Source-oriented internal microgrid balance optimization method and system for source-grid-load-storage virtual power plant

Abstract

The application provides a source network load storage virtual power plant-oriented internal microgrid balance optimization method and system, relates to the virtual power plant field, and the method comprises the following steps: obtaining first target power information from a power system, and performing cleaning, standardization, clustering and hierarchical classification processing to obtain second target power information; constructing a load model and designing a time sequence network according to the second target power information to construct a time sequence model; determining adjustable variables and adjustable capacity ranges to obtain boundary constraint conditions; building a double-layer optimization architecture facing user demand and user equipment end response scheduling, and constructing a multi-task learning model; solving the multi-task learning model, and outputting a target solution to adjust the internal microgrid balance of the virtual power plant. The target solution obtained by solving the multi-task learning model can represent the resource allocation instructions required by each device under the optimization target, and then the response strategy is determined to flexibly adjust the user load, so that the multi-objective optimization of source network load storage resources is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual power plants, in particular to an internal microgrid balance optimization method and system for a source-grid-load-storage virtual power plant. BACKGROUND

[0002] A virtual power plant is a power source coordination and management system that participates in the power market and grid operation as a special power plant through advanced information communication technology and software systems. The core of the virtual power plant concept can be summarized as "communication" and "aggregation". The key technologies of the virtual power plant mainly include coordination control technology, intelligent metering technology and information communication technology. The virtual power plant can provide management and auxiliary services for distribution networks and transmission networks. In actual operation, it is crucial to control the energy supply and demand balance of the internal microgrid of the virtual power plant. Microgrid balance is conducive to maintaining the stability of voltage and frequency, and can also ensure that the power demand of users can be met at any time.

[0003] At present, the virtual power plant technology is in the initial development stage and has some limitations. In related technologies, the cost of the energy storage system is high, and the energy storage efficiency and service life are key factors that restrict performance. It is difficult to dynamically feedback the optimal aggregation of multi-element load and demand response, accurately predict load changes and optimize scheduling, and the degree of technical integration between source, grid, load and storage is not high, making it difficult to achieve efficient collaboration. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an internal microgrid balance optimization method and system for a source-grid-load-storage virtual power plant, which solves the problem of insufficient internal microgrid balance and unsatisfactory optimization scheduling of the current virtual power plant technology.

[0005] To achieve the above purpose, the present application is realized by the following technical solutions:

[0006] In a first aspect, the embodiments of the present application provide an internal microgrid balance optimization method for a source-grid-load-storage virtual power plant, which comprises: obtaining, from a power system, first target power information representing historical load data and historical time series data according to required data types; performing cleaning, standardization, clustering and hierarchical classification processing on the first target power information to obtain second target power information; constructing a neural network learning framework in combination with a random process fitting Brownian motion according to the second target power information, and adjusting hyperparameters for optimization to obtain a load model; designing a time series network according to the second target power information, constructing a time series model to establish time correlation; determining adjustable variables and adjustable capacity ranges corresponding to optimization targets based on the load model and the time series model to obtain boundary constraint conditions; building a double-layer optimization architecture for user demand and user equipment end response scheduling, and constructing a multi-task learning model based on the boundary constraint conditions; solving the multi-task learning model to output a target solution to adjust the internal microgrid balance of the virtual power plant; wherein the target solution represents resource allocation instructions required by each device under the optimization target.

[0007] In a second aspect, the embodiments of the present application provide an internal microgrid balance optimization system for a source-grid-load-storage virtual power plant, which comprises: an obtaining module, a processing module, a first constructing module, a second constructing module, a constraint determining module, a third constructing module and a solving module.

[0008] Specifically, the obtaining module is configured to obtain, from a power system, first target power information representing historical load data and historical time series data according to required data types; the processing module is configured to perform cleaning, standardization, clustering and hierarchical classification processing on the first target power information to obtain second target power information; the first constructing module is configured to construct a neural network learning framework in combination with a random process fitting Brownian motion according to the second target power information, and adjust hyperparameters for optimization to obtain a load model; the second constructing module is configured to design a time series network according to the second target power information, and construct a time series model to establish time correlation; the constraint determining module is configured to determine adjustable variables and adjustable capacity ranges corresponding to optimization targets based on the load model and the time series model to obtain boundary constraint conditions; the third constructing module is configured to build a double-layer optimization architecture for user demand and user equipment end response scheduling, and construct a multi-task learning model based on the boundary constraint conditions; and the solving module is configured to solve the multi-task learning model to output a target solution to adjust the internal microgrid balance of the virtual power plant; wherein the target solution represents resource allocation instructions required by each device under the optimization target.

[0009] In a third aspect, an electronic device is provided, which comprises a processor, a memory, and a program stored in the memory and capable of running on the processor, and the program, when executed by the processor, implements the internal microgrid balancing optimization method for a source-grid-load-storage virtual power plant in the first aspect.

