Virtual power plant power scheduling method and device based on model predictive control

Abstract

The application provides a virtual power plant power scheduling method and device based on model predictive control, which comprises the following steps: establishing a regional space topology correlation model according to multi-source data in a virtual power plant coverage area; performing feature extraction on the multi-source data in the virtual power plant coverage area to obtain a regional feature set; performing feature fusion on the regional features to obtain a regional portrait feature; constructing a regional dynamic portrait according to the regional portrait feature; performing probabilistic regional attribute identification on the virtual power plant coverage area according to the dynamic regional portrait to obtain an attribute probability result; forming a regional portrait label according to the attribute probability result; constructing a regional power state prediction model according to the regional portrait label; calculating a regional regulation capacity index according to the regional portrait label; and constructing a virtual power plant power optimization scheduling model according to a regional power prediction result, the regional regulation capacity index and a regional new energy state.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a virtual power plant power dispatching method and apparatus based on model predictive control. Background Technology

[0002] With the rapid development of new energy power generation technologies, distributed energy resources, and new load equipment, the power system is gradually shifting from the traditional centralized energy supply model to a multi-source collaborative interaction model. Virtual power plants aggregate various types of energy resources, such as distributed photovoltaics, wind power, energy storage equipment, adjustable loads, and electric vehicles, and achieve resource synergy optimization through digital control and intelligent dispatching technologies. This has become an important technical means to improve the capacity for new energy absorption and enhance the flexible regulation capability of the power grid.

[0003] During the operation of a virtual power plant, significant differences exist in user types, energy resource composition, and energy consumption behavior across different power supply areas, resulting in varying load variation patterns and renewable energy absorption characteristics in each area. Therefore, accurate identification of the characteristics of different areas within the virtual power plant is crucial for achieving precise load forecasting, optimal resource allocation, and dynamic dispatch control.

[0004] Existing methods for analyzing the regional characteristics of virtual power plants typically divide regions based on the distribution network topology, user categories, and historical load curves. Specifically, as disclosed in Chinese patent application CN120728597A, regional boundaries are determined based on substations, distribution nodes, and user access relationships. Then, combined with user types, the regions are divided into residential, commercial, or industrial areas, and corresponding load forecasting models are established based on different region types.

[0005] For example, existing methods generally assume that the load in commercial areas is mainly concentrated during the daytime working hours, and since commercial buildings are usually equipped with rooftop photovoltaics, the output of new energy sources is highly matched with the load changes. For residential areas, it is generally assumed that the load is concentrated during the evening electricity consumption period, while the output of new energy sources mainly occurs during the daytime, and there is a significant time mismatch between the two.

[0006] However, with the development of integrated energy systems, traditional analysis methods based on fixed regional types have gradually revealed their shortcomings.

[0007] First, existing regional classification methods are coarse-grained, typically using discrete labels such as "residential area," "commercial area," and "industrial area" to describe regional attributes, which makes it difficult to accurately represent the energy behavior of complex real-world areas. For example, mixed-use developments exhibit characteristics of both commercial and residential loads, and their load variation patterns differ from those of traditional commercial areas and traditional residential areas. Industrial parks may also contain production loads, distributed photovoltaic systems, and energy storage devices, making their energy interactions more complex and difficult to accurately describe using a single label.

[0008] Secondly, existing methods primarily rely on historical load data and user type information for region identification, lacking comprehensive analysis of multi-source information such as renewable energy output, energy storage operation status, environmental factors, and changes in user behavior. This results in insufficient dimensions for representing regional characteristics. For example, two regions belonging to the same commercial area may have significant differences in building type, photovoltaic installed capacity, energy storage configuration, and user activity patterns. However, existing methods typically treat them as the same type of region, reducing the adaptability of subsequent prediction models.

[0009] Secondly, existing regional attributes are typically expressed using deterministic labels, directly defining a region as a specific type of region. However, in actual operation, regional energy attributes are dynamic and subject to change. For example, with the increase in electric vehicle charging facilities, the integration of distributed renewable energy sources, and the expansion of commercial facilities, previously residential areas may gradually transform into integrated energy areas, and fixed regional labels cannot reflect these changes in regional status in real time.

