Dynamic aggregation peak regulation method and system based on distributed photovoltaic power prediction and computing equipment

Abstract

The invention provides a dynamic aggregation peak regulation method and system based on distributed photovoltaic power prediction and computing equipment, and is applied to the technical field of dispatching and control of a power system. The dynamic aggregation peak regulation method comprises the following steps: acquiring data of a photovoltaic power station through a photovoltaic power station monitoring system, a meteorological station and a geographic information system; performing photovoltaic power prediction by using the distributed photovoltaic power prediction model; load demand prediction is carried out on the load in the future time period; performing network division and dynamic aggregation on the photovoltaic area; and carrying out peak regulation strategy optimization based on the output of the dynamic aggregation. According to the method, the peak regulation pressure of power grid operation can be reduced, and the stability and economical efficiency of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system scheduling and control, and particularly to a dynamic aggregation peak shaving method and system and a computing device based on distributed photovoltaic power prediction. Background Art

[0002] With the transformation of the global energy structure and the rapid development of clean energy technologies, photovoltaic power generation, as an important renewable energy source, has been widely used. However, the output of photovoltaic power generation is greatly affected by climate conditions and has volatility and intermittency. This instability poses challenges to the balanced scheduling of the power grid, especially when there are significant temporal differences between load demand and photovoltaic power generation, making peak shaving more difficult.

[0003] Traditional power system scheduling methods mainly rely on the flexibility of fossil fuel generating units. However, with the increasing proportion of renewable energy sources such as photovoltaic power, the limitations of traditional power system scheduling methods are becoming increasingly apparent. The response speed of fossil fuel generating units is relatively slow and it is difficult to cope with the rapid changes in photovoltaic power generation. In addition, maintaining a high reserve capacity to cope with uncertainties not only increases the system operation cost but also reduces the energy utilization efficiency. In this case, the phenomenon of curtailment of photovoltaic power gradually becomes serious, and part of the photovoltaic power generation is wasted, further affecting the effective utilization of renewable energy.

[0004] Therefore, in order to achieve more accurate load regulation and optimize the operation of the power grid, it is necessary to propose a dynamic peak shaving method based on photovoltaic power prediction to improve the flexibility and efficiency of system peak shaving. Summary of the Invention

[0005] The present invention aims to provide a dynamic aggregation peak shaving method and system and a computing device based on distributed photovoltaic power prediction, which can improve the photovoltaic power consumption capacity, reduce the peak shaving pressure of the power grid operation, and improve the stability and economy of the system.

[0006] According to one aspect of the present invention, a dynamic aggregation peak shaving method based on distributed photovoltaic power prediction is proposed, including:

[0007] Obtaining data of a photovoltaic power station through a photovoltaic power station monitoring system, a meteorological station, and a geographic information system;

[0008] Performing photovoltaic power prediction using a distributed photovoltaic power prediction model;

[0009] Performing load demand prediction on the load for a future time period;

[0010] Performing network partitioning and dynamic aggregation on the photovoltaic area;

[0011] Optimizing a peak shaving strategy based on the output of the dynamic aggregation.

[0012] According to some embodiments, photovoltaic power prediction using a distributed photovoltaic power prediction model includes:

[0013] Processing the time series features of distributed photovoltaic power using a long short-term memory network model;

[0014] Using a graph convolutional network model to process the spatial features of distributed photovoltaic power;

[0015] Fusing the time features and the spatial features together to generate a power prediction value for a photovoltaic power station.

[0016] According to some embodiments, the update formula of the graph convolutional network is:

[0017] H (l+1) =σ(D -1 / 2 AD -1 / 2 H (l) W (l) )

[0018] where σ is the activation function, D is the degree matrix of the graph, D -1 / 2 is a diagonal matrix, A is the adjacency matrix between photovoltaic power stations, representing the connection relationship between nodes, H (l) is the input feature of the l-th layer of the graph convolutional network, and W (l) is the trainable parameter matrix of the graph convolutional network, used to transform node features.