[0010] In a fourth aspect, a computer readable storage medium is provided, which stores a program or instructions, and the program or instructions, when executed by a processor, implements the internal microgrid balancing optimization method for a source-grid-load-storage virtual power plant in the first aspect.

[0011] The present application provides an internal microgrid balancing optimization method and system for a source-grid-load-storage virtual power plant. Compared with the prior art, the present application has the following beneficial effects:

[0012] Based on the first target power information representing historical load data and historical time series data, the present application constructs a load model and a time series model respectively, analyzes the load data and time series data of the power system, determines the adjustable variables and adjustable capacity range with respect to the set optimization target, and determines the adjustable potential. In order to effectively schedule and control the source-grid-load-storage resources, optimize the microgrid balancing facing the user demand and the corresponding user equipment end, the present application constructs a double-layer optimization architecture and establishes a multi-task learning model, determines the boundary constraint conditions of the multi-task learning model based on the determined adjustable variables and adjustable capacity range; the target solution obtained by solving the multi-task learning model can represent the resource allocation instructions required by each device under the optimization target, and then determine the response strategy to flexibly adjust the user load, interact on both sides of supply and demand, realize multi-objective optimization of source-grid-load-storage resources, and is beneficial to internal microgrid balancing. BRIEF DESCRIPTION OF DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0014] Figure 1 is a flowchart of an internal microgrid balancing optimization method for a source-grid-load-storage virtual power plant provided by the embodiments of the present application;

[0015] Figure 2 is Figure 1 is an exemplary flowchart of S120 in

[0016] Figure 3 is a structural diagram of an internal microgrid balancing optimization system for a source-grid-load-storage virtual power plant provided by the embodiments of the present application;

[0017] Figure 4 is another structure schematic diagram of a source network load storage oriented virtual power plant internal microgrid balance optimization system provided by the embodiment of the present application;

[0018] Figure 5 is a structure schematic diagram of an electronic device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application are described clearly and completely. Obviously, the described embodiments are some embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0020] It should be noted that, in this document, the relationship terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0021] The embodiments of the present application provide a source network load storage oriented virtual power plant internal microgrid balance optimization method and system, which solve the problems of insufficient internal microgrid balance and unsatisfactory optimization scheduling of current virtual power plant technology.

[0022] The technical solutions in the embodiments of the present application have the following general ideas to solve the above technical problems:

[0023] A virtual power plant is a power source coordination management system participating in power market and power grid operation as a special power plant through advanced information communication technology and software system. The core of the concept of the virtual power plant can be summarized as "communication" and "aggregation". The key technologies of the virtual power plant mainly include coordination control technology, intelligent metering technology and information communication technology. The virtual power plant can provide management and auxiliary services for distribution network and transmission network. In actual operation process, it is crucial to control the energy supply and demand balance of the internal microgrid of the virtual power plant. The microgrid balance is conducive to maintaining the stability of voltage and frequency, and can ensure that the power demand of users can be met at any time.

[0024] At present, the virtual power plant technology is in the initial development stage, and there are some limitations. In related technologies, the cost of the energy storage system is high, and the energy storage efficiency and service life are key factors restricting performance. The optimal aggregation of multi-element load and demand response cannot dynamically feedback, it is difficult to accurately predict load changes and optimize scheduling, the degree of technical integration between source, network, load and storage is not high, and efficient collaboration is difficult to achieve.

[0025] In order to better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the drawings of the specification and specific embodiments.

[0026] First, a kind of internal microgrid balance optimization method for source network load storage virtual power plant provided by the embodiment of the present application will be introduced.

[0027] The flowchart of the internal microgrid balance optimization method for source network load storage virtual power plant provided by the embodiment of the present application is shown as Figure 1 The internal microgrid balance optimization method for source network load storage virtual power plant can include the following steps S110-S170.

[0028] S110, according to the required data type, the first target power information representing the historical load data and the historical time sequence data is obtained from the power system.

[0029] S120, the first target power information is cleaned, standardized, clustered and classified to obtain the second target power information.

[0030] S130, according to the second target power information, a neural network learning framework is constructed by combining the random process of fitting Brown motion, and the hyperparameters are adjusted for optimization to obtain a load model.

[0031] S140, according to the second target power information, a time sequence network is designed, a time sequence model is constructed to establish time correlation.

[0032] S150, based on the load model and the time sequence model, the adjustable variables and adjustable capacity range corresponding to the optimization target are determined to obtain the boundary constraint condition.

[0033] S160, a double-layer optimization architecture for user demand and user equipment end response scheduling is built, and a multi-task learning model is constructed based on the boundary constraint condition.

[0034] S170, the multi-task learning model is solved, and the target solution is output to adjust the internal microgrid balance of the virtual power plant; wherein the target solution is used to represent the resource allocation instructions required by each device under the optimization target.

