Cloud edge cooperative regulation and control method and device based on large model, equipment and medium

Abstract

The embodiment of the invention discloses a cloud edge cooperative regulation and control method and device based on a large model, equipment and a medium. The method comprises the following steps: receiving side end data uploaded by a collection device, the side end data being data which is collected by the collection device and is related to the side end device; performing photovoltaic prediction and load prediction through a multi-mode large model according to the edge data, the historical data and the weather forecast data to generate a photovoltaic prediction curve and a load prediction curve; according to the photovoltaic prediction curve, the load prediction curve and preset equipment information, a regulation and control strategy is generated through the multi-mode large model, and feasibility verification is carried out on the regulation and control strategy; and if the feasibility verification of the regulation and control strategy is passed, issuing the regulation and control strategy to the edge end equipment through the acquisition equipment so as to regulate and control the edge end equipment. According to the embodiment of the invention, the flexibility and accuracy of the regulation and control strategy can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a cloud-edge collaborative control method, device, equipment and medium based on a large model. Background Art

[0002] In order to accelerate the construction of a new power system with new energy as the main body and realize the maximum utilization of energy and the maximization of economic benefits, it is urgent to build a source-grid-load-storage integration project to improve the absorption capacity of new energy and reduce the operation risk of the power grid through the interaction and coordination of various energy and resources. Park microgrid is an important landing point for the source-grid-load-storage integration project. The reasonable regulation of photovoltaic, energy storage, controllable load and other equipment is of great significance for achieving local energy self-sufficiency, reducing long-distance transmission losses and maximizing economic benefits.

[0003] However, the current construction of park microgrids usually lacks effective control methods or the existing control methods are inflexible, which mainly leads to the following problems: 1. The distributed new energy output load has alternating peaks, resulting in insufficient local absorption capacity, serious reverse overload, and insufficient carrying capacity of low-voltage distribution network lines; 2. Photovoltaic, storage, and charging are operated separately, the photovoltaic abandonment rate is high, and it is impossible to fully utilize the low-valley electricity price and photovoltaic power generation, the rate of return is low, and the cost recovery period is long; 3. The control strategy cannot be close to the actual photovoltaic output and user load data, resulting in a large deviation between the control strategy and the optimal solution. At the same time, when the weather environment changes, the photovoltaic output changes, the equipment ages, and the load power consumption characteristics change dynamically over time, the existing control strategy is prone to failure. Summary of the invention

[0004] The embodiments of the present invention provide a cloud-edge collaborative control method, device, equipment and medium based on a large model, aiming to solve the problem that the control strategy of the existing campus microgrid is not flexible enough and has low accuracy.

[0005] In a first aspect, an embodiment of the present invention provides a cloud-edge collaborative control method based on a large model, which is applied to the cloud, and includes: Receiving edge data uploaded by a collection device, wherein the edge data is data collected by the collection device and related to the edge device; Performing photovoltaic prediction and load prediction through a multimodal large model according to the edge data, historical data and weather forecast data to generate a photovoltaic prediction curve and a load prediction curve; Generate a control strategy through the multi-modal large model according to the photovoltaic prediction curve, the load prediction curve and the preset equipment information, and verify the feasibility of the control strategy; If the feasibility verification of the control strategy is passed, the control strategy is sent to the edge device via the acquisition device to control the edge device.

[0006] In a second aspect, an embodiment of the present invention further provides a cloud-edge collaborative control device based on a large model, which is applied to the cloud, and includes: A first receiving unit is used to receive edge data uploaded by a collection device, wherein the edge data is data collected by the collection device and related to the edge device; A photovoltaic load prediction unit, used to perform photovoltaic prediction and load prediction through a multi-modal large model according to the edge data, historical data and weather forecast data, so as to generate a photovoltaic prediction curve and a load prediction curve; A strategy generating unit, configured to generate a control strategy through the multi-modal large model according to the photovoltaic prediction curve, the load prediction curve and the preset equipment information, and to verify the feasibility of the control strategy; The control unit is used to send the control strategy to the edge device via the acquisition device if the feasibility verification of the control strategy is passed, so as to control the edge device.