[0019] According to some embodiments, the formula for distributed photovoltaic power prediction is:

[0020]

[0021] where H GCN is the output feature of the graph convolutional network, h LSTM is the output feature of the long short-term memory network, FC represents the fully connected layer, ⊕ represents the feature concatenation operation, is the predicted power of the i-th power station at time t + k.

[0022] According to some embodiments, network partitioning and dynamic aggregation of a photovoltaic area includes:

[0023] Using the raster tool in the geographic information system to perform spatial partitioning according to the geographical features of the photovoltaic power stations, the distribution of the photovoltaic power stations, and the installed capacity information. Each raster network unit in the raster tool represents an independent management area.

[0024] According to some embodiments, network partitioning and dynamic aggregation of a photovoltaic area further includes:

[0025] Based on the photovoltaic power prediction result and the load demand prediction result, the aggregation center of the photovoltaic cluster is determined through a fusion clustering algorithm. The formula for dynamic aggregation is as follows:

[0026]

[0027] where, is the predicted photovoltaic power generation value of grid G j at time t + k, is the predicted load demand value of grid G j at time t + k. The difference between the two represents the imbalance degree of this grid area.

[0028] According to some embodiments, the peak shaving strategy is optimized based on the output of the dynamic aggregation, including:

[0029] Through a feedback control mechanism, the operating state of the power grid monitored in real time is adjusted. The feedback control formula is:

[0030] S(t) = S(t - 1) + α·(L(t) - P(t))

[0031] where, S(t) is the feedback adjustment coefficient, L(t) and P(t) are the real-time load and photovoltaic power generation respectively, and α is the adjustment coefficient.

[0032] According to some embodiments, the data obtained for the photovoltaic power station includes the historical power generation data, meteorological data, and geographical information data of the photovoltaic power station.

[0033] According to another aspect of the present invention, a system for dynamic aggregation peak shaving based on distributed photovoltaic power prediction is provided, including:

[0034] A data collection module for obtaining data related to photovoltaic power prediction;

[0035] A distributed photovoltaic power prediction module responsible for processing the time series characteristics and spatial characteristics of distributed photovoltaic power;

[0036] A load demand prediction module for predicting the load demand of the power grid in the future time period;

[0037] A grid division and dynamic aggregation module that uses a geographic information system and combines a fusion clustering algorithm to perform spatial division and define the power generation and load characteristics within each spatial area;

[0038] A peak shaving strategy optimization module that dynamically schedules resources in the time dimension.

[0039] According to another aspect of the present invention, a computing device is provided, including:

[0040] A processor; and

[0041] A memory stores a computer program which, when executed by the processor, causes the processor to execute any of the foregoing methods.

[0042] According to an exemplary embodiment of the present invention, data of a photovoltaic power station is acquired. Through data collection, the quality and reliability of model training can be improved, and multi-dimensional data is collected to provide rich and detailed basic materials for subsequent analysis. A distributed photovoltaic power prediction model is used for photovoltaic power prediction, which can quickly respond to changes in weather or other external factors, timely adjust the prediction results, and ensure the real-time performance and sensitivity of the system; load demand prediction is performed based on the attention mechanism. The attention mechanism enables the model to automatically learn important data within a time period, thereby more accurately capturing the load fluctuation law, effectively coping with load changes, and improving the prediction accuracy; a geographic information system is used for spatial division, and combined with a fusion clustering algorithm to determine the optimal photovoltaic cluster aggregation center, ensuring the scientificity and flexibility of the partition. Based on the output of the dynamic aggregation, the peak shaving strategy is optimized, comprehensively considering information in both time and space dimensions, formulating a more reasonable power dispatching plan, and maximizing the global supply-demand balance.

[0043] According to some embodiments, the distributed photovoltaic power prediction model combines a long short-term memory network and a graph convolutional network. The long short-term memory network is used to capture time dynamic characteristics, and the graph convolutional network is used to capture spatial correlations. The combination of the two significantly improves the accuracy of photovoltaic power prediction and provides reliable data support for the dynamic dispatching of the power grid.