[0035] The foregoing is a specific implementation of the internal microgrid balance optimization method for the source-grid-load-storage virtual power plant provided by the embodiment of the present application. It can be understood that the first target power information based on the historical load data and the historical time series data is used to construct a load model and a time series model, respectively, to analyze the load data and the time series data of the power system, determine the adjustable variables and the adjustable capacity range related to the set optimization target, and determine the adjustable potential.

[0036] Further, to effectively schedule and control the source-grid-load-storage resources and optimize the microgrid balance facing the user demand and the corresponding user equipment end, the present application constructs a double-layer optimization architecture and establishes a multi-task learning model, determines the boundary constraint conditions of the multi-task learning model based on the determined adjustable variables and adjustable capacity range; the target solution obtained by solving the multi-task learning model can represent the resource allocation instructions required by each device under the optimization target, and then determine the response strategy to flexibly adjust the user load, interact on both sides of supply and demand, realize multi-objective optimization of source-grid-load-storage resources, and be beneficial to internal microgrid balance.

[0037] In one example, the load model includes a drift network and a diffusion network, the drift network is used to capture the uncertainty in the power grid output data and the load data; the diffusion network is used to capture the uncertainty in the preset load prediction model; wherein the load prediction model is used to generate predicted future data.

[0038] In some embodiments, as shown in Figure 2 The foregoing cleaning, standardization, clustering and hierarchical classification processing of the first target power information to obtain the second target power information, that is, the S120 specifically can include the following steps:

[0039] S210, cleaning the first target power information by Lagrange interpolation method, eliminating the abnormal values and noises of the data, and identifying and correcting the errors and missing values, to determine the integrity and accuracy of the data, to obtain the first intermediate information;

[0040] S220, calculating the scaling factor based on the mapping target range and the original range, to convert the data by a preset scaling formula to make the data have the same scale, to obtain the second intermediate information;

[0041] S230, using an unsupervised learning method combining contrastive learning and meta-learning, generating positive sample pairs of the same data through SimCLR enhancement, and using different data as negative sample pairs, clustering the second intermediate information to obtain clustering results;

[0042] S240, according to the clustering results, classifying the load-side adjustable resources to obtain the second target power information.

[0043] In the embodiments of the present application, it can be understood that the first target power information can include load side time series data, user power consumption and power consumption behavior and other related data and potential uncertainty interference factor data investigated from the power system. The present application cleans the load curve data by Lagrange interpolation method, eliminates the abnormal values and noises in the data, and identifies and corrects the errors and missing values in the load curve, so as to ensure the integrity and accuracy of the data.

[0044] In one example, in the data standardization process, the original range of the data can be defined by determining the minimum value and the maximum value in the data set corresponding to the first target power information; the scaling factor can be calculated based on the preset mapping target range and the original range; and the data conversion can be performed by calculating the preset scaling formula to make the data have the same scale.

[0045] In some embodiments, the aforementioned designing a time series network according to the second target power information and constructing a time series model to establish a time correlation, i.e., the aforementioned S140, can specifically include the following steps:

[0046] S310, performing time series decomposition, autocorrelation analysis and feature extraction on the time series data in the second target power information to obtain feature information representing the load data;

[0047] S320, processing the time dependence of the feature information by using a preset long short-term memory network or a gated recurrent unit to obtain an initial model;

[0048] S330, selecting a suitable loss function and adjusting network parameters to train the initial model to obtain a time series model.

[0049] In some embodiments, after the aforementioned designing a time series network according to the second target power information and constructing a time series model to establish a time correlation, i.e., after the aforementioned S140, the internal microgrid balance optimization method for the source-grid-load-storage virtual power plant can further include:

[0050] S141, defining nodes and edges, so that each node represents a load unit, and the node position represents the position of the corresponding load unit in the virtual power grid, and the edge represents the relationship between the nodes; wherein the weight of the edge is set based on the historical load correlation.

[0051] S142, using one of a graph convolution network (GCN) and a graph attention network (GAT) as a basic architecture, introducing an incremental learning algorithm, aggregating the information of neighbor nodes through multi-layer graph convolution operation, introducing residual connection and layer normalization to improve the training effect and convergence speed of the model.

[0052] S143、According to the change of the load data, dynamically adjust the weight of the edge, use the stream processing framework to process and update the graph data in real time, so that the model can reflect the latest load changes.