[0007] In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the computer program.

[0008] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0009] The embodiment of the present invention provides a cloud-edge collaborative control method, device, equipment and medium based on a large model. The method includes: receiving edge data uploaded by a collection device, wherein the edge data is data related to the edge device collected by the collection device; performing photovoltaic prediction and load prediction through a multimodal large model based on the edge data, historical data and weather forecast data to generate a photovoltaic prediction curve and a load prediction curve; generating a control strategy through the multimodal large model based on the photovoltaic prediction curve, the load prediction curve and preset equipment information, and verifying the feasibility of the control strategy; if the feasibility verification of the control strategy is passed, the control strategy is sent to the edge device via the collection device to control the edge device. The technical solution of the embodiment of the present invention performs photovoltaic prediction and load prediction through a multimodal large model based on the edge data, historical data and weather forecast data uploaded by the acquisition device, generates a photovoltaic prediction curve and a load prediction curve, generates a control strategy through the multimodal large model based on the two generated prediction curves and preset equipment information, and sends the control strategy that has passed the feasibility verification to the edge device to control the edge device. In the entire control process, multiple types of data such as edge data, historical data and weather data are comprehensively considered, and the photovoltaic prediction curve and the load prediction curve are first generated through the multimodal large model, and the control strategy is generated through the multimodal large model based on the two prediction curves, and the control strategy that has passed the feasibility verification is sent to the edge device to control the edge device, which can improve the flexibility and accuracy of the control strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.

[0011] Figure 1 It is an overall schematic diagram of a cloud-edge collaborative control system based on a large model provided by an embodiment of the present invention; Figure 2 A schematic diagram of a flow chart of a cloud-edge collaborative control method based on a large model provided in an embodiment of the present invention; Figure 3 A schematic diagram of a sub-process of a cloud-edge collaborative control method based on a large model provided in an embodiment of the present invention; Figure 4 A schematic diagram of another sub-process of a cloud-edge collaborative control method based on a large model provided in an embodiment of the present invention; Figure 5A schematic diagram of another sub-process of a cloud-edge collaborative control method based on a large model provided in an embodiment of the present invention; Figure 6 A schematic flow chart of a cloud-edge collaborative control method based on a large model provided in another embodiment of the present invention; Figure 7 A schematic diagram of a flow chart of a cloud-edge collaborative control method based on a large model provided in yet another embodiment of the present invention; Figure 8 A schematic block diagram of a cloud-edge collaborative control device based on a large model provided in an embodiment of the present invention; Fig. 9 A schematic block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described feature diagrams, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other feature diagrams, wholes, steps, operations, elements, components and / or collections thereof.

[0014] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0015] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0016] As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0017] See also Figure 1 , Figure 1 It is an overall schematic diagram of a cloud-edge collaborative control system based on a large model provided by an embodiment of the present invention. The cloud-edge collaborative control system based on a large model in an embodiment of the present invention includes a cloud, edge devices and collection devices. The edge devices are concentrators. The cloud includes a control system and a collection system. The edge devices include metering and control devices and user devices. The user devices include photovoltaic devices, area energy storage devices and controllable load devices. Figure 1 As shown, there is remote communication between the cloud and the acquisition device, and the acquisition system in the cloud stores the full amount of the edge data related to the edge device collected by the acquisition device; there is local communication between the acquisition device and the edge device. It can be understood that the acquisition device is used to collect edge data related to the edge device, and can also receive the control strategy issued by the control system, and transmit the control strategy to the edge device to control the edge device. Specifically, the charging, discharging and power generation of the edge device are regulated. It should be noted that in this embodiment, a corresponding software program is configured in the cloud to implement the functions of the control system, which can improve the flexibility and accuracy of the control strategy.

[0018] See also Figure 2 , Figure 2 Schematic diagram of a cloud-edge collaborative control method based on a large model provided by an embodiment of the present invention. Figure 2 As shown, the method includes the following steps S110-S140.

