A grid dispatching method based on microgrid access

Abstract

The present invention discloses a grid dispatching method based on microgrid access, which belongs to the technical field of grid dispatching. By considering the amount of electricity transmitted by the microgrid, the energy utilization efficiency is improved and the energy consumption of grid power generation is reduced. Then, a trained electricity consumption prediction model is used to predict the power supply of the grid power supply area in the future period, and a grid power generation strategy is generated based on the predicted electricity consumption. This can effectively improve the power generation efficiency of the grid, avoid overproduction, and further reduce unnecessary energy consumption.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid dispatching, and in particular relates to a power grid dispatching method based on microgrid access. Background Art

[0002] Grid dispatch is the core link of power system operation. It ensures the safe, stable, economical and efficient operation of the power system by scientifically formulating power generation plans, optimizing power flow, real-time monitoring of system status and rapid response to emergencies, thereby meeting the power needs of all sectors of society and ensuring national energy security and social and economic development. In the existing technology, the power generation plan of the power grid is usually arranged by experienced personnel based on their own experience. In order to avoid insufficient power generation, more expected power generation is usually planned, which will lead to energy waste. Summary of the invention

[0003] The present invention provides a power grid dispatching method based on microgrid access, which is used to solve the problem of energy waste in the prior art.

[0004] A power grid dispatching method based on microgrid access, comprising:

[0005] Obtaining a target amount of electricity to be delivered by at least one microgrid to the power grid; wherein the target amount of electricity represents the amount of electricity that the microgrid will supply to the power grid in a future period, which is uploaded by the microgrid;

[0006] Collect historical power consumption influencing factors and historical power consumption corresponding to the power grid power supply area, and obtain sample data after pre-processing the historical power consumption influencing factors and historical power consumption;

[0007] A neural network model is used to construct a power consumption prediction model corresponding to the power supply area of ​​the power grid, and the power consumption prediction model is trained using sample data to obtain a trained power consumption prediction model;

[0008] The trained power consumption prediction model is used to predict the power supply of the power grid power supply area in the future period to obtain the predicted power consumption;

[0009] A power grid generation strategy is generated according to the target power consumption and the predicted power consumption, and the power grid is dispatched according to the power grid generation strategy.

[0010] In a possible implementation manner, collecting historical power consumption influencing factors and historical power consumption corresponding to the power grid supply area includes:

[0011] Collect seasonal data, weather data, temperature data, date type data and historical power consumption corresponding to the power supply area of ​​the power grid in historical time, and use seasonal data, weather data, temperature data and date type data as factors affecting historical power consumption;

[0012] The season data includes season classification codes, the weather data includes weather classification codes, the date type data includes working day codes or non-working day codes, and the historical time is in days.

[0013] In a possible implementation manner, after preprocessing the historical electricity consumption influencing factors and the historical electricity consumption, sample data is obtained, including:

[0014] Based on the historical electricity consumption influencing factors and historical electricity consumption, the historical electricity consumption for N consecutive days and the historical electricity consumption influencing factors on the N+1th day are taken as training samples, and the historical electricity consumption on the N+1th day is taken as training labels. The training samples and training labels are combined into sample data, and multiple different sample data are obtained.

[0015] In a possible implementation, using a neural network model to construct a power consumption prediction model corresponding to a power grid power supply area includes: using a BP neural network model to construct a power consumption prediction model corresponding to a power grid power supply area.

[0016] In a possible implementation, the power consumption prediction model is trained using sample data to obtain a trained power consumption prediction model, including:

[0017] Randomly initializing the hyperparameters of the electricity consumption prediction model to obtain a hyperparameter code, and repeatedly obtaining a plurality of different hyperparameter codes;

[0018] Obtain the error function value corresponding to each hyperparameter encoding, and take the hyperparameter encoding with the smallest error function value as the optimal encoding;

[0019] An individual collaborative search method is used to perform a first neighborhood search on the hyperparameter encoding to obtain a hyperparameter encoding after the first neighborhood search;