[0053] In some embodiments, the aforementioned double-layer optimization architecture is built for user demand and user equipment end response scheduling, and a multi-task learning model is constructed based on boundary constraint conditions, i.e., the aforementioned S160 can specifically include the following steps:

[0054] S410, based on the user demand and the user equipment end response scheduling sub-problem, a double-layer optimization architecture is built, and load types are designed for different optimization sub-problems;

[0055] S420, a double-layer coding method for continuous-discrete mixed variables is designed, and the continuous-discrete mixed variables include: load side user end load adjustment and energy storage adjustment continuous variables, and process scheduling discrete variables under each user;

[0056] S430, at least one of the load side adjustable capacity, user satisfaction, load side energy loss, minimized response cost, and energy storage cost is determined as an optimization target;

[0057] S440, based on the optimization target, the variables are limited by the boundary constraint condition and the model constraint is performed, and a multi-task learning model is constructed; the multi-task learning model includes an upper-layer user demand sub-model and a lower-layer load side response scheduling sub-model.

[0058] In some embodiments, the aforementioned multi-task learning model is constructed based on the optimization target, the variables are limited by the boundary constraint condition, and the model constraint is performed, i.e., the aforementioned S440 can specifically include the following steps:

[0059] S510, based on the double-layer optimization architecture, the upper-layer continuous decision variable is taken as the input of the lower-layer optimization problem, the discrete scheduling variable in the form of multi-layer coding is reconstructed, and is regarded as an independent scene for simultaneous optimization.

[0060] S520, in the decision space, according to the problem characteristics and the population distribution, it is judged whether there is similarity between tasks, and a similarity measurement index is set as a multi-task learning benchmark.

[0061] In some embodiments, after the aforementioned multi-task learning model is constructed based on the optimization target, the variables are limited by the boundary constraint condition, and the model constraint is performed, i.e., after the aforementioned S440, the internal microgrid balance optimization method for the source network load storage virtual power plant can further include:

[0062] S441, based on the weighting and shared network structure of the loss function between tasks, the shared transfer learning is used to jointly optimize all tasks for multi-task learning model training;

[0063] S442, defining the shareable knowledge type by the main task and the auxiliary task in the multi-task scene, deciding to migrate the local knowledge or the whole knowledge to establish the knowledge migration mechanism;

[0064] S443, taking the model input result after the knowledge migration as the environment update solution reference, selecting top-K individuals from the new input individuals and the historical individuals as the initial path scheduling selection scheme of the next generation, and evaluating and feeding back the result to judge the effective degree of the knowledge to realize the dynamic updating mechanism of the migrated knowledge.

[0065] In some embodiments, the internal microgrid balancing optimization method for the source-grid-load-storage virtual power plant provided by the present application can meet the needs of different use scenarios; the internal microgrid balancing optimization method for the source-grid-load-storage virtual power plant can further include dynamic environment transformation trend estimation, current and historical environment interaction relationship detection, and feedback mechanism design.

[0066] Specifically, in the dynamic environment transformation trend estimation process, the present application designs an environment detection mechanism with fixed time nodes to detect the dynamic changes of the target space; the current environment is at t+1 time, and t time is taken as the measurement reference to evaluate the interaction relationship between the uncertainty coupling parameters of the before and after time environment and to sense the environment change trend.

[0067] In the process of current and historical environment interaction relationship detection, the present application re-evaluates the difference of the area information contained by multiple uncertainty coupling parameters before and after the environment according to the uncertainty coupling parameter characteristics; different environment characteristics are extracted and detected, and if it is detected that the environment has no intersection and no significant distribution difference, the current environment is represented as a static optimization problem; if it is detected that the environment has an intersection and a significant difference, the historical environment characteristics are used as knowledge to migrate and guide the algorithm solving under the current environment.

[0068] In the feedback mechanism design process, the present application designs a cross-domain migration strategy with a feedback correction mechanism to evaluate the forward and reverse information of the optimization results before and after the knowledge migration and feed back to the source domain task that issues the knowledge; in addition, to ensure fairness and justice in decision optimization, the model is adjusted and optimized in the training process, the migrated knowledge is corrected and updated, and it is migrated to other tasks to adapt to different similarity tasks, effectively improve the problem solving quality, and shorten the task solving time.

[0069] In one example, for a cold storage type load: based on the internal microgrid balance optimization method for a source-grid-load-storage virtual power plant provided in the present application, the refrigeration equipment is operated by energy storage technology during the off-peak electricity price period to store cold energy to the target temperature interval; during the peak electricity price period, the stored cold energy is used to maintain the cold storage temperature, reducing the operation time of the refrigeration equipment; the optimal operation strategy of the refrigeration equipment is generated by a dynamic optimization model combined with the real-time electricity price signal, the load distribution is flexibly adjusted, and the peak load shifting and cost minimization are realized.

[0070] In one example, for a steam type load: based on the internal microgrid balance optimization method for a source-grid-load-storage virtual power plant provided in the present application, a multi-energy coordinated optimization strategy is determined to realize flexible adjustment and supply-demand balance of the steam load; during the off-peak electricity price period, excess steam is stored by an efficient heat storage device to reduce the high peak operation pressure; during the peak electricity price period, the stored steam is released to meet the production or living demand, reducing the operation time of the boiler and stabilizing the load peak.