[0019] S110. Receive edge data uploaded by a collection device, wherein the edge data is data collected by the collection device and related to the edge device.

[0020] In an embodiment of the present invention, the collection system in the cloud receives the edge data uploaded by the collection device, wherein the edge data is data related to the edge device collected by the collection device.

[0021] S120. Perform photovoltaic prediction and load prediction using a multimodal large model according to the edge data, historical data, and weather forecast data to generate a photovoltaic prediction curve and a load prediction curve.

[0022] In an embodiment of the present invention, the control system in the cloud performs photovoltaic forecasting and load forecasting through a multimodal large model based on the edge data, historical data and weather forecast data to generate photovoltaic forecast curves and load forecast curves, wherein the edge data includes photovoltaic ledger information, user industry information and date attribute information, and the historical data includes historical photovoltaic output curves, historical load curves and historical weather data. It should be noted that in this embodiment, the historical data is stored in the historical edge database in the acquisition system, and the weather forecast data is data obtained by externally calling the weather forecast, specifically including radiation, rainfall, temperature, humidity, cloud cover and other data information.

[0023] See also Figure 3 In one embodiment, for example, in an embodiment of the present invention, the step S120 includes the following steps S121-S122.

[0024] S121, performing photovoltaic prediction through the multimodal large model according to the photovoltaic ledger information, the historical photovoltaic output curve, the historical weather data and the weather forecast data to generate the photovoltaic prediction curve; S122. Perform load forecasting using the multimodal large model according to the user industry information, the date attribute information, the historical load curve, and the weather forecast data to generate the load forecast curve.

[0025] In an embodiment of the present invention, the control system in the cloud includes a photovoltaic prediction module and a load prediction module, wherein the photovoltaic prediction curve generated in step S121 is generated in the photovoltaic prediction module. Figure 4 As shown, step S121 specifically includes steps S1211-S1215: S1211, sequentially performing data cleaning and feature extraction on the historical photovoltaic output curve to obtain historical photovoltaic output features; S1212, vectorize and encode the photovoltaic ledger information to generate a photovoltaic ledger encoding vector; S1213, vectorize and encode the weather forecast data to generate a weather forecast encoding vector; S1214, generate a similar day output curve according to the historical photovoltaic output curve, the historical weather data and the weather forecast data, and vector encode the similar day output curve to obtain a day output curve encoding vector; S1215, modify the day output curve encoding vector through the multimodal large model according to the historical photovoltaic output characteristics, the photovoltaic ledger encoding vector and the weather forecast encoding vector to generate the photovoltaic prediction curve. It should be noted that in this embodiment, the multimodal large model in the photovoltaic prediction module is based on the Transform network model, and the input is the features extracted from different data sources, for example, the historical photovoltaic output characteristics extracted from the historical photovoltaic output curve, and the photovoltaic ledger encoding vector generated according to the photovoltaic ledger information. It should also be noted that in this embodiment, the multimodal large model in the photovoltaic prediction module can learn information from multiple data sources by performing feature-level data fusion through the attention mechanism.

[0026] Further, step S1214 specifically includes: performing a similarity search on the weather forecast data in the weather database where the historical weather data is located to obtain the weather daily curve data of the corresponding day, and performing a similarity search in the historical edge database where the historical photovoltaic output curve is located based on the weather daily curve data of the corresponding day to generate the output curve for the similar day; or correcting the weather forecast data based on the real environment observation data to obtain the corrected weather forecast data, and performing a similarity search on the corrected weather forecast data in the weather database where the historical weather data is located to obtain the weather daily curve data of the corresponding day, and performing the step of performing a similarity search in the historical edge database where the historical photovoltaic output curve is located based on the weather daily curve data of the corresponding day to generate the output curve for the similar day. It should be noted that, in this embodiment, the control system also includes a weather forecast data correction module, and the weather forecast data correction module is used to correct the weather forecast data. In this embodiment, since the radiance information in the weather forecast data plays a decisive role in photovoltaic prediction, it is possible to add an environmental monitoring device with cloud cover monitoring, use the environmental monitoring device to monitor the real environment observation data, and correct the weather forecast data through the weather forecast data correction module according to the real environment observation data to obtain the corrected weather forecast data. It should also be noted that in this embodiment, the weather forecast data correction module first formats the input real environment observation data and the weather forecast data to convert the data into the same format and unit, and then performs feature extraction, normalizes the extracted features, and adjusts the feature scale so as to more effectively learn the information in the features, and finally corrects the weather forecast data to obtain the corrected weather forecast data.