[0020] A population information fusion collaborative search method is used to perform a second neighborhood search on the hyperparameter encoding after the first neighborhood search to obtain the hyperparameter encoding after the second neighborhood search;

[0021] According to the optimal coding, a curve guidance method is used to search the optimal direction of the hyperparameter coding after the second neighborhood search to obtain the hyperparameter coding after the optimal direction search;

[0022] A cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search to obtain the hyperparameter encoding after the global search;

[0023] Determine whether the current number of training times has reached the maximum number of training times. If so, obtain the power consumption prediction model after training based on the hyperparameter encoding after the global search, otherwise return to the step of obtaining the optimal encoding.

[0024] In a possible implementation, an individual collaborative search method is used to perform a first neighborhood search on a hyperparameter encoding to obtain a hyperparameter encoding after the first neighborhood search, including:

[0025]

[0026]

[0027] in, Indicates t During the training i Hyperparameter encoding, Indicates the hyperparameter encoding after the first neighborhood search , represents the first random number between (0,1), represents the optimal encoding, represents the information interaction intensity adjustment parameter, Represents a random one with hyperparameter encoding Different other hyperparameter encodings, i =1,2,…,N, where N represents the total number of hyperparameter encodings, represents the first adjustment factor, and , represents the maximum number of optimizations, represents the location quality factor, and , Represents hyperparameter encoding The corresponding error function value is, Indicates t The second adjustment factor in the training process, and , Indicates t -1 times the second adjustment factor during the training process, and at the beginning of the training, the second adjustment factor is set to 0.01.

[0028] In a possible implementation, a population information fusion collaborative search method is used to perform a second neighborhood search on the hyperparameter encoding after the first neighborhood search to obtain the hyperparameter encoding after the second neighborhood search, including:

[0029]

[0030]

[0031] in, Indicates t During the training m The hyperparameter encoding after the first neighbor search, Indicated in t+1 hyperparameter encoding during training The update amount, Represents the hyperparameter encoding after the second neighborhood search , m =1,2,…,N, represents the inertia weight, Indicated in t Hyperparameter encoding during training The update amount, represents the second random number uniformly distributed in [0,1], Represents a random one with hyperparameter encoding Different other hyperparameter encodings, represents the interaction control factor, represents the third random number uniformly distributed in [0,1], Represents the central location of all hyperparameter encodings.

[0032] In a possible implementation, according to the optimal coding, a curve guidance method is used to perform an optimal direction search on the hyperparameter coding after the second neighborhood search to obtain the hyperparameter coding after the optimal direction search, including:

[0033]

[0034] in, Indicates t During the training n The hyperparameter encoding after the second neighborhood search, Indicates t +1 training session n The hyperparameter encoding after the second neighborhood search, represents the fourth random number between [0,2π], represents a random number between [0,π], represents the first control coefficient, and , π represents pi, represents the golden section number, and , represents the optimal encoding, represents the second control coefficient, and , Represents the central location of all hyperparameter encodings.

[0035] In a possible implementation, a cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search to obtain the hyperparameter encoding after the global search, including:

[0036] The cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search, and the search value is:

[0037]

[0038] in, Indicates t During the training k Hyperparameter encoding after optimal direction search, Represents hyperparameter encoding The corresponding search value, represents a random hyperparameter encoding, represents the sixth random number between (0,1), represents the maximum number of training times, represents pi;

[0039] Determine whether the error function value of the search value is reduced. If so, use the search value as the hyperparameter encoding after the global search. Otherwise, use the hyperparameter encoding after the original optimal direction search directly as the hyperparameter encoding after the global search.

[0040] In a possible implementation, generating a power grid generation strategy according to the target power consumption and the predicted power consumption, and scheduling the power grid according to the power grid generation strategy includes:

[0041] The predicted power consumption minus the target power consumption is used to obtain the basic power generation of the power grid;

[0042] Adding the basic power generation of the power grid to the preset fault-tolerant power generation to obtain the planned power generation of the power grid;

[0043] Taking the planned power generation of the power grid as a target, a power grid power generation strategy is generated, and the power grid is dispatched according to the power grid power generation strategy.