[0044] It should be understood that the above general description and the following detailed description are only exemplary and do not limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments.

[0046] Figure 1 The flowchart of the method for dynamic aggregation peak shaving based on distributed photovoltaic power prediction according to an exemplary embodiment is shown.

[0047] Figure 2 The schematic block diagram of the dynamic aggregation peak shaving system based on distributed photovoltaic power prediction according to an exemplary embodiment is shown.

[0048] Figure 3 The block diagram of a computing device according to an exemplary embodiment of the present invention is shown. DETAILED DESCRIPTION

[0049] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. Like reference numerals refer to like or similar parts throughout the figures, and thus their repetitive description will be omitted.

[0050] In addition, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present invention. However, those skilled in the art will realize that the technical solutions of the present invention may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be used. In other cases, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of the present invention.

[0051] The block diagrams shown in the drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0052] The flowcharts shown in the drawings are merely illustrative and do not necessarily include all of the content and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps may be decomposed, while some operations / steps may be combined or partially combined, so the actual execution order may change according to the actual situation.

[0053] It should be understood that although terms such as first, second, and third may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another. Thus, the first component discussed below may be referred to as the second component without departing from the teachings of the concept of the present invention. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0054] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of the relevant data need to comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entrances are provided for the user to choose to authorize or reject.

[0055] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of exemplary embodiments, and the modules or processes in the accompanying drawings are not necessarily essential for implementing the present invention. Therefore, they cannot be used to limit the protection scope of the present invention.

[0056] Traditional power system scheduling mainly relies on fossil fuels such as coal and natural gas. However, with the increasing proportion of renewable energy sources such as photovoltaic power, the limitations of traditional scheduling methods are gradually emerging. Due to their physical characteristics, fossil fuel generating units have a slow start-up and shutdown process and are difficult to quickly respond to the rapid changes of renewable energy sources such as photovoltaic power generation. Moreover, traditional generating units are usually designed for stable operation, and frequent start-up and shutdown not only increase the maintenance cost but also may shorten the equipment life.

[0057] Photovoltaic and other renewable energy sources have significant volatility and intermittency characteristics, which pose new challenges to the balanced scheduling of power systems. Existing scheduling mechanisms may not be able to effectively integrate a large number of dispersed distributed power sources, resulting in local overload or waste. Even if there is sufficient photovoltaic power generation capacity, part of the generated electricity may have to be abandoned due to insufficient grid acceptance capacity or improper scheduling, that is, the phenomenon of "photovoltaic curtailment" occurs. With the increasing proportion of photovoltaic energy in the power system, how to effectively manage its volatility and intermittency has become a key challenge to ensure the stable operation of the power grid.

[0058] Therefore, the present invention proposes a dynamic aggregation peak shaving method based on distributed photovoltaic power prediction to solve the problems such as large peak shaving pressure on the power grid, difficult optimization of power resources allocation, and insufficient real-time response ability caused by the volatility of photovoltaic power generation.

[0059] Before describing the embodiments of the present application, some terms or concepts related to the embodiments of the present application are explained.

[0060] GIS (Geographic Information System), a geographic information system is a tool and technology for capturing, storing, operating, analyzing, managing, and presenting all forms of geographic data. By combining spatial data and attribute data, it solves complex spatial problems.

[0061] The exemplary embodiments of the present invention are described below with reference to the accompanying drawings.

[0062] The present invention conducts intelligent inspection on the anti-external force damage of transmission lines based on a visual model. The visual model uses the Florence-2 model and combines the specific requirements of transmission line inspection to develop an intelligent inspection method based on a visual architecture to improve the recognition accuracy and inspection efficiency.

[0063] Figure 1 A method flow chart of dynamic aggregation peak shaving based on distributed photovoltaic power prediction according to an exemplary embodiment is shown.

[0064] In S101, data of the photovoltaic power station is obtained through a photovoltaic power station monitoring system, a weather station, and a geographic information system.