[0027] Furthermore, if Figure 5As shown, step S122 specifically includes steps S1221-S1225: S1221, performing data cleaning and feature extraction on the historical load curve in sequence to obtain historical load characteristics; S1222, performing vectorized encoding on the user industry information to obtain a user industry encoding vector; S1223, performing vectorized encoding on the weather forecast data to generate a weather forecast encoding vector; S1224, performing similarity search in a constructed user energy consumption portrait database based on the user industry information, the date attribute information, the historical load curve and the weather forecast data to obtain a similar daily load curve, and performing vectorized encoding on the similar daily load curve to generate a daily load curve encoding vector; S1225, modifying the daily load curve encoding vector through the multimodal large model based on the historical load characteristics, the user industry encoding vector and the weather forecast encoding vector to generate the load prediction curve. It should be noted that, in this embodiment, the user energy consumption portrait database is constructed based on the pre-collected user industry information, different weather energy consumption curves, load curves under different date attributes, large load energy consumption curves, etc. It can be understood that the load curves under different date attributes include workday load curves, holiday load curves, major event load curves, etc., that is, it is necessary to comprehensively characterize the user energy consumption characteristics. The user industry information, the date attribute information, the historical load curve and the weather forecast data are input into the constructed user energy consumption portrait database for similarity retrieval to search for the most similar daily load curve to obtain a similar daily load curve. On the basis of the similar daily load curve, the curve is refined and corrected using the reasoning ability of the multimodal large model to generate the load prediction curve. It should also be noted that, in this embodiment, the load prediction curve generated in step S122 is generated in the load prediction module. The multimodal large model in the load prediction module and the multimodal large model of the photovoltaic prediction module are the same model, both of which are based on Transform as the basic network model, and the input is the features extracted from different data sources, such as historical load features and user industry coding vectors. Understandably, the multimodal large model maps the features of different data sources into the same feature space, allowing the large model to integrate information from different modalities in this space, and then captures the correlation between different modalities through the attention mechanism, allowing the large model to more accurately understand the complex relationship between different data sources, thereby obtaining a more refined load forecast curve.

[0028] Furthermore, in this embodiment, the historical photovoltaic output curve is considered when performing photovoltaic prediction, and the historical load curve is considered when performing load prediction, because historical data can be used to calculate equipment attenuation, perform short-term prediction, and model training. It should be noted that in this embodiment, when performing photovoltaic prediction, the model obtained after training the multimodal large model is the LSTM-ATT model.

[0029] S130, generating a control strategy through the multimodal large model according to the photovoltaic prediction curve, the load prediction curve and the preset equipment information, and verifying the feasibility of the control strategy.

[0030] In an embodiment of the present invention, the preset device information includes a list of adjustable devices and online status, a list of adjustable quantities, device constraints, and a device operation mode, wherein the list of adjustable devices and the online status of the device that is offline for a long time are not adjustable, and the device that is offline for a short time is attempted to be regulated; the list of adjustable quantities includes adjustable quantities such as photovoltaic active power, photovoltaic reactive power, energy storage charging, and energy storage discharging; the device constraints include constraints such as photovoltaic adjustable upper and lower limits, energy storage charging and discharging upper and lower limits; the device operation mode includes a safe operation mode, a peak-valley arbitrage mode, etc. It should be noted that in this embodiment, the feasibility verification of the control strategy is specifically offline verification. During the verification process, it is checked whether the device constraints are met. If the device constraints are met, it indicates that the feasibility verification of the control strategy has passed. On the contrary, if the device constraints are not met, it indicates that the feasibility verification of the control strategy has failed. It should also be noted that in this embodiment, the loss function will be continuously optimized during the generation of the control strategy.