[0010] Furthermore, existing technologies typically focus only on load and renewable energy characteristics within a region, neglecting the mutual influences between different regions caused by the interconnections of distribution networks. For example, rapid load growth in one region may lead to voltage fluctuations or power flow changes in adjacent regions, and traditional regional analysis methods cannot effectively describe the inter-regional relationships.

[0011] Therefore, existing virtual power plant regional characteristic analysis technologies suffer from problems such as coarse regional classification, insufficient feature expression, weak dynamic adaptability, and lack of regional correlation, making it difficult to meet the needs of refined prediction and intelligent scheduling of virtual power plants under the condition of high proportion of new energy access. Summary of the Invention

[0012] This invention provides a virtual power plant power dispatching method based on model predictive control (MMC). It addresses the shortcomings of existing virtual power plant regional characteristic analysis techniques, such as coarse regional classification, insufficient feature representation, weak dynamic adaptability, and missing regional correlations. These shortcomings make it difficult to meet the needs of refined prediction and intelligent dispatching of virtual power plants under conditions of high renewable energy integration. By integrating distribution network topology data, regional load operation data, renewable energy output data, energy storage operation data, and environmental behavior data, a multi-dimensional regional feature system is constructed, encompassing spatial attributes, load behavior attributes, energy interaction attributes, and dynamic evolution attributes. Furthermore, a dynamic feature embedding model is used to extract deep features of regional energy behavior, generating a dynamic regional profile that characterizes the overall energy status of the region. Simultaneously, a probabilistic regional attribute identification method replaces the traditional fixed-label classification method, enabling accurate identification of complex regions, multi-type regions, and dynamically evolving regions. This improves the accuracy of regional characteristic identification and provides more accurate data support for virtual power plant load prediction, resource dispatching, and renewable energy consumption optimization.

[0013] This invention provides a virtual power plant power dispatching method based on model predictive control, comprising the following steps.

[0014] Based on multi-source data within the coverage area of ​​the virtual power plant, a regional spatial topology correlation model is established.

[0015] Feature extraction is performed on multi-source data within the coverage area of ​​the virtual power plant to obtain a set of regional features.

[0016] The region features are fused to obtain the region profile features;

[0017] Based on the aforementioned regional profile features, a dynamic regional profile is constructed;

[0018] Based on the dynamic regional profile, probabilistic regional attribute identification is performed on the virtual power plant coverage area to obtain attribute probability results.

[0019] Based on the attribute probability results, regional profile tags are generated;

[0020] Based on the regional profile labels, a regional power state prediction model is constructed;

[0021] Calculate the regional regulation capacity index based on the regional profile labels;

[0022] Based on regional power forecast results, regional regulation capacity indicators, and regional renewable energy status, a virtual power plant power optimization scheduling model is constructed.

[0023] According to the present invention, a virtual power plant power dispatching method based on model predictive control is provided, wherein establishing a regional spatial topology correlation model based on multi-source data within the coverage area of ​​the virtual power plant includes:

[0024] Based on the connection relationships and electrical characteristics between nodes within the coverage area of ​​the virtual power plant, the coverage area of ​​the virtual power plant is divided into regional units to obtain a set of regions; among which, electrical characteristics include user nodes with the same power supply path and similar electrical relationships;

[0025] The regional spatial topology association model is established by taking each regional unit in the aforementioned regional spatial topology association model as a node in the graph model.

[0026] According to the present invention, a virtual power plant power dispatching method based on model predictive control is provided, wherein feature extraction is performed on multi-source data within the coverage area of ​​the virtual power plant to obtain a regional feature set, including:

[0027] Feature extraction is performed on regional load operation data to obtain load behavior characteristics;

[0028] Feature extraction is performed on distributed renewable energy operation data to obtain renewable energy characteristics;

[0029] Feature extraction is performed on the operating data of energy storage devices to obtain energy storage regulation characteristics;

[0030] Feature extraction is performed on the power grid topology data to obtain spatial correlation features.

[0031] According to the present invention, a virtual power plant power dispatching method based on model predictive control is provided, wherein constructing a regional dynamic profile based on the regional profile features includes:

[0032] The spatial correlation features of the region are obtained by propagating spatial correlation features of the region profile features through a graph neural network.