[0065] According to the exemplary embodiment, key data related to photovoltaic power prediction is obtained, including historical power generation data of the photovoltaic power station, meteorological data such as solar radiation, temperature, humidity, wind speed, etc., and geographic information data such as the longitude, latitude, and altitude of the power station. The data of the photovoltaic power station is collected through channels such as a photovoltaic power station monitoring system, a weather station, and a geographic information system (GIS). The obtained data of the photovoltaic power station is preprocessed, through cleaning and standardization processing, to ensure the integrity, consistency, and accuracy of the data, providing high-quality input for the training and application of subsequent distributed photovoltaic power prediction models.

[0066] In S103, a distributed photovoltaic power prediction model is used for photovoltaic power prediction.

[0067] According to the exemplary embodiment, a distributed photovoltaic power prediction model is used for photovoltaic power prediction. The distributed photovoltaic power prediction model includes a long short-term memory network model (LSTM) and a graph convolutional network model (GCN). The long short-term memory network model is responsible for processing the time series characteristics of distributed photovoltaic power, and the graph convolutional network model is responsible for processing the spatial characteristics of distributed photovoltaic power. By fusing the time characteristics and spatial characteristics together, a power prediction value of the photovoltaic power station is generated.

[0068] LSTM is a special recurrent neural network (RNN), specifically designed to overcome the problem of gradient vanishing or explosion encountered by traditional RNNs when processing long time series data. The long short-term memory network includes a forget gate, an input gate, a memory cell, and an output gate. Information is stored in the memory cell through the input gate, the forget gate determines the information to be discarded from the memory cell, and the result of the output gate is used as the time series prediction result of photovoltaic power. The update process of the LSTM network can be expressed as:

[0069] Forget gate: f t = σ(W f · [h t-1 , x t + b f )

[0070] Where:

[0071] σ is the sigmoid activation function;

[0072] W f is the weight matrix of the forget gate;

[0073] h t-1 is the hidden state at the previous moment;

[0074] x t is the input at the current moment;

[0075] bf is the bias term of the forget gate.

[0076] Input gate: i t = σ(W i · [h t-1 , x t + b i )

[0077]

[0078] Wherein:

[0079] W i is the weight matrix of the input gate, b i is the bias term vector of the input gate;

[0080] W c is the weight matrix of the intermediate variable, b c is the bias vector of the intermediate variable.

[0081] Memory cell update:

[0082] Output gate: o t = σ(W o · [h t-1 , x t + b o )

[0083] h t = o t · tanh(C t )

[0084] Wherein:

[0085] W o is the weight matrix of the output gate

[0086] b o is the bias vector of the output gate

[0087] Finally, the output h of the LSTM t will be used as the time series prediction result of the photovoltaic power.

[0088] The LSTM in this example stores information through memory cells and uses input gates, forget gates, and output gates to control the inflow, retention, and outflow of information, thereby effectively managing long-term dependencies.

[0089] According to some embodiments, the update formula of the graph convolutional network is:

[0090] H (l+1) = σ(D -1 / 2 AD -1 / 2 H (l) W (l) )

[0091] Wherein:

[0092] σ is the activation function;

[0093] D is the degree matrix of the graph, and D -1 / 2 is a diagonal matrix;

[0094] A is the adjacency matrix between photovoltaic power stations, representing the connection relationship between nodes;

[0095] H (l) is the input feature of the l-th layer of the graph convolutional network;

[0096] W (l) is the trainable parameter matrix of the graph convolutional network, used to transform node features.

[0097] By the distributed photovoltaic power prediction model, the time feature of LSTM and the spatial feature of GCN are fused together to generate the power prediction value of the photovoltaic power station. Let H GCN be the output feature of GCN, and h LSTM be the output feature of LSTM. Then the distributed photovoltaic power prediction formula is:

[0098]

[0099] Wherein, is the predicted power of the i-th power station at time t + k. FC represents the fully connected layer, represents the feature concatenation operation.