[0044] The present invention provides a grid dispatching method based on microgrid access, which improves energy utilization efficiency and reduces energy consumption of grid power generation by considering the amount of electricity transmitted by the microgrid. It then uses a trained electricity consumption prediction model to predict the power supply of the grid power supply area in future time periods, and generates a grid power generation strategy based on the predicted electricity consumption. This can effectively improve the power generation efficiency of the grid, avoid overproduction, and further reduce unnecessary energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0046] Figure 1A flowchart of a grid dispatching method based on microgrid access provided by an embodiment of the present invention.

[0047] Figure 2 A flowchart of using sample data to train the electricity consumption prediction model is provided in an embodiment of the present invention.

[0048] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0049] Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention.

[0050] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0051] like Figure 1 As shown, an embodiment of the present invention provides a grid dispatching method based on microgrid access, including:

[0052] S11, obtaining a target amount of electricity to be transmitted from at least one microgrid to the power grid; wherein the target amount of electricity represents the amount of electricity that the microgrid will supply to the power grid in the future period, which is uploaded by the microgrid;

[0053] A microgrid is a small, independent, controllable power system (such as a photovoltaic microgrid or a wind microgrid) that can operate autonomously and can also be interconnected with the main power grid. A microgrid is usually composed of distributed energy resources (DERs), loads, energy storage systems, and power electronic equipment. Generally, the electric energy stored in a microgrid can be transmitted to the power grid to achieve energy conservation. Therefore, the present invention incorporates the target amount of electricity that the microgrid transmits to the power grid into the production plan to achieve energy conservation.

[0054] It is worth noting that uploading by the microgrid means that the data is uploaded by the microgrid staff.

[0055] S12, collecting historical power consumption influencing factors and historical power consumption corresponding to the power grid power supply area, and pre-processing the historical power consumption influencing factors and historical power consumption to obtain sample data;

[0056] The historical electricity consumption influencing factors may be some factors that affect the electricity consumption in the power grid supply area, such as weather, working days, etc. By collecting these historical electricity consumption influencing factors, electricity consumption prediction can be achieved.

[0057] S13, using a neural network model to construct a power consumption prediction model corresponding to the power grid power supply area, and using sample data to train the power consumption prediction model to obtain a trained power consumption prediction model;

[0058] Using a neural network model to construct a power consumption prediction model corresponding to the power grid supply area may include: using a BP neural network, a convolutional neural network or other neural networks with prediction functions to construct a power consumption prediction model. After the power consumption prediction model is constructed, the hyperparameters of the power consumption prediction model need to be optimized before the prediction function can be realized. Therefore, it is necessary to train the power consumption prediction model using sample data to obtain the trained power consumption prediction model.

[0059] S14, using the trained power consumption prediction model to predict the power supply of the power grid power supply area in the future period to obtain the predicted power consumption;

[0060] Based on the format of sample data, the factors affecting current electricity consumption are collected and used as inputs of the trained electricity consumption prediction model to obtain the predicted electricity consumption.

[0061] S15. Generate a power grid generation strategy based on the target power consumption and the predicted power consumption, and dispatch the power grid according to the power grid generation strategy.

[0062] According to the target power consumption and the predicted power consumption, the amount of power that still needs to be produced can be determined, so as to determine how many generating sets need to be started to meet the demand.

[0063] The present invention provides a grid dispatching method based on microgrid access, which improves energy utilization efficiency and reduces energy consumption of grid power generation by considering the amount of electricity transmitted by the microgrid. It then uses a trained electricity consumption prediction model to predict the power supply of the grid power supply area in future time periods, and generates a grid power generation strategy based on the predicted electricity consumption. This can effectively improve the power generation efficiency of the grid, avoid overproduction, and further reduce unnecessary energy consumption.

