a new first weight coefficient and a new second weight coefficient. For each preset recommendation algorithm, the target recommendation algorithm is determined according to the new first weight coefficient, the new second weight coefficient, the calculation time and the cross-section limit value, and the scheduling optimization module obtains a new initial unit combination according to the target recommendation algorithm. The result feedback module determines whether the new initial unit combination meets the preset iteration stop condition. If the new initial unit combination meets the preset iteration stop condition, the new initial unit combination is used as the target unit combination. If the new initial unit combination does not meet the preset iteration stop condition, return to execute the above steps until the determined new initial unit combination meets the preset iteration stop condition, and the new initial unit combination that meets the preset iteration stop condition is output.

[0086] Optionally, Q-learning or a policy gradient method in reinforcement learning is used to optimize the first weight coefficient and the second weight coefficient. Example: Use Q-learning to explore the optimal weight setting in this scheduling scenario to maximize the combined score of computing time and cross-section limit violation.

[0087] In the embodiment of the present application, when the initial unit combination does not meet the preset iteration stop condition, the first weight coefficient and the second weight coefficient are optimized according to the operation information and the scheduling scenario type of the initial unit combination to obtain a new first weight coefficient and a new second weight coefficient, and based on the new first weight coefficient and the new second weight coefficient, a target recommendation algorithm is obtained; according to the target recommendation algorithm and the attribute information of each unit combination, a new initial unit combination with the minimum target value of the target to be optimized is determined from each unit combination; when the new initial unit combination does not meet the preset iteration stop condition, return to execute the step of optimizing the first weight coefficient and the second weight coefficient according to the operation information and the scheduling scenario type of the initial unit combination to obtain a new initial unit combination, until the new initial unit combination meets the preset iteration stop condition, and the new initial unit combination that meets the preset iteration stop condition is used as the target unit combination. The embodiment of the present application adaptively adjusts the first weight coefficient and the second weight coefficient dynamically based on the operation information and the scheduling scenario type of the initial unit combination to improve the accuracy of the determined target unit combination, facilitate adaptation to different scheduling scenario requirements, and improve the flexibility of scheduling under different scheduling scenarios.

[0088] Figure 5 FIG. 1 is a flow chart of a method for determining a preset recommendation algorithm in an embodiment. Figure 5 As shown, the following steps are included:

[0089] S501 : Determine a first evaluation value of a recommended algorithm to be added according to sample multi-source data of a power grid system and the recommended algorithm to be added.

[0090] Optionally, the recommendation algorithm to be added may be a new deep learning model, a hybrid optimization algorithm, etc.

[0091] In an embodiment of the present application, sample multi-source data is input into the recommended algorithm to be added, and the sample calculation time of the recommended algorithm to be added and the sample section over-limit value of the power grid system are obtained. The corresponding first sample weight coefficient and second sample weight coefficient are determined based on the scheduling scenario type of the sample multi-source data. The sample calculation time and the sample section over-limit value are weighted and summed based on the first sample weight coefficient and the second sample weight coefficient to obtain a first evaluation value.

[0092] In a possible implementation, the recommendation algorithm to be added may also be tested using automated machine learning tools or processes such as Hyperopt and TPOT. For example, TPOT is applied to the recommendation algorithm to be added to automatically test the performance of the recommendation algorithm to be added in the scheduling scenario to obtain a first evaluation value.

[0093] S502: Determine a second evaluation value of each preset recommendation algorithm according to the sample multi-source data and each preset recommendation algorithm.

[0094] In the embodiment of the present application, the second evaluation value may be determined based on the above S501, or an evaluation index may be added based on the above S501, such as the accuracy of each preset recommendation algorithm.

[0095] S503: Determine a target evaluation value according to each second evaluation value.

[0096] In the embodiment of the present application, the mean of the second evaluation values ​​may be used as the target evaluation value; the median of the second evaluation values ​​may be used as the target evaluation value; or the second evaluation values ​​may be weighted and summed to obtain the target evaluation value.

[0097] S504: When the first evaluation value is less than the target evaluation value, the recommendation algorithm to be added is used as a preset recommendation algorithm.