[0100] This embodiment can effectively fuse the time feature and the spatial feature to generate the power prediction value of the photovoltaic power station. By processing the historical power data and meteorological data through the LSTM network, the time series prediction result h LSTM is generated. By processing the connection relationship between power stations through the GCN network, the spatial feature H GCN is generated. The time feature and the spatial feature are concatenated h LSTM ⊕ H GCN and then the fused feature is processed through the fully connected layer to generate the final power prediction value.

[0101] In S105, the load demand of the future time period is predicted.

[0102] According to the exemplary embodiment, load demand forecasting is performed to predict the load demand of the power grid in a future time period, so as to ensure that the power grid can schedule power resources in a timely manner according to load changes. Based on the attention mechanism, load demand forecasting is carried out using a deep learning model, which can process complex and variable load data and generate high-precision load forecasting results.

[0103] Attention weight α t The calculation formula is as follows:

[0104]

[0105] In the formula, "exp" represents the natural exponential function, "tanh" is the hyperbolic tangent activation function, W1 and W2 are weight matrices, and h t-1 represents the hidden state at the previous moment, and X(t) is the input at the current moment.

[0106] Attention weight α t represents the attention degree of the model to the input features at time t, X(t) is the input feature vector at time t, and h t is the hidden layer state vector.

[0107] Load demand prediction value The formula is as follows:

[0108]

[0109] Where:

[0110] σ is the activation function;

[0111] α t is the attention weight;

[0112] h t is the photovoltaic power prediction result.

[0113] In S107, network partitioning and dynamic aggregation are performed on the photovoltaic area.

[0114] According to the exemplary embodiment, the photovoltaic area is network partitioned. Based on the geographical characteristics of the photovoltaic power station, the distribution of the photovoltaic power station, and the installed capacity information, the grid tool in the geographic information system is used for spatial partitioning. Each grid network unit represents an independent management area, and each independent management area contains one or more photovoltaic power stations and related attribute information such as geographical location and installed capacity.

[0115] Using GIS for spatial partitioning and combining with the fusion clustering algorithm, the photovoltaic cluster partitioning result output by the dynamic aggregation defines the power generation and load characteristics in each grid area, ensuring the scientificity and flexibility of the photovoltaic cluster partitioning.

[0116] Based on the results of photovoltaic power prediction and load demand prediction, through a fusion clustering algorithm combining the K-means and spectral clustering algorithms, the aggregation center of the photovoltaic cluster is determined, and dynamic aggregation is carried out with the goal of minimizing the power imbalance degree of each grid network unit. The formula is as follows:

[0117]

[0118] Among them, is the predicted photovoltaic power generation value of grid G j at time t + k, is the predicted load demand value of grid G j at time t + k. The difference between the two represents the imbalance degree of this grid area.

[0119] In this exemplary embodiment, by adjusting the clustering center, the above objective function is minimized, and the clustering result is iteratively optimized until the convergence condition is reached.

[0120] In S109, based on the output of the dynamic aggregation, the peak shaving strategy is optimized.

[0121] According to the exemplary embodiment, the peak shaving strategy is optimized by using the power generation and load characteristics of each grid network unit, and the power dispatching plan for each time period is calculated and adjusted. The spatial configuration of power resources is determined through dynamic aggregation, and the peak shaving strategy optimization dynamically schedules these resources in the time dimension, thereby achieving global supply-demand balance.

[0122] Let the peak shaving strategy for each time period be S(t), then the peak shaving optimization objective is:

[0123]

[0124] Among them:

[0125] T is the total number of time periods;

[0126] N is the number of grid areas;

[0127] P^Gj(t) is the predicted photovoltaic power generation value of grid Gj at time t;

[0128] L^Gj(t) is the predicted load demand value of grid Gj at time t.

[0129] Through the feedback control mechanism, the peak shaving strategy optimization module can adjust according to the real-time monitored grid operation status:

[0130] S(t) = S(t - 1) + α·(L(t) - P(t))

[0131] Among them, S(t) is the feedback adjustment coefficient, and L(t) and P(t) are the real-time load and photovoltaic power generation respectively.