[0033] Based on the regional state sequence, the temporal evolution features of the region are extracted using a temporal attention mechanism.

[0034] By integrating the spatial correlation features and temporal evolution features of the region, a dynamic regional profile is obtained.

[0035] According to the present invention, a virtual power plant power dispatching method based on model predictive control includes, in which, based on the dynamic regional profile, probabilistic regional attribute identification is performed on the virtual power plant coverage area to obtain attribute probability results, including:

[0036] A fully connected mapping is performed on the dynamic region profile to obtain the region state vector;

[0037] Based on the region state and a preset set of region attribute categories, the probability of a region belonging to each type is calculated using a classification model to obtain the attribute probability results.

[0038] According to the present invention, a virtual power plant power dispatching method based on model predictive control is provided, wherein constructing a regional power state prediction model based on the regional profile label includes:

[0039] Based on the region profile labels, construct the region power state input features;

[0040] A regional power state prediction model is established based on the regional power state input features and regional profile labels.

[0041] According to the present invention, a virtual power plant power dispatching method based on model predictive control is provided. The virtual power plant power optimization dispatching model includes regional power balance constraints and regional differentiated dispatching objective functions. The regional differentiated dispatching objective functions include electricity purchase cost objectives, renewable energy consumption objectives, regional regulation balance objectives, and regional operation risk objectives.

[0042] The present invention also provides a virtual power plant power dispatching device based on model predictive control, comprising the following modules:

[0043] The spatial topology module is used to establish a regional spatial topology association model based on multi-source data within the coverage area of ​​the virtual power plant;

[0044] The feature set module is used to extract features from multi-source data within the coverage area of ​​the virtual power plant to obtain a regional feature set.

[0045] The portrait feature module is used to perform feature fusion on the region features to obtain region portrait features;

[0046] The dynamic profiling module is used to construct a dynamic profiling of the region based on the region's profiling features.

[0047] The attribute recognition module is used to perform probabilistic regional attribute recognition on the virtual power plant coverage area based on the dynamic regional profile, and obtain the attribute probability result.

[0048] The profile tagging module is used to generate regional profile tags based on the attribute probability results;

[0049] The prediction model module is used to construct a regional power state prediction model based on the regional profile labels.

[0050] The regulation capacity calculation module is used to calculate the regional regulation capacity index based on the regional profile labels.

[0051] The optimization scheduling module is used to construct a virtual power plant power optimization scheduling model based on regional power forecast results, regional regulation capacity indicators, and regional renewable energy status.

[0052] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the virtual power plant power dispatching method based on model predictive control as described above.

[0053] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the virtual power plant power dispatching method based on model predictive control as described above.

[0054] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the virtual power plant power dispatching method based on model predictive control as described above.

[0055] This invention provides a virtual power plant power dispatching method based on model predictive control. The method establishes a regional spatial topology association model based on multi-source data within the virtual power plant's coverage area; extracts features from the multi-source data within the virtual power plant's coverage area to obtain a regional feature set; fuses these regional features to obtain regional profile features; constructs a dynamic regional profile based on these regional profile features; identifies probabilistic regional attributes of the virtual power plant's coverage area based on the dynamic regional profile to obtain attribute probability results; forms regional profile labels based on the attribute probability results; constructs a regional power state prediction model based on the regional profile labels; calculates regional regulation capacity indicators based on the regional profile labels; and constructs a virtual power plant power optimization dispatching model based on the regional power prediction results, regional regulation capacity indicators, and regional renewable energy status. This method achieves accurate identification of complex regions, multi-type regions, and dynamically evolving regions, improving the accuracy of regional characteristic identification and providing more accurate data support for virtual power plant load forecasting, resource dispatching, and renewable energy consumption optimization. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating the virtual power plant power scheduling method based on model predictive control provided by the present invention.

[0058] Figure 2 This is a schematic diagram of the structure of the virtual power plant power dispatching device based on model predictive control provided by the present invention.

[0059] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0061] The following is combined with Figures 1 to 3 This invention is described.