[0098] In the embodiment of the present application, the first evaluation value of the recommended algorithm to be added is determined based on the sample multi-source data of the power grid system and the recommended algorithm to be added, the second evaluation value of each preset recommended algorithm is determined based on the sample multi-source data and each preset recommended algorithm, and the target evaluation value is determined based on each second evaluation value. When the first evaluation value is less than the target evaluation value, the recommended algorithm to be added is used as the preset recommended algorithm. In the embodiment of the present application, the recommended algorithm to be added is tested and screened autonomously, which improves the efficiency and accuracy of each preset recommended algorithm, so that when determining the unit combination based on each preset recommended algorithm, a certain real-time performance and calculation efficiency can be guaranteed.

[0099] In an exemplary embodiment, the method further includes: eliminating a preset recommendation algorithm corresponding to a second evaluation value that is not less than a preset evaluation value.

[0100] In the embodiment of the present application, the performance of each preset recommendation algorithm is continuously evaluated through a rolling time window mechanism (for example, monthly or quarterly), and the calculation time, cross-section limit value, accuracy and other indicators of different preset recommendation algorithms in actual applications are recorded. Based on these indicators, the preset recommendation algorithms that no longer meet the performance requirements are removed to maintain the advancement of each preset recommendation algorithm.

[0101] Optionally, you can also use model pruning or quantization to optimize the deep learning model in the preset recommendation algorithm to reduce the computing time and storage requirements of the preset recommendation algorithm. Example: For the existing deep learning model in the preset recommendation algorithm, use the quantization tool to reduce the number of bits of the parameters in the deep learning model and optimize its execution speed to better adapt to real-time scheduling.

[0102] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0103] Based on the same inventive concept, the embodiment of the present application also provides a device for determining the unit combination of a power grid system for implementing the method for determining the unit combination of a power grid system involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the device for determining the unit combination of one or more power grid systems provided below can refer to the limitations of the method for determining the unit combination of a power grid system above, and will not be repeated here.

[0104] In an exemplary embodiment, Figure 6 As shown, a device for determining a unit combination of a power grid system is provided, comprising: a first determining module 11, a second determining module 12 and a third determining module 13, wherein:

[0105] A first determination module 11 is used to determine an initial recommendation algorithm corresponding to a scheduling scenario type according to multi-source data of the power grid system, the scheduling scenario type of the multi-source data, and each preset recommendation algorithm;

[0106] The second determination module 12 is used to determine the initial unit combination with the minimum target value of the target to be optimized from each unit combination according to the initial recommendation algorithm and the attribute information of each unit combination;

[0107] The third determination module 13 is used to determine the target unit combination according to the initial unit combination and the preset iteration stop condition.

[0108] In one embodiment, the third determination module 13 is specifically used to input multi-source data into the preset recommendation algorithm for each preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the section over-limit value of the power grid system; determine the first weight coefficient corresponding to the calculation time according to the scheduling scenario type of the multi-source data, and determine the second weight coefficient corresponding to the section over-limit value; determine the initial recommendation algorithm according to the calculation time of each preset recommendation algorithm, the section over-limit value of the power grid system, the first weight coefficient and the second weight coefficient.

[0109] In one embodiment, the third determination module 13 is specifically configured to determine the scheduling scenario type corresponding to the multi-source data according to a preset scenario classifier and the multi-source data.

[0110] In one embodiment, the third determination module 13 is specifically used to optimize the first weight coefficient and the second weight coefficient according to the operating information and the scheduling scenario type of the initial unit combination to obtain a new first weight coefficient and a new second weight coefficient when the initial unit combination does not meet the preset iteration stop condition, and obtain a target recommendation algorithm based on the new first weight coefficient and the new second weight coefficient; determine a new initial unit combination with the minimum target value of the target to be optimized from each unit combination according to the target recommendation algorithm and the attribute information of each unit combination; when the new initial unit combination does not meet the preset iteration stop condition, return to execute the step of optimizing the first weight coefficient and the second weight coefficient according to the operating information and the scheduling scenario type of the initial unit combination to obtain a new initial unit combination until the new initial unit combination meets the preset iteration stop condition, and use the new initial unit combination that meets the preset iteration stop condition as the target unit combination.

[0111] In one embodiment, the unit combination determination device of the power grid system further includes:

[0112] A fourth determination module, configured to determine a first evaluation value of the recommended algorithm to be added based on the sample multi-source data of the power grid system and the recommended algorithm to be added;

[0113] A fifth determination module, used to determine a second evaluation value of each preset recommendation algorithm according to the sample multi-source data and each preset recommendation algorithm;

[0114] A sixth determination module, used to determine a target evaluation value according to each second evaluation value;

[0115] The adding module is used to use the recommendation algorithm to be added as the preset recommendation algorithm when the first evaluation value is less than the target evaluation value.