[0132] By working together through dynamic aggregation and peak shaving strategy optimization, the power system's supply-demand balance can be effectively managed. Dynamic aggregation determines the spatial configuration of power resources, while the peak shaving strategy optimization module conducts dynamic scheduling in the time dimension to ensure the global supply-demand balance. The feedback control mechanism further enhances the system's flexibility and adaptability, enabling the power system to operate stably in a complex and changing environment.

[0133] The dynamic aggregation peak shaving method based on distributed photovoltaic power prediction in this exemplary embodiment combines LSTM (Long Short-Term Memory Network) and GCN (Graph Convolutional Network) to design a photovoltaic power prediction model. LSTM is used to capture time dynamic characteristics, and GCN is used to capture spatial correlations. The combination of the two significantly improves the accuracy of photovoltaic power prediction, providing reliable data support for the dynamic scheduling of the power grid.

[0134] The present invention introduces an attention mechanism for load demand prediction. By dynamically allocating weights to input features, the attention mechanism can identify and focus on the key factors affecting load demand, thereby generating more accurate prediction results. This method improves the flexibility and prediction accuracy of the model, enabling the power grid to more effectively respond to fluctuations in load demand.

[0135] The dynamic peak shaving strategy optimization method continuously adjusts the peak shaving strategy according to the photovoltaic power and load demand prediction results through real-time monitoring and feedback control mechanism. This method can minimize the power imbalance degree in each grid area, optimize the allocation of power resources, improve the utilization rate of photovoltaic power generation, enhance the operation stability of the power grid, and reduce the fluctuation risk caused by supply-demand imbalance.

[0136] Figure 2 A schematic block diagram showing a dynamic aggregation peak shaving system based on distributed photovoltaic power prediction according to an exemplary embodiment is shown.

[0137] According to an exemplary embodiment, the dynamic aggregation peak shaving system based on distributed photovoltaic power prediction consists of a data collection module, a distributed photovoltaic power prediction module, a load demand prediction module, a grid division and dynamic aggregation module, and a peak shaving strategy optimization module.

[0138] The data collection module is used to obtain key data related to photovoltaic power prediction, including historical power generation data of photovoltaic power plants, meteorological data, and geographical information data. The data collection module also cleans and standardizes the collected data to ensure the integrity, consistency, and accuracy of the data, providing high-quality input for the training and application of subsequent prediction models.

[0139] The distributed photovoltaic power prediction module consists of an LSTM network and a GCN network, which are responsible for processing the time-series features and spatial features of distributed photovoltaic power respectively. The LSTM is used to capture the trend of photovoltaic power generation over time, such as considering factors like daily cycles and seasonality; the GCN is used to evaluate the impact of geographical location on photovoltaic power generation, such as the interaction between adjacent regions or the terrain occlusion effect. The distributed photovoltaic power prediction module combines time and spatial features to provide high-precision power prediction results, supporting more effective resource planning and scheduling decisions.

[0140] The load demand prediction module is mainly used to predict the load demand of the power grid in the future time period to ensure that the power grid can schedule power resources in a timely manner according to the load changes. The load demand prediction module adopts a deep learning model based on the attention mechanism, which can process complex and variable load data and generate high-precision load prediction results.

[0141] The grid division and dynamic aggregation module uses a geographic information system and combines a fusion clustering algorithm to perform spatial division and define the power generation and load characteristics within each spatial region, ensuring the scientificity and flexibility of the photovoltaic cluster division. According to the geographical characteristics of the power grid, the distribution of photovoltaic power stations, and information such as installed capacity, the target area is initially divided into grids. Each grid unit represents an independent area in the power grid, containing one or more photovoltaic power stations and related attribute information such as geographical location and installed capacity. The grid division and dynamic aggregation module identifies and quantifies the unique attributes of distributed power sources and loads within each sub-grid, such as average power generation capacity and peak load time, and aggregates distributed energy resources with similar characteristics into virtual power plants for unified management and scheduling, while minimizing the power imbalance between regions.