[0062] Figure 1 This is a flowchart illustrating the virtual power plant power dispatching method based on model predictive control provided by the present invention, as shown below. Figure 1 As shown, the method includes the following:

[0063] Step 101: Establish a regional spatial topology association model based on multi-source data within the coverage area of ​​the virtual power plant.

[0064] In step 101, the multi-source data includes distribution network topology data, regional load operation data, distributed new energy operation data, energy storage device operation data, and environmental impact data.

[0065] The power distribution network topology data is used to describe the electrical connection relationships between regions and between nodes within a region; the regional load operation data is used to describe user-side energy consumption behavior; the distributed renewable energy operation data is used to describe renewable energy output characteristics; the energy storage device operation data is used to describe the regional adjustability; and the environmental impact data is used to describe the impact of external factors on regional energy behavior.

[0066] Specifically:

[0067] The power distribution network topology data includes substation node information, power line connection relationships, power distribution transformer information, user access node information, and node voltage level information.

[0068] The regional load operation data includes historical load power sequences, real-time load power sequences, user load types, and load change cycles.

[0069] The new energy operation data includes photovoltaic output power, wind power output power, new energy installed capacity, and new energy output fluctuation characteristics.

[0070] The energy storage operation data includes rated energy storage capacity, current state of charge, charging and discharging power, and adjustment response time.

[0071] The environmental impact data includes temperature data, humidity data, meteorological conditions, and date type information.

[0072] In one embodiment, step 101 includes steps A1 to A2.

[0073] Step A1: Based on the connection relationships and electrical characteristics between nodes within the virtual power plant coverage area, divide the virtual power plant coverage area into regional units to obtain a set of regions; among which, electrical characteristics include user nodes with the same power supply path and similar electrical relationships.

[0074] Step A2: Use each regional unit in the regional spatial topology association model as a node in the graph model to establish the regional spatial topology association model.

[0075] Specifically, in step A1, the set of regions is represented as:

[0076]

[0077] in, Indicates the first Each regional unit Indicates the number of regions.

[0078] In step A2, the regional spatial topological association model is represented as follows:

[0079]

[0080] in, For a set of region nodes, It is a set of connections between regions. This is the regional correlation weight matrix. The method for calculating the degree of electrical influence between different areas includes:

[0081]

[0082] in, Indicates the region With the region Electrical distance between them This represents the line impedance parameters between regions. This represents the parameters affecting the inter-regional power flow.

[0083] By establishing a regional spatial topological relationship model, the analysis of regional characteristics is no longer limited to the interior of a single region, but can take into account the mutual influence between regions.

[0084] Step 102: Extract features from multi-source data within the coverage area of ​​the virtual power plant to obtain a regional feature set; wherein, the regional feature set includes load behavior features, new energy features, energy storage regulation features, and spatial correlation features.

[0085] In one embodiment, step 102 includes steps B1 to B4:

[0086] Step B1: Extract features from the regional load operation data to obtain load behavior characteristics;

[0087] Step B2: Extract features from the distributed new energy operation data to obtain new energy features;

[0088] Step B3: Extract features from the operating data of the energy storage device to obtain energy storage regulation features;

[0089] Step B4: Extract features from the power grid topology data to obtain spatial correlation features.

[0090] It should be noted that there is no constraint on the execution order between steps B1 and B4.

[0091] Specifically, regarding step B1, load peak characteristics, peak-to-valley difference characteristics, load fluctuation characteristics, and periodic change characteristics are extracted from the regional historical load sequence in the regional load operation data.

[0092] The regional historical load sequence is denoted as:

[0093]

[0094] The characteristics of peak load are:

[0095]

[0096] Peak load characteristics are used to describe the patterns of peak electricity demand in a region.

[0097] The characteristics of peak-valley difference are:

[0098]

[0099] Peak-valley difference characteristics are used to describe the magnitude of regional load variation.

[0100] Load fluctuation characteristics are expressed using the load standard deviation as follows:

[0101]

[0102] in, Indicates the number of sampling periods. This indicates the average load in the region.

[0103] Periodic variation characteristics are derived by analyzing daily, weekly, and seasonal cycles to form periodic characteristic parameters.