[0064] In a possible implementation manner, collecting historical power consumption influencing factors and historical power consumption corresponding to the power grid supply area includes:

[0065] Collect seasonal data, weather data, temperature data, date type data and historical power consumption corresponding to the power supply area of ​​the power grid in historical time, and use seasonal data, weather data, temperature data and date type data as factors affecting historical power consumption;

[0066] The season data includes season classification codes, the weather data includes weather classification codes, the date type data includes working day codes or non-working day codes, and the historical time is in days.

[0067] It is worth noting that in addition to the above data, other data can also be used as factors affecting electricity consumption. When the factors affecting electricity consumption are non-digital data, these non-digital data should be converted into digital data and normalized to facilitate data learning and analysis of the electricity consumption prediction model.

[0068] In a possible implementation manner, after preprocessing the historical electricity consumption influencing factors and the historical electricity consumption, sample data is obtained, including:

[0069] Based on the historical electricity consumption influencing factors and historical electricity consumption, the historical electricity consumption for N consecutive days and the historical electricity consumption influencing factors on the N+1th day are taken as training samples, and the historical electricity consumption on the N+1th day is taken as training labels. The training samples and training labels are combined into sample data, and multiple different sample data are obtained.

[0070] In a possible implementation, a neural network model is used to construct a power consumption prediction model corresponding to a power grid power supply area, including: using a BP (Back Propagation) neural network model to construct a power consumption prediction model corresponding to a power grid power supply area.

[0071] In the prior art, intelligent optimization algorithms such as gradient descent or genetic algorithm are generally used to train neural network models, which often have technical problems such as being easily trapped in local optimality and slow training speed, ultimately resulting in poor data analysis capabilities. Therefore, an embodiment of the present invention provides an algorithm for training the power consumption prediction model to solve the technical problems existing in the prior art, improve the accuracy of power consumption prediction, and ultimately improve the accuracy of power grid power generation.

[0072] like Figure 2 As shown, the power consumption prediction model is trained using sample data to obtain a trained power consumption prediction model, including:

[0073] S21. Randomly initialize the hyperparameters of the electricity consumption prediction model to obtain a hyperparameter code, and repeatedly obtain a plurality of different hyperparameter codes;

[0074] For example, the weight parameters of the electricity consumption prediction model are randomly initialized between the upper limit and the lower limit of the weight parameters, and the weight parameters after initialization constitute a hyperparameter encoding.

[0075] Optionally, a chaotic mapping method may be used to initialize the hyperparameters of the electricity consumption prediction model to improve the training speed of the algorithm.

[0076] S22, obtaining the error function value corresponding to each hyperparameter encoding, and taking the hyperparameter encoding with the smallest error function value as the optimal encoding;

[0077] Optionally, a cross entropy error function may be used to obtain the error function value corresponding to each hyperparameter encoding, or a root mean square error function may be used to obtain the error function value of each hyperparameter encoding.

[0078] S23, performing a first neighborhood search on the hyperparameter code using an individual collaborative search method to obtain a hyperparameter code after the first neighborhood search;

[0079] S24, performing a second neighborhood search on the hyperparameter encoding after the first neighborhood search using a population information fusion collaborative search method to obtain the hyperparameter encoding after the second neighborhood search;

[0080] S25, according to the optimal coding, using a curve guidance method to perform an optimal direction search on the hyperparameter coding after the second neighborhood search, to obtain the hyperparameter coding after the optimal direction search;

[0081] S26, using a cosine function perturbation update method to perform a global search on the hyperparameter encoding after the optimal direction search to obtain the hyperparameter encoding after the global search;

[0082] S27. Determine whether the current number of training times has reached the maximum number of training times. If so, obtain the power consumption prediction model after training based on the hyperparameter encoding after the global search. Otherwise, return to the step of obtaining the optimal encoding.