[0142] The peak shaving strategy optimization module performs dynamic scheduling of resources in the time dimension to ensure the supply-demand balance of the power system. The photovoltaic cluster division results output by the dynamic aggregation module define the power generation and load characteristics within each grid area, and the dynamic aggregation module determines the spatial configuration of power resources. The peak shaving strategy optimization module performs dynamic scheduling of these resources in the time dimension, uses these characteristic data to calculate and adjust the power scheduling plan for each time period, thereby achieving the global supply-demand balance.

[0143] The dynamic aggregation peak shaving system based on distributed photovoltaic power prediction according to the present invention not only improves the utilization rate of distributed photovoltaic power generation through the close cooperation between modules, but also enhances the ability of the power grid to cope with load fluctuations, ensuring the safety and reliability of power supply.

[0144] Figure 3 A block diagram of a computing device according to an exemplary embodiment of the present invention is shown.

[0145] As Figure 3As shown, computing device 30 includes a processor 12 and a memory 14. Computing device 30 may also include a bus 22, a network interface 16, and an I / O interface 18. The processor 12, the memory 14, the network interface 16, and the I / O interface 18 may communicate with each other via the bus 22.

[0146] The processor 12 may include one or more general-purpose CPUs (Central Processing Units), microprocessors, or application-specific integrated circuits, etc., for executing relevant program instructions.

[0147] The memory 14 may include a machine system-readable medium in the form of volatile memory, such as random access memory (RAM), read-only memory (ROM), and / or cache memory. The memory 14 is used to store one or more programs containing instructions as well as data. The processor 12 can read the instructions stored in the memory 14 to execute the method according to the embodiments of the present invention described above.

[0148] The computing device 30 may also communicate with one or more networks via the network interface 16. The network interface 16 may be a wireless network interface.

[0149] The bus 22 may include an address bus, a data bus, a control bus, etc. The bus 22 provides a path for exchanging information between components.

[0150] It should be noted that in the specific implementation process, the computing device 30 may also include other components necessary for normal operation. In addition, those skilled in the art can understand that the above devices may also only include the components necessary to implement the solution of the embodiments of this specification, and do not have to include all the components shown in the figure.

[0151] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical cards, nanosystems (including molecular memory ICs), network storage devices, cloud storage devices, or any type of medium or device suitable for storing instructions and / or data.

[0152] The embodiments of the present invention also provide a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps of any one of the methods described in the above method embodiments.

[0153] Those skilled in the art can clearly understand that the technical solution of the present invention can be implemented by means of software and / or hardware. The "units" and "modules" in this specification refer to software and / or hardware that can complete specific functions independently or in cooperation with other components, where the hardware can be, for example, a field programmable gate array, an integrated circuit, etc.

[0154] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0155] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0156] In the several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some service interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical or other form.

[0157] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0158] In addition, the functional units in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0159] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention.

[0160] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0161] The above specifically shows and describes the exemplary embodiments of the present invention. It should be understood that the present invention is not limited to the detailed structures, setting manners, or implementation methods described herein; on the contrary, the present invention is intended to cover various modifications and equivalent settings included within the spirit and scope of the appended claims.

[0162] Those skilled in the art can clearly understand that the technical solution of the present invention can be implemented by means of software and / or hardware. The "units" and "modules" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field-programmable gate array, an integrated circuit, etc.

[0163] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequences, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0164] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0165] In several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling, or communication connection between each other can be through some service interfaces. The indirect coupling or communication connection of devices or units can be in electrical or other forms.

[0166] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0167] In addition, each functional unit in various embodiments of the present invention can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0168] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present invention.

[0169] In the above embodiments, the descriptions of each embodiment have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0170] The above specifically shows and describes the exemplary embodiments of the present invention. It should be understood that the present invention is not limited to the detailed structures, setting methods, or implementation methods described here; on the contrary, the present invention is intended to cover various modifications and equivalent settings included within the spirit and scope of the appended claims.