[0104] Regarding step B2, based on the output power of the new energy source:

[0105]

[0106] Calculate the penetration rate of new energy sources:

[0107]

[0108] in, For the maximum output power of new energy, This represents the maximum load in the region.

[0109] The new energy fluctuation index is:

[0110]

[0111] The new energy fluctuation index is used to describe the stability of new energy output.

[0112] Regarding step B3, based on the energy storage operating status, construct:

[0113]

[0114] in, Indicates ability, Indicates discharge capability, Indicates the state of charge.

[0115] Regarding step B4, to avoid judging the renewable energy absorption capacity solely based on time overlap, this invention further incorporates regional spatial correlation to establish an energy coupling evaluation index, namely, spatial correlation characteristics:

[0116]

[0117] in, To determine the degree of time matching between new energy sources and load, To ensure the degree of matching between new energy sources and load space, For electrical connection matching degree, These are the weighting coefficients.

[0118] The degree of time matching is as follows:

[0119]

[0120] The degree of spatial matching is determined based on the distance between the new energy node and the load node:

[0121]

[0122] The electrical matching degree is determined based on the line impedance and power flow capacity.

[0123] Step 103: Perform feature fusion on the regional features to obtain regional profile features.

[0124] In step 103, for any region Construct regional profile features:

[0125]

[0126] in, Indicates load behavior characteristics, Indicates the characteristics of new energy sources. Indicates energy storage regulation characteristics, It represents spatial correlation characteristics.

[0127] Through the above steps 102 and 103, a multi-dimensional regional profile feature that can describe regional energy behavior is formed.

[0128] Step 104: Construct a dynamic regional profile based on the regional profile features.

[0129] In one embodiment, step 104 includes steps C1 to C2.

[0130] Step C1: Propagate spatial correlation features of regional profile features through graph neural networks to obtain regional spatial correlation features;

[0131] Step C2: Extract the temporal evolution features of the region based on the regional state sequence using a temporal attention mechanism;

[0132] Step C3: Integrate regional spatial correlation features and regional temporal evolution features to obtain a dynamic regional profile.

[0133] Specifically, in step C1, for any region node Its initial node features are represented as follows:

[0134]

[0135] in, To create a regional profile feature. This represents the initial state characteristics of the regional nodes.

[0136] Achieving inter-region feature propagation through graph convolution:

[0137]

[0138] in, Indicates the first Regions in layered networks The hidden state, Representation and region A set of neighboring regions that are related. Indicates the inter-regional correlation weights. Indicates the first Layer network parameters, This represents a non-linear activation function.

[0139] Define the regional spatial association characteristics as , .

[0140] Through the above processing, each regional node not only includes its own energy behavior characteristics, but also incorporates the spatial impact of adjacent regions on it.

[0141] For example, when a region has a high proportion of new energy access, a large-scale charging load access, and load growth during peak hours, the impact information can be propagated to related regions through regional topology, thereby improving the accuracy of regional status description.

[0142] In step C2, for the region Obtain the regional state sequence within a continuous time window:

[0143]

[0144] in, Indicates the length of the historical time window. Indicates the current time.

[0145] Calculate the influence weights of different historical moments on the current region state using a time attention mechanism:

[0146]

[0147] in, This represents the query vector at the current moment. Represents the feature vector of historical states. This represents the time attention weight.

[0148] Regional temporal evolution features are obtained based on temporal attention weights:

[0149]

[0150] in, This indicates the temporal evolution characteristics of the region.

[0151] The above methods can identify regional load growth trends, new energy access change trends, and user behavior migration trends, avoiding the misjudgment of regional attributes caused by traditional methods that rely solely on fixed historical periods.

[0152] In step C3, the regional spatial correlation features and regional temporal evolution features are fused:

[0153]

[0154] in, Indicates spatial correlation characteristics, Represents dynamic characteristics over time. This represents the final dynamic image of the region.

[0155] Furthermore, through fully connected mapping:

[0156]

[0157] Obtain the region state vector for region attribute identification.

[0158] in, For the mapping matrix, This is the bias parameter.

[0159] The regional state vector can simultaneously express the regional energy composition, user load behavior, new energy absorption capacity, spatial correlation impact, and dynamic change trend.