[0116] In one embodiment, the unit combination determination device of the power grid system further includes:

[0117] The elimination module is used to eliminate the preset recommendation algorithms corresponding to the second evaluation value that is not less than the preset evaluation value.

[0118] Each module in the above-mentioned unit combination determination device of the power grid system can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can 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 each of the above modules.

[0119] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of any of the above method embodiments when executing the computer program.

[0120] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above method embodiments are implemented.

[0121] In one embodiment, a computer program product is provided, including a computer program, which implements the steps of any of the above method embodiments when executed by a processor.

[0122] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0123] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0124] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0125] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for determining unit combination of a power grid system, characterized in that: The method comprises: Determining an initial recommendation algorithm corresponding to the scheduling scenario type according to the multi-source data of the power grid system, the scheduling scenario type of the multi-source data, and each preset recommendation algorithm; According to the initial recommendation algorithm and the attribute information of each unit combination, determining an initial unit combination with the minimum target value of the target to be optimized from each unit combination; A target unit combination is determined according to the initial unit combination and a preset iteration stop condition.

2. The method according to claim 1, characterized in that The determining, according to the multi-source data of the power grid system, the dispatch scenario type of the multi-source data and each preset recommendation algorithm, an initial recommendation algorithm corresponding to the dispatch scenario type includes: For each of the preset recommendation algorithms, inputting the multi-source data into the preset recommendation algorithm to obtain the calculation time of the preset recommendation algorithm and the cross-section limit value of the power grid system; Determine a first weight coefficient corresponding to the calculation time according to the scheduling scenario type of the multi-source data, and determine a second weight coefficient corresponding to the cross-limit value of the section; The initial recommendation algorithm is determined according to the calculation time of each of the preset recommendation algorithms, the cross-section limit value of the power grid system, the first weight coefficient and the second weight coefficient.

3. The method according to claim 2, characterized in that The method further comprises: According to a preset scenario classifier and the multi-source data, a scheduling scenario type corresponding to the multi-source data is determined.

4. The method according to claim 2, characterized in that: The step of determining the target unit combination according to the initial unit combination and a preset iteration stop condition includes: When the initial unit combination does not satisfy the preset iteration stop condition, optimizing the first weight coefficient and the second weight coefficient according to the operation information of the initial unit combination and the scheduling scenario type to obtain a new first weight coefficient and a new second weight coefficient, and obtaining a target recommendation algorithm based on the new first weight coefficient and the new second weight coefficient; According to the target recommendation algorithm and the attribute information of each unit combination, a new initial unit combination with the minimum target value of the target to be optimized is determined from each unit combination; In the case that the new initial unit combination does not satisfy the preset iteration stop condition, return to execute the step of optimizing the first weight coefficient and the second weight coefficient according to the operation information of the initial unit combination and the scheduling scenario type to obtain a new initial unit combination, until the new initial unit combination satisfies the preset iteration stop condition, and use the new initial unit combination that satisfies the preset iteration stop condition as the target unit combination.

5. The method according to claim 1, characterized in that The method further comprises: Determining a first evaluation value of the recommended algorithm to be added according to the sample multi-source data of the power grid system and the recommended algorithm to be added; Determine a second evaluation value of each of the preset recommendation algorithms according to the sample multi-source data and each of the preset recommendation algorithms; Determining a target evaluation value according to each of the second evaluation values; When the first evaluation value is less than the target evaluation value, the recommendation algorithm to be added is used as the preset recommendation algorithm.

6. The method according to claim 5, characterized in that The method further comprises: The preset recommendation algorithms corresponding to the second evaluation value that is not less than the preset evaluation value are eliminated.

7. A device for determining unit combination of a power grid system, characterized in that: The device comprises: A first determination module, configured to determine an initial recommendation algorithm corresponding to the scheduling scenario type according to multi-source data of the power grid system, the scheduling scenario type of the multi-source data, and each preset recommendation algorithm; A second determination module is used to determine, from the unit combinations, an initial unit combination with the minimum target value of the target to be optimized; The third determination module is used to determine the target unit combination according to the initial unit combination and a preset iteration stop condition.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images