[0083] In a possible implementation, an individual collaborative search method is used to perform a first neighborhood search on a hyperparameter encoding to obtain a hyperparameter encoding after the first neighborhood search, including:

[0084]

[0085]

[0086] in, Indicates t During the training i Hyperparameter encoding, Indicates the hyperparameter encoding after the first neighborhood search , represents the first random number between (0,1), represents the optimal encoding, represents the information interaction intensity adjustment parameter, Represents a random one with hyperparameter encoding Different other hyperparameter encodings, i =1,2,…,N, where N represents the total number of hyperparameter encodings, represents the first adjustment factor, and , represents the maximum number of optimizations, represents the location quality factor, and , Represents hyperparameter encoding The corresponding error function value is, Indicates t The second adjustment factor in the training process, and , Indicates t -1 times the second adjustment factor during the training process, and at the beginning of the training, the second adjustment factor is set to 0.01.

[0087] The individual collaborative search method provided by the embodiment of the present invention takes into account the influence of the position of the hyperparameter encoding itself, can integrate population information, realize local area search between different hyperparameter encodings, strengthen information exchange, realize exploration of more areas, and enhance training effects.

[0088] In a possible implementation, a population information fusion collaborative search method is used to perform a second neighborhood search on the hyperparameter encoding after the first neighborhood search to obtain the hyperparameter encoding after the second neighborhood search, including:

[0089]

[0090]

[0091] in, Indicates t During the training m The hyperparameter encoding after the first neighbor search, Indicated in t +1 hyperparameter encoding during training The update amount, Represents the hyperparameter encoding after the second neighborhood search , m =1,2,…,N, represents the inertia weight, Indicated in t Hyperparameter encoding during training The update amount, represents the second random number uniformly distributed in [0,1], Represents a random one with hyperparameter encoding Different other hyperparameter encodings, represents the interaction control factor, represents the third random number uniformly distributed in [0,1], Represents the central position of all hyperparameter encodings. The central position of all hyperparameter encodings refers to the hyperparameter encoding composed of the average value of each dimension.

[0092] The embodiment of the present invention provides a population information fusion collaborative search method, which can effectively utilize hyperparameter encoding information, individual historical information and group information interaction, so that a certain global optimization can be performed while effectively searching unknown local areas.

[0093] In a possible implementation, according to the optimal coding, a curve guidance method is used to perform an optimal direction search on the hyperparameter coding after the second neighborhood search to obtain the hyperparameter coding after the optimal direction search, including:

[0094]

[0095] in, Indicates t During the training n The hyperparameter encoding after the second neighborhood search, Indicates t +1 training session n The hyperparameter encoding after the second neighborhood search, represents the fourth random number between [0,2π], represents a random number between [0,π], represents the first control coefficient, and , π represents pi, represents the golden section number, and , represents the optimal encoding, represents the second control coefficient, and , Represents the central location of all hyperparameter encodings.

[0096] The curve guidance method provided in the embodiment of the present invention is based on the optimal coding and combines the central position of the hyperparameter coding, so that the entire population searches towards the optimal area. The hyperparameter coding is also disturbed to a certain extent, which expands the search range in the solution space while ensuring the search speed of the algorithm.

[0097] Optionally, since the curve guidance method has low search accuracy in the later stage of the algorithm, a greedy strategy can be used to control the curve guidance process after the number of training times reaches 2T / 3.

[0098] In a possible implementation, a cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search to obtain the hyperparameter encoding after the global search, including:

[0099] The cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search, and the search value is:

[0100]

[0101] in, Indicates t During the training k Hyperparameter encoding after optimal direction search, Represents hyperparameter encoding The corresponding search value, Represents a random hyperparameter encoding. The random hyperparameter encoding can be an existing hyperparameter encoding or a randomly generated hyperparameter encoding. represents the sixth random number between (0,1), represents the maximum number of training times, represents pi;

[0102] Determine whether the error function value of the search value is reduced. If so, use the search value as the hyperparameter encoding after the global search. Otherwise, use the hyperparameter encoding after the original optimal direction search directly as the hyperparameter encoding after the global search.

[0103] The cosine function perturbation update method provided in the embodiment of the present invention can not only significantly improve the global search of the algorithm, but also prevent the algorithm from falling into a stagnant state, thereby ensuring the search efficiency of the algorithm.