[0160] Step 105: Based on the dynamic region profile, perform probabilistic region attribute identification on the virtual power plant coverage area to obtain attribute probability results.

[0161] In step 105, a set of regional attribute categories is pre-set:

[0162]

[0163] The categories include, but are not limited to, residential areas, commercial and office areas, industrial production areas, integrated energy areas, new energy high-penetration areas, and energy storage and regulation areas.

[0164] Based on the region state vector, the probability of a region belonging to each type is calculated using a classification model:

[0165]

[0166] in, Indicates the region Category The probability of.

[0167] For example, for a certain region, the recognition result can be represented as:

[0168] Regional attributes probability Commercial office type 0.62 Residential living type 0.18 Integrated energy type 0.15 Industrial production type 0.05

[0169] This method can preserve the uncertainty of regional attributes.

[0170] Step 106: Based on the attribute probability results, generate regional profile tags.

[0171]

[0172] Each probability value corresponds to a regional energy attribute.

[0173] At the same time, regional dynamic characteristics With regional attribute probability Together they serve as inputs for the subsequent load forecasting and scheduling model of the virtual power plant.

[0174] Step 107: Construct a regional power state prediction model based on the regional profile labels.

[0175] In one embodiment, step 107 includes steps D1 and D2:

[0176] Step D1: Construct regional power state input features based on the region profile labels;

[0177] Step D2: Based on the regional power state input features and regional profile labels, establish a regional power state prediction model.

[0178] In step D1, for the first Each region, in time Constructing regional power state input features at every moment:

[0179]

[0180] in, Create dynamic region profiles; Historical load sequence for the region; Output power for new energy sources; In energy storage operation mode; This is an adjustable resource status.

[0181] In step D2, the regional power state prediction model is as follows:

[0182]

[0183] in, Forecasted net load for the region over the next period; Regional power prediction model.

[0184] The net load of the area is defined as follows:

[0185]

[0186] when This indicates that there is a demand for electricity in the area; when This indicates that the region has a surplus of new energy sources.

[0187] Step 108: Calculate the regional adjustment capability index based on the regional profile labels.

[0188] Because different regions have different energy structures, their power regulation capabilities vary. Therefore, regional regulation capability indicators are further calculated based on dynamic regional profiles.

[0189] Build

[0190] in, This is an indicator of regional regulation capacity. Energy storage capacity; Adjustable load capacity.

[0191] Calculate the energy storage regulation capability and define the energy storage response capability:

[0192]

[0193] in, This represents the remaining energy storage capacity; This is the maximum charge / discharge power; For weights.

[0194] Calculate load regulation capacity and determine the proportion of adjustable load based on load behavior characteristics in the regional profile:

[0195]

[0196] in, Adjustable load power; This represents the total load of the region.

[0197] Form a comprehensive evaluation of regulatory capacity:

[0198]

[0199] in, To enhance the region's capacity to absorb new energy sources; These are the weight parameters.

[0200] Output region adjustment capability matrix:

[0201]

[0202] Step 109: Based on the regional power prediction results, regional regulation capacity indicators, and regional renewable energy status, construct a virtual power plant power optimization scheduling model.

[0203] In step 109, regional power balance constraints are established:

[0204]

[0205] in, In order to exchange power with the main power grid; For energy storage and power regulation; This is the adjustable load response power.

[0206] Establish a regionally differentiated scheduling objective function:

[0207]

[0208] in, To achieve the target of electricity purchase cost, reduce the cost of electricity purchase from the power grid; To achieve the goal of absorbing renewable energy and reduce the curtailment of renewable energy; To achieve the goal of regional regulation and balance, priority should be given to utilizing regions with high regulation capacity; It serves as a regional operational risk target to prevent voltage overruns, load overloads, and backfeeding of renewable energy.

[0209]

[0210]

[0211]

[0212]

[0213] in .