Claims

1. A dynamic aggregation peak shaving method based on distributed photovoltaic power prediction, characterized in that Including: Obtaining data of a photovoltaic power station through a photovoltaic power station monitoring system, a meteorological station, and a geographic information system; Performing photovoltaic power prediction using a distributed photovoltaic power prediction model; Performing load demand prediction on the load for a future time period; Conducting network partitioning and dynamic aggregation for the photovoltaic area; Optimizing the peak shaving strategy based on the output of the dynamic aggregation.

2. The dynamic aggregation peak shaving method according to claim 1, wherein Performing photovoltaic power prediction using a distributed photovoltaic power prediction model includes: Processing the time series characteristics of distributed photovoltaic power using a long short-term memory network model; Using a graph convolutional network model to process the spatial characteristics of distributed photovoltaic power; Fusing the time characteristics and the spatial characteristics together to generate a power prediction value for the photovoltaic power station.

3. The dynamic aggregation peak shaving method according to claim 2, wherein The update formula of the graph convolutional network is: H (l+1) = σ(D -1 / 2 AD -1 / 2 H (l) W (l) ) Among them, σ is the activation function, D is the degree matrix of the graph, D -1 / 2 is a diagonal matrix, A is the adjacency matrix between photovoltaic power stations, representing the connection relationship between nodes, H (l) is the input feature of the l-th layer of the graph convolutional network, W (l) is the trainable parameter matrix of the graph convolutional network, which is used to transform node features.

4. The dynamic aggregation peak shaving method according to claim 2, wherein The formula for the distributed photovoltaic power prediction is: Among them, H GCN is the output feature of the graph convolutional network, h LSTM is the output feature of the long short-term memory network, FC represents the fully connected layer, and ⊕ represents the feature concatenation operation. is the predicted power of the i-th power station at time t + k.

5. The dynamic aggregation peak shaving method according to claim 1, wherein Conducting network partitioning and dynamic aggregation for the photovoltaic area includes: Using the raster tool in the geographic information system to perform spatial partitioning according to the geographical characteristics of the photovoltaic power station, the distribution of the photovoltaic power station, and the installed capacity information, and each raster network unit in the raster tool represents an independent management area.

6. The dynamic aggregation peak shaving method according to claim 1, characterized in that Conducting network partitioning and dynamic aggregation for the photovoltaic area further includes: Determining the aggregation center of the photovoltaic clusters through the fusion clustering algorithm for the photovoltaic power prediction result and the load demand prediction result, and the formula used for dynamic aggregation: Among them, is the grid G j the predicted value of the photovoltaic power generation at time t + k, is the grid G j the predicted value of the load demand at time t + k, and the difference between the two represents the imbalance degree of this grid area.

7. The dynamic aggregation peak shaving method according to claim 1, characterized in that, Optimizing the peak shaving strategy based on the output of the dynamic aggregation includes: Adjusting the real-time monitored power grid operation state through a feedback control mechanism, and the feedback control formula: S(t) = S(t - 1)+α·(L(t)-P(t)) Wherein, S(t) is the feedback adjustment coefficient, L(t) and P(t) are the real-time load and photovoltaic power generation respectively, and α is the adjustment coefficient.

8. The dynamic aggregation peak shaving method according to claim 1, wherein Obtaining data of the photovoltaic power station includes historical power generation data, meteorological data, and geographic information data of the photovoltaic power station.

9. A system for dynamic aggregated peak shaving based on distributed photovoltaic power prediction, characterized in that, Including: A data collection module for obtaining data related to photovoltaic power prediction; A distributed photovoltaic power prediction module responsible for processing the time series characteristics and spatial characteristics of distributed photovoltaic power; A load demand prediction module for predicting the load demand of the power grid for a future time period; A grid partitioning and dynamic aggregation module using the geographic information system and combining with the fusion clustering algorithm to perform spatial partitioning and define the power generation and load characteristics within each spatial area; A peak shaving strategy optimization module for dynamically scheduling resources in the time dimension.

10. A computing device, characterized in that, Including: A processor; And A memory storing a computer program, and when the computer program is executed by the processor, the processor executes the method according to any one of claims 1 - 8.

Images