[0071] In one example, for a ventilation type load: based on the internal microgrid balance optimization method for a source-grid-load-storage virtual power plant provided in the present application, an intelligent ventilation control strategy is determined to realize efficient energy utilization and operation cost optimization by adjusting the operation mode of the ventilation equipment in different periods; during the off-peak electricity price period, the operation power of the ventilation equipment is increased to store cold air or exhaust hot air; during the peak electricity price period, the operation power of the ventilation equipment is reduced to save electricity cost; natural ventilation is utilized according to the regional climate conditions (such as night low temperature or wind speed), and the wind speed and operation time of the ventilation system are dynamically adjusted according to the indoor temperature and humidity demand, load prediction results and electricity price changes.

[0072] In one example, for a lighting type load: based on the internal microgrid balance optimization method for a source-grid-load-storage virtual power plant provided in the present application, an intelligent lighting and natural light utilization strategy is determined to reduce the lighting electricity load and energy cost by dynamically adjusting the brightness and period of the lighting equipment; the lighting load is divided into a core area and an auxiliary area, and the lighting equipment in different areas is independently adjusted by an intelligent control system to preferentially guarantee the lighting demand of the core area; the indoor lighting brightness is dynamically adjusted by a sensor according to the outdoor light intensity to reduce the power consumption of the lighting equipment during the day or sunny weather; during the off-peak electricity price period, the overall lighting of the unnecessary area is performed to meet the centralized production or maintenance demand, and only the necessary lighting load is maintained during the peak electricity price period.

[0073] In some embodiments, the present application provides an internal microgrid balance optimization system 600 for a source-grid-load-storage virtual power plant, as shown in Figure 3 The internal microgrid balance optimization system 600 can include the following modules:

[0074] The acquisition module 610 is configured to acquire first target power information representing historical load data and historical time series data from a power system according to a required data type.

[0075] The processing module 620 is configured to perform cleaning, standardization, clustering, and hierarchical classification on the first target power information to obtain second target power information.

[0076] The first construction module 630 is configured to construct a neural network learning framework according to the second target power information, combine a random process of fitting Brownian motion, adjust hyperparameters for optimization, and obtain a load model.

[0077] The second construction module 640 is configured to design a time series network according to the second target power information, construct a time series model to establish a time correlation.

[0078] The constraint determination module 650 is configured to determine adjustable variables and adjustable capacity ranges corresponding to an optimization target based on the load model and the time series model, and obtain boundary constraint conditions.

[0079] The third construction module 660 is configured to build a double-layer optimization architecture facing user demand and user equipment end response scheduling, and construct a multi-task learning model based on the boundary constraint conditions.

[0080] The solving module 670 is configured to solve the multi-task learning model, and output a target solution to adjust internal micro-grid balance of a virtual power plant. The target solution is used to represent resource allocation instructions required by each device under the optimization target.

[0081] According to embodiments of the present application, any multiple modules of the acquisition module 610, the processing module 620, the first construction module 630, the second construction module 640, the constraint determination module 650, the third construction module 660, and the solving module 670 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of the modules can be combined with at least part of the functions of other modules, and implemented in one module.

[0082] In some embodiments, the processing module 620 can be specifically configured to:

[0083] The first target power information is cleaned by the Lagrange interpolation method to eliminate abnormal values and noise of the data, and to identify and correct errors and missing values, so as to determine the integrity and accuracy of the data, and obtain first intermediate information.

[0084] A scaling factor is calculated based on a mapping target range and an original range, so that data conversion is performed by a preset scaling formula to make the data have the same scale, and second intermediate information is obtained.

[0085] An unsupervised learning method combining contrastive learning and meta-learning is adopted, a positive sample pair of the same data is generated through SimCLR enhancement, and different data is used as a negative sample pair to cluster the second intermediate information to obtain a clustering result.

[0086] According to the clustering result, the load side adjustable resources are classified and graded to obtain the second target power information.

[0087] In some embodiments, the second construction module 640 can be specifically used for:

[0088] The time series data in the second target power information is subjected to time series decomposition, autocorrelation analysis and feature extraction to obtain feature information representing the load data.

[0089] The time dependence of the feature information is processed through a preset long short-term memory network or a gated recurrent unit to model an initial model.

[0090] A suitable loss function is selected and the network parameters are adjusted to train the initial model to obtain a time series model.

[0091] In some embodiments, as shown in Figure 4 The internal microgrid balance optimization system 600 can further include an incremental learning module 680, which can be specifically used for:

[0092] Defining nodes and edges, so that each node represents a load unit, and the node position represents the position of the corresponding load unit in the virtual power grid, and the edge represents the relationship between the nodes; wherein the weight of the edge is set based on the historical load correlation;

[0093] One of a graph convolution network GCN and a graph attention network GAT is used as a basic architecture, an incremental learning algorithm is introduced, information of neighbor nodes is aggregated through multi-layer graph convolution operation, residual connection and layer normalization are introduced to improve the training effect and convergence speed of the model;

[0094] According to the change of the load data, the weight of the edge is dynamically adjusted, a stream processing framework is used to process and update the graph data in real time, so that the model can reflect the latest load change.