[0214] This invention provides a virtual power plant power dispatching method based on model predictive control. By integrating distribution network topology data, regional load operation data, renewable energy output data, energy storage operation data, and environmental behavior data, a multi-dimensional regional feature system is constructed, including spatial attributes, load behavior attributes, energy interaction attributes, and dynamic evolution attributes. Furthermore, a dynamic feature embedding model is used to extract deep features of regional energy behavior, generating a dynamic regional profile that characterizes the overall energy status of the region. Simultaneously, a probabilistic regional attribute identification method replaces the traditional fixed-label classification method, achieving accurate identification of complex regions, multi-type regions, and dynamically evolving regions. This improves the accuracy of regional characteristic identification and provides more accurate data support for virtual power plant load forecasting, resource dispatching, and renewable energy consumption optimization.

[0215] The virtual power plant power dispatching device based on model predictive control provided by the present invention will be described below. The virtual power plant power dispatching device based on model predictive control described below can be referred to in correspondence with the virtual power plant power dispatching method based on model predictive control described above.

[0216] Figure 2This is a schematic diagram of the process of the virtual power plant power dispatching device based on model predictive control provided by the present invention, as shown below. Figure 2 As shown, the device includes the following:

[0217] Spatial topology module 201 is used to establish a regional spatial topology association model based on multi-source data within the coverage area of ​​the virtual power plant;

[0218] Feature set module 202 is used to extract features from multi-source data within the coverage area of ​​the virtual power plant to obtain a regional feature set;

[0219] The portrait feature module 203 is used to perform feature fusion on the region features to obtain region portrait features;

[0220] The dynamic portrait module 204 is used to construct a dynamic portrait of the region based on the region portrait features;

[0221] The attribute recognition module 205 is used to perform probabilistic regional attribute recognition on the virtual power plant coverage area based on the dynamic regional profile, and obtain the attribute probability result.

[0222] The profile tag module 206 is used to generate regional profile tags based on the attribute probability results;

[0223] The prediction model module 207 is used to construct a regional power state prediction model based on the regional profile labels.

[0224] The regulation capacity calculation module 208 is used to calculate the regional regulation capacity index based on the regional profile label;

[0225] The optimized scheduling module 209 is used to construct a virtual power plant power optimized scheduling model based on regional power prediction results, regional regulation capacity indicators, and regional renewable energy status.

[0226] This invention provides a virtual power plant power dispatching device based on model predictive control. By integrating distribution network topology data, regional load operation data, renewable energy output data, energy storage operation data, and environmental behavior data, it constructs a multi-dimensional regional feature system that includes spatial attributes, load behavior attributes, energy interaction attributes, and dynamic evolution attributes. Furthermore, it utilizes a dynamic feature embedding model to perform deep feature extraction on regional energy behavior, generating a dynamic regional profile that can characterize the comprehensive energy status of the region. Simultaneously, it replaces the traditional fixed label classification method with a probabilistic regional attribute identification method, achieving accurate identification of complex regions, multi-type regions, and dynamically evolving regions. This improves the accuracy of regional characteristic identification and provides more accurate data support for virtual power plant load forecasting, resource dispatching, and renewable energy consumption optimization.

[0227] Figure 3An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logic instructions in the memory 830 to execute a virtual power plant power dispatching method based on model predictive control, including:

[0228] Based on multi-source data within the coverage area of ​​the virtual power plant, a regional spatial topology correlation model is established.

[0229] Feature extraction is performed on multi-source data within the coverage area of ​​the virtual power plant to obtain a set of regional features.

[0230] The region features are fused to obtain the region profile features;

[0231] Based on the aforementioned regional profile features, a dynamic regional profile is constructed;

[0232] Based on the dynamic regional profile, probabilistic regional attribute identification is performed on the virtual power plant coverage area to obtain attribute probability results.

[0233] Based on the attribute probability results, regional profile tags are generated;

[0234] Based on the regional profile labels, a regional power state prediction model is constructed;

[0235] Calculate the regional regulation capacity index based on the regional profile labels;

[0236] Based on regional power forecast results, regional regulation capacity indicators, and regional renewable energy status, a virtual power plant power optimization scheduling model is constructed.

[0237] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0238] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the virtual power plant power dispatching method based on model predictive control provided by the above methods.

[0239] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the virtual power plant power dispatching method based on model predictive control provided by the above methods.