[0031] S140: If the feasibility verification of the control strategy is passed, the control strategy is sent to the edge device via the acquisition device to control the edge device.

[0032] In an embodiment of the present invention, if the device constraint condition is met, indicating that the control strategy meets the control requirements, the control strategy is sent to the edge device via the acquisition device to control the edge device.

[0033] Figure 6 A flowchart of a cloud-edge collaborative control method based on a large model is provided for another embodiment of the present invention, such as Figure 6 As shown, in this embodiment, the method includes steps S110-S150. That is, in this embodiment, after step S140 of the above embodiment, the method further includes step S150.

[0034] S150. If the feasibility verification of the control strategy fails, the control strategy is modified through the strategy optimization algorithm model to obtain a modified control strategy, and the modified control strategy is sent to the edge device through the acquisition device to control the edge device.

[0035] In an embodiment of the present invention, if the device constraint condition is not met, indicating that the control strategy does not meet the control requirements, the control strategy is corrected by a strategy optimization algorithm model to obtain a corrected control strategy, wherein the strategy optimization algorithm model is a differential integrated moving average autoregressive model. The corrected control strategy is obtained, and the corrected control strategy is sent to the edge device. It should be noted that in this embodiment, step S140 and step S150 are completed in the strategy generation module in the control system.

[0036] Figure 7 A flowchart of a cloud-edge collaborative control method based on a large model is provided for another embodiment of the present invention, such as Figure 7 As shown, in this embodiment, the method includes steps S110-S190. That is, in this embodiment, after step S150 of the above embodiment, the method further includes steps S160-S190.

[0037] S160, receiving the real photovoltaic output curve and the real load curve uploaded by the acquisition device; S170, performing consistency determination on the actual photovoltaic output curve and the photovoltaic prediction curve to obtain a photovoltaic determination result; S180, performing consistency determination on the real load curve and the load forecast curve to obtain a load determination result; S190. If the accuracy of the photovoltaic discrimination result / the load discrimination result is lower than a preset threshold, the strategy self-evolution mode is entered to train the multimodal large model with recent historical data to obtain an updated multimodal large model, and the performance of the updated multimodal large model is evaluated. The updated multimodal large model that passes the performance evaluation is deployed on the cloud.

[0038] In an embodiment of the present invention, after receiving the real photovoltaic output curve and the real load curve uploaded by the acquisition device, the control system will respectively perform consistency judgment on the real photovoltaic output curve and the photovoltaic prediction curve, the real load curve and the load prediction curve, and obtain the photovoltaic judgment result and the load judgment result. If the accuracy of the photovoltaic judgment result / the load judgment result is lower than the preset threshold, the strategy self-evolution mode is entered to train the multimodal large model with recent historical data to obtain an updated multimodal large model, and the performance of the updated multimodal large model is evaluated, and the updated multimodal large model is deployed on the cloud after the performance evaluation is qualified to keep the model continuously effective. Understandably, if the accuracy of the photovoltaic judgment result / the load judgment result is not lower than the preset threshold, the edge device can be continuously monitored. It should be noted that in this embodiment, for the updated multimodal large model that fails the performance evaluation, the alarm information is fed back to the control system. The updated multimodal large model is mainly evaluated from the two aspects of power balance and economy. It should also be noted that, in this embodiment, steps S160-S190 are completed in the strategy self-evolution module in the control system. The reason for setting the strategy self-evolution module is that the deployed multi-modal large model will easily fail due to factors such as seasonal changes, equipment aging, photovoltaic pollution, and changes in user power consumption patterns. The large model in the strategy self-evolution module can be a particle swarm algorithm, linear regression, and algorithms based on graph convolutional neural networks