[0095] In some embodiments, the third construction module 660 can specifically include the following units:

[0096] The architecture building unit 661 is used to build a double-layer optimization architecture based on user demand and user equipment end response scheduling sub-problems, and design load types for different optimization sub-problems;

[0097] The dual-layer coding unit 662 is used to design a dual-layer coding method for continuous-discrete mixed variables, which include: continuous variables of load regulation and energy storage regulation at the load-side user end, and discrete variables of scheduling of each user's next process.

[0098] The objective determination unit 663 is used to determine at least one of the following as the optimization objective: load-side adjustability, user satisfaction, load-side energy loss, minimization of response cost, and energy storage cost.

[0099] The learning model building unit 664 is used to construct a multi-task learning model based on the optimization objective, by limiting variables and constraining the model through boundary constraints. The multi-task learning model includes an upper-layer user demand sub-model and a lower-layer load-side response scheduling sub-model.

[0100] In some embodiments, the learning model building unit 664 can be specifically used for:

[0101] Based on a two-layer optimization architecture, the upper-layer continuous decision variables are used as the input to the lower-layer optimization problem, and the discrete scheduling variables are reconstructed into a multi-layer coding form, and are optimized as independent scenarios simultaneously.

[0102] In the decision space, based on the characteristics of the problem and the population distribution, we judge whether there is similarity between tasks and set a similarity metric as a benchmark for multi-task learning.

[0103] In some embodiments, such as Figure 4 As shown, the internal microgrid balancing optimization system 600 may further include a feedback optimization module 690, which can be specifically used for:

[0104] Based on a weighted and shared network structure with inter-task loss functions, shared transfer learning jointly optimizes all tasks to train a multi-task learning model.

[0105] Define the shareable knowledge types by defining the main task and auxiliary tasks in a multi-task scenario, and decide whether to transfer local or overall knowledge in order to establish a knowledge transfer mechanism.

[0106] Using the model input results after knowledge transfer as the benchmark for environmental update, the top-K individuals are selected from the new input individuals and historical individuals respectively as the initial path scheduling selection scheme for the next generation. The results are evaluated and feedback is provided to determine the effectiveness of the knowledge in order to realize the dynamic update mechanism of transferred knowledge.

[0107] Figure 4 Each module in the system shown has the function of implementing each step in the aforementioned internal microgrid balance optimization method for source-grid-load-storage virtual power plants, and can achieve the corresponding technical effects. For the sake of brevity, it will not be elaborated here.

[0108] In some embodiments, the present application provides an electronic device, a structural schematic diagram of which is shown in Figure 5 .

[0109] The electronic device can include a processor 710 and a memory 720 storing computer program instructions.

[0110] In particular, the processor 710 described above can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured to implement one or more integrated circuits that embody the embodiments of the present application.

[0111] The memory 720 can include a mass storage for data or instructions. By way of example and not limitation, the memory 720 can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory 720 can include removable or non-removable (or fixed) media. Where appropriate, the memory 720 can be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, the memory 720 is non-volatile solid-state memory.

[0112] The memory 720 can include read-only memory (ROM), random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, in general, the memory 720 includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software that, when executed (by one or more processors), is configured to perform the operations described above in any of the embodiments of the source network load storage virtual power plant-oriented internal microgrid balancing optimization method.

[0113] The processor 710 implements the source network load storage virtual power plant-oriented internal microgrid balancing optimization method in any of the embodiments described above by reading and executing computer program instructions stored in the memory 720.

[0114] In one example, the electronic device can further include a communication interface 730 and a bus 700. As shown in Figure 5 , the processor 710, the memory 720, and the communication interface 730 are connected through the bus 700 and complete communication among each other.

[0115] The communication interface 730 is mainly configured to implement the communication between the modules, devices, units and / or equipment in the embodiments of the present application.

[0116] Bus 700 includes hardware, software, or both, to couple and / or interface the components of the online data traffic billing device to each other and / or to other devices. By way of example, and not limitation, bus can be an Accelerated Graphics Port (AGP) or other graphics bus, a Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand (IB) interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association local (VLB) bus, or some other suitable bus or a combination of two or more of these. Where appropriate, bus 700 can include one or more buses. Although the present application is described and illustrated with a particular bus, it is not intended to be limited to this arrangement.

[0117] In addition, in combination with the above-mentioned source-oriented network-load-storage virtual power plant internal microgrid balance optimization method in the embodiments, the present application can provide a computer storage medium to implement. The computer storage medium has computer program instructions stored thereon; the computer program instructions are executed by a processor to implement any one of the above-mentioned source-oriented network-load-storage virtual power plant internal microgrid balance optimization methods in the embodiments.