[0240] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0241] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0242] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A virtual power plant power dispatching method based on model predictive control, characterized in that, include: Based on multi-source data within the coverage area of ​​the virtual power plant, a regional spatial topology correlation model is established. Feature extraction is performed on multi-source data within the coverage area of ​​the virtual power plant to obtain a set of regional features. The region features are fused to obtain the region profile features; Based on the aforementioned regional profile features, a dynamic regional profile is constructed; Based on the dynamic regional profile, probabilistic regional attribute identification is performed on the virtual power plant coverage area to obtain attribute probability results. Based on the attribute probability results, regional profile tags are generated; Based on the regional profile labels, a regional power state prediction model is constructed; Calculate the regional regulation capacity index based on the regional profile labels; Based on regional power forecast results, regional regulation capacity indicators, and regional renewable energy status, a virtual power plant power optimization scheduling model is constructed.

2. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The step of establishing a regional spatial topology association model based on multi-source data within the coverage area of ​​the virtual power plant includes: Based on the connection relationships and electrical characteristics between nodes within the coverage area of ​​the virtual power plant, the coverage area of ​​the virtual power plant is divided into regional units to obtain a set of regions; among which, electrical characteristics include user nodes with the same power supply path and similar electrical relationships; The regional spatial topology association model is established by taking each regional unit in the aforementioned regional spatial topology association model as a node in the graph model.

3. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The feature extraction of multi-source data within the coverage area of ​​the virtual power plant yields a regional feature set, including: Feature extraction is performed on regional load operation data to obtain load behavior characteristics; Feature extraction is performed on distributed renewable energy operation data to obtain renewable energy characteristics; Feature extraction is performed on the operating data of energy storage devices to obtain energy storage regulation characteristics; Feature extraction is performed on the power grid topology data to obtain spatial correlation features.

4. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The step of constructing a dynamic regional profile based on the regional profile features includes: The spatial correlation features of the region are obtained by propagating spatial correlation features of the region profile features through a graph neural network. Based on the regional state sequence, the temporal evolution features of the region are extracted using a temporal attention mechanism. By integrating the spatial correlation features and temporal evolution features of the region, a dynamic regional profile is obtained.

5. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The step of performing probabilistic regional attribute identification on the virtual power plant coverage area based on the dynamic regional profile to obtain attribute probability results includes: A fully connected mapping is performed on the dynamic region profile to obtain the region state vector; Based on the region state and a preset set of region attribute categories, the probability of a region belonging to each type is calculated using a classification model to obtain the attribute probability results.

6. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The step of constructing a regional power state prediction model based on the regional profile labels includes: Based on the region profile labels, construct the region power state input features; A regional power state prediction model is established based on the regional power state input features and regional profile labels.

7. The virtual power plant power dispatching method based on model predictive control according to claim 1, characterized in that, The virtual power plant power optimization scheduling model includes regional power balance constraints and regional differentiated scheduling objective functions. The regional differentiated scheduling objective functions include electricity purchase cost targets, renewable energy consumption targets, regional regulation balance targets, and regional operation risk targets.

8. A virtual power plant power dispatching device based on model predictive control, characterized in that, include: The spatial topology module is used to establish a regional spatial topology association model based on multi-source data within the coverage area of ​​the virtual power plant; The feature set module is used to extract features from multi-source data within the coverage area of ​​the virtual power plant to obtain a regional feature set. The portrait feature module is used to perform feature fusion on the region features to obtain region portrait features; The dynamic profiling module is used to construct a dynamic profiling of the region based on the region's profiling features. The attribute recognition module is used to perform probabilistic regional attribute recognition on the virtual power plant coverage area based on the dynamic regional profile, and obtain the attribute probability result. The profile tagging module is used to generate regional profile tags based on the attribute probability results; The prediction model module is used to construct a regional power state prediction model based on the regional profile labels. The regulation capacity calculation module is used to calculate the regional regulation capacity index based on the regional profile labels. The optimization scheduling module is used to construct a virtual power plant power optimization scheduling model based on regional power forecast results, regional regulation capacity indicators, and regional renewable energy status.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the virtual power plant power dispatching method based on model predictive control as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the virtual power plant power dispatching method based on model predictive control as described in any one of claims 1 to 7.

Images