[0104] Optionally, an annealing simulation algorithm can be used to control the cosine function perturbation update method, thereby maintaining population diversity and being more conducive to searching for the optimal solution. After each search, each hyperparameter encoding can also be processed for out-of-bounds to ensure parameter validity.

[0105] In a possible implementation, generating a power grid generation strategy according to the target power consumption and the predicted power consumption, and scheduling the power grid according to the power grid generation strategy includes:

[0106] The predicted power consumption minus the target power consumption is used to obtain the basic power generation of the power grid;

[0107] Adding the basic power generation of the power grid to the preset fault-tolerant power generation to obtain the planned power generation of the power grid;

[0108] Taking the planned power generation of the power grid as a target, a power grid power generation strategy is generated, and the power grid is dispatched according to the power grid power generation strategy.

[0109] Optionally, the planned power generation of the power grid can be directly used as the power generation strategy of the power grid, so that the staff can generate power and realize the regulation of the power generation.

[0110] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0112] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0114] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program described can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0115] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A grid dispatching method based on microgrid access, characterized in that: include: Obtaining a target amount of electricity to be delivered by at least one microgrid to the power grid; wherein the target amount of electricity represents the amount of electricity that the microgrid will supply to the power grid in a future period, which is uploaded by the microgrid; Collect historical power consumption influencing factors and historical power consumption corresponding to the power grid power supply area, and obtain sample data after pre-processing the historical power consumption influencing factors and historical power consumption; A neural network model is used to construct an electricity consumption prediction model corresponding to the power supply area of ​​the power grid, and the electricity consumption prediction model is trained using sample data. The steps of obtaining the trained electricity consumption prediction model are as follows: randomly initialize the hyperparameters of the electricity consumption prediction model to obtain the hyperparameter coding, and repeatedly obtain multiple different hyperparameter codings; obtain the error function value corresponding to each hyperparameter coding, and take the hyperparameter coding with the smallest error function value as the optimal coding; use the individual collaborative search method to perform a first neighborhood search on the hyperparameter coding to obtain the hyperparameter coding after the first neighborhood search; use the population information fusion collaborative search method to search the first neighborhood The hyperparameter coding after the search is subjected to a second neighborhood search to obtain the hyperparameter coding after the second neighborhood search; according to the optimal coding, the curve guidance method is used to perform an optimal direction search on the hyperparameter coding after the second neighborhood search to obtain the hyperparameter coding after the optimal direction search; the cosine function perturbation update method is used to perform a global search on the hyperparameter coding after the optimal direction search to obtain the hyperparameter coding after the global search; it is determined whether the current number of training times reaches the maximum number of training times, and if so, the power consumption prediction model after training is obtained according to the hyperparameter coding after the global search, otherwise, the step of obtaining the optimal coding is returned; The trained power consumption prediction model is used to predict the power supply of the power grid power supply area in the future period to obtain the predicted power consumption; Based on the target power consumption and predicted power consumption, a power grid generation strategy is generated, and the power grid is dispatched according to the power grid generation strategy.

2. The grid dispatching method based on microgrid access according to claim 1, characterized in that: Collect historical power consumption influencing factors and historical power consumption corresponding to the power grid supply area, including: Collect seasonal data, weather data, temperature data, date type data and historical power consumption corresponding to the power supply area of ​​the power grid in historical time, and use seasonal data, weather data, temperature data and date type data as factors affecting historical power consumption; The season data includes season classification codes, the weather data includes weather classification codes, the date type data includes working day codes or non-working day codes, and the historical time is in days.

3. The grid dispatching method based on microgrid access according to claim 2 is characterized in that: After preprocessing the historical electricity consumption influencing factors and historical electricity consumption, sample data is obtained, including: Based on the historical electricity consumption influencing factors and historical electricity consumption, the historical electricity consumption for N consecutive days and the historical electricity consumption influencing factors on the N+1th day are taken as training samples, and the historical electricity consumption on the N+1th day is taken as training labels. The training samples and training labels are combined into sample data, and multiple different sample data are obtained.

4. The grid dispatching method based on microgrid access according to claim 1, characterized in that: A neural network model is used to construct a power consumption prediction model corresponding to the power supply area of ​​the power grid, including: using a BP neural network model to construct a power consumption prediction model corresponding to the power supply area of ​​the power grid.

5. The grid dispatching method based on microgrid access according to claim 1, characterized in that: The individual collaborative search method is used to perform a first neighborhood search on the hyperparameter encoding, and the hyperparameter encoding after the first neighborhood search is obtained, including: in, Indicates t During the training i Hyperparameter encoding, Indicates the hyperparameter encoding after the first neighborhood search , represents the first random number between (0,1), represents the optimal encoding, represents the information interaction intensity adjustment parameter, Represents a random one with hyperparameter encoding Different other hyperparameter encodings, i =1,2,…,N, where N represents the total number of hyperparameter encodings, represents the first adjustment factor, and , represents the maximum number of optimizations, represents the location quality factor, and , Represents hyperparameter encoding The corresponding error function value is, Indicates t The second adjustment factor in the training process, and , Indicates t -1 times the second adjustment factor during the training process, and at the beginning of the training, the second adjustment factor is set to 0.

01.

6. The grid dispatching method based on microgrid access according to claim 5, characterized in that: The population information fusion collaborative search method is used to perform a second neighborhood search on the hyperparameter encoding after the first neighborhood search, and the hyperparameter encoding after the second neighborhood search is obtained, including: in, Indicates t During the training m The hyperparameter encoding after the first neighbor search, Indicated in t +1 hyperparameter encoding during training The update amount, Represents the hyperparameter encoding after the second neighborhood search , m =1,2,…,N, represents the inertia weight, Indicated in t Hyperparameter encoding during training The update amount, represents the second random number uniformly distributed in [0,1], Represents a random one with hyperparameter encoding Different other hyperparameter encodings, represents the interaction control factor, represents the third random number uniformly distributed in [0,1], Represents the central location of all hyperparameter encodings.

7. The grid dispatching method based on microgrid access according to claim 6, characterized in that: According to the optimal coding, a curve guidance method is used to search the optimal direction of the hyperparameter coding after the second neighborhood search to obtain the hyperparameter coding after the optimal direction search, including: in, Indicates t During the training process n The hyperparameter encoding after the second neighborhood search, Indicates t +1 training session n The hyperparameter encoding after the second neighborhood search, represents the fourth random number between [0,2π], represents a random number between [0,π], represents the first control coefficient, and , π represents pi, represents the golden section number, and , represents the optimal encoding, represents the second control coefficient, and , Represents the central location of all hyperparameter encodings.

8. The grid dispatching method based on microgrid access according to claim 7 is characterized in that: The cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search, and the hyperparameter encoding after the global search is obtained, including: The cosine function perturbation update method is used to perform a global search on the hyperparameter encoding after the optimal direction search, and the search value is: in, Indicates t During the training process k Hyperparameter encoding after optimal direction search, Represents hyperparameter encoding The corresponding search value, represents a random hyperparameter encoding, represents the sixth random number between (0,1), represents the maximum number of training times, represents pi; Determine whether the error function value of the search value is reduced. If so, use the search value as the hyperparameter encoding after the global search. Otherwise, use the hyperparameter encoding after the original optimal direction search directly as the hyperparameter encoding after the global search.

9. The grid dispatching method based on microgrid access according to claim 1, characterized in that: Generating a power grid generation strategy according to the target power consumption and the predicted power consumption, and dispatching the power grid according to the power grid generation strategy, including: The predicted power consumption minus the target power consumption is used to obtain the basic power generation of the power grid; Adding the basic power generation of the power grid to the preset fault-tolerant power generation to obtain the planned power generation of the power grid; Taking the planned power generation of the power grid as a target, a power grid power generation strategy is generated, and the power grid is dispatched according to the power grid power generation strategy.

Images