[0118] It needs to be clear that the present application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above-mentioned embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after understanding the spirit of the present application.

[0119] The functional blocks shown in the structural block diagrams above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, functional cards, and the like. When implemented in software, the elements of the present application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. A "machine-readable medium" includes any medium that can store or transport information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, and the like. The code segments can be downloaded via computer networks such as the Internet, intranets, and the like.

[0120] It is also noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be performed simultaneously.

[0121] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer program instructions can also be stored in a computer readable medium that can direct a computer, other programmable data processing apparatus, or other processing devices to operate in a particular manner, such that the instructions stored in the computer readable medium produce an article of manufacture including instructions which implement the function / act specified in the flowchart and / or block diagram block or blocks.

[0122] In summary, compared with the prior art, the present application has the following beneficial effects:

[0123] 1. The application constructs a double-layer optimization architecture and establishes a multi-task learning model. Based on the determined adjustable variables and adjustable capacity range, the boundary constraint conditions of the multi-task learning model can be determined. The target solution obtained by solving the multi-task learning model can represent the resource allocation instructions required by each device under the optimization target, and then determine the response strategy to flexibly adjust the user load, interact on both sides of supply and demand, realize the multi-objective optimization of source, network, load and storage resources, and benefit the internal microgrid balance.

[0124] 2. The application predicts future load demand and power generation capacity by analyzing historical operation data and real-time monitoring data, and uses machine learning and deep learning algorithms to improve load prediction accuracy and optimize scheduling efficiency. The scheduling strategy can be adjusted according to real-time data to cope with load changes and renewable energy fluctuations; realize multi-objective optimization of source, network, load and storage resources, including maximizing economic benefits, minimizing carbon emissions and balancing supply and demand.

[0125] The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A source-oriented network-load-storage virtual power plant internal microgrid balance optimization method, characterized in that, The method comprises the following steps: According to the required data type, obtain the first target power information representing historical load data and historical time series data from the power system; Clean, standardize, cluster and classify the first target power information to obtain the second target power information; According to the second target power information, combine the random process of fitting Brown motion to construct a neural network learning framework, and adjust the hyperparameters for optimization to obtain a load model; According to the second target power information, design a time series network to construct a time series model to establish a time correlation; Based on the load model and the time series model, determine the adjustable variables and adjustable capacity range corresponding to the optimization target to obtain boundary constraint conditions; Build a double-layer optimization architecture facing user demand and user equipment end response scheduling, and build a multi-task learning model based on the boundary constraint conditions; Solve the multi-task learning model to output a target solution to adjust the internal microgrid balance of the virtual power plant; wherein the target solution represents the resource allocation instructions required by each device under the optimization target; the virtual power plant is a power supply coordination management system participating in the power market and the power grid operation; The method of building a double-layer optimization architecture facing user demand and user equipment end response scheduling, and building a multi-task learning model based on the boundary constraint conditions comprises: Based on the user demand and user equipment end response scheduling subproblem, build a double-layer optimization architecture, and design a load type for different optimization subproblems; Design a double-layer coding method for continuous-discrete mixed variables, including: load side user end load adjustment and energy storage adjustment continuous variables, and process scheduling discrete variables under each user; Determine at least one of the load side adjustable capacity, user satisfaction, load side energy loss, minimum response cost and energy storage cost as the optimization target; Based on the optimization target, limit the variables and perform model constraints through the boundary constraint conditions to construct a multi-task learning model; the multi-task learning model includes an upper-layer user demand submodel and a lower-layer load side response scheduling submodel.

2. The source-oriented load and energy storage virtual power plant internal microgrid balancing optimization method of claim 1, wherein, The method of cleaning, standardizing, clustering and classifying the first target power information to obtain the second target power information comprises: Clean the first target power information by Lagrange interpolation method to eliminate abnormal values and noises of the data, and identify and correct errors and missing values to determine the integrity and accuracy of the data, to obtain first intermediate information; Calculate the scaling factor based on the mapping target range and the original range to convert the data through a preset scaling formula to make the data have the same scale, to obtain second intermediate information; Use an unsupervised learning method combining contrastive learning and meta-learning to generate positive sample pairs of the same data through SimCLR enhancement, and use different data as negative sample pairs to cluster the second intermediate information to obtain clustering results; According to the clustering results, classify the load side adjustable resources to obtain the second target power information.

3. The source-oriented load and energy storage virtual power plant internal microgrid balancing optimization method of claim 1, wherein, The method of designing a time series network according to the second target power information, constructing a time series model to establish a time correlation comprises: time series decomposition, autocorrelation analysis and feature extraction are performed on time series data in the second target power information to obtain characteristic information representing the load data; An initial model is obtained by processing the time dependence of the characteristic information through a preset long short-term memory network or a gated recurrent unit. An appropriate loss function is selected and network parameters are adjusted to train the initial model to obtain a time series model.

4. The source-oriented load and energy storage virtual power plant internal microgrid balancing optimization method of claim 1, wherein, After designing a time series network according to the second target power information and constructing a time series model to establish a time correlation, the method further comprises: Nodes and edges are defined, each node represents a load unit, and the node position represents the position of the corresponding load unit in the virtual power grid, and the edge represents the relationship between the nodes; wherein the weight of the edge is set based on historical load correlation; One of a graph convolution network (GCN) and a graph attention network (GAT) is used as a basic architecture, an incremental learning algorithm is introduced, neighbor node information is aggregated through multi-layer graph convolution operation, and residual connection and layer normalization are introduced to improve the training effect and convergence speed of the model; According to the change of the load data, the weight of the edge is dynamically adjusted, and a stream processing framework is used to process and update the graph data in real time, so that the model can reflect the latest load change.

5. The source-oriented load and energy storage virtual power plant internal microgrid balancing optimization method of claim 1, wherein, The multi-task learning model is constructed based on the optimization target and by limiting variables and performing model constraints through the boundary constraint condition, comprising: Based on the double-layer optimization architecture, the upper-layer continuous decision variable is used as the input of the lower-layer optimization problem, the discrete scheduling variable is reconstructed in the form of multi-layer coding, and is regarded as an independent scene for simultaneous optimization. In the decision space, according to the problem characteristics and population distribution, it is judged whether there is similarity between tasks, and a similarity measurement index is set as a multi-task learning benchmark.

6. The source-oriented load and energy storage virtual power plant internal microgrid balancing optimization method of claim 1, wherein, After the multi-task learning model is constructed based on the optimization target and by limiting variables and performing model constraints through the boundary constraint condition, the method further comprises: Based on the weighting and shared network structure of the task loss function, the joint optimization of all tasks is performed through the shared transfer learning to train the multi-task learning model; The type of shareable knowledge is defined through the main task and the auxiliary task in the multi-task scene, the local knowledge or the overall knowledge is determined to establish the knowledge transfer mechanism; The model input result after knowledge transfer is used as an environment update solution benchmark, top-K individuals are selected from the new input individuals and the historical individuals as the initial path scheduling selection scheme of the next generation, and the result is evaluated and fed back to judge the effective degree of the knowledge to realize the dynamic updating mechanism of the transferred knowledge.

7. A source-oriented network-load-storage virtual power plant internal microgrid balance optimization system, characterized in that, comprises: An acquisition module is configured to acquire, from a power system, first target power information representing historical load data and historical time series data according to a required data type; A processing module is configured to perform cleaning, standardization, clustering and hierarchical classification processing on the first target power information to obtain second target power information; A first construction module is configured to construct a neural network learning framework according to the second target power information and in combination with a random process fitting Brownian motion, and adjust hyperparameters for optimization to obtain a load model. a second construction module, configured to design a timing network according to the second target power information, and construct a timing model to establish a time correlation; a constraint determination module, configured to determine adjustable variables and adjustable capacity ranges corresponding to an optimization target based on the load model and the timing model, and obtain a boundary constraint condition; a third construction module, configured to build a double-layer optimization architecture facing user demand and user equipment end response scheduling, and construct a multi-task learning model based on the boundary constraint condition; a solution module, configured to solve the multi-task learning model, and output a target solution to adjust internal micro-grid balance of the virtual power plant; the target solution is used to represent resource allocation instructions required by each device under the optimization target; the virtual power plant is a power source coordination management system participating in power market and power grid operation; the building of the double-layer optimization architecture facing user demand and user equipment end response scheduling, and the construction of the multi-task learning model based on the boundary constraint condition, comprises: building a double-layer optimization architecture based on user demand and user equipment end response scheduling sub-problems, and designing load types for different optimization sub-problems; designing a double-layer coding method facing continuous-discrete mixed variables, the continuous-discrete mixed variables including: load side user end load adjustment and energy storage adjustment continuous variables, and process scheduling discrete variables under each user; determining at least one of the following as the optimization target: load side adjustable capacity, user satisfaction, load side energy loss, minimized response cost, and energy storage cost; based on the optimization target, limiting variables and performing model constraint through the boundary constraint condition, and constructing a multi-task learning model; the multi-task learning model includes an upper-layer user demand sub-model and a lower-layer load side response scheduling sub-model.

8. An electronic device, comprising: comprise: a processor, a memory, and a program stored on the memory and executable on the processor, the program being executed by the processor to implement the internal micro-grid balance optimization method for the source network load storage virtual power plant according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, the computer readable storage medium stores a program or instructions, the program or instructions being executed by the processor to implement the internal micro-grid balance optimization method for the source network load storage virtual power plant according to any one of claims 1 to 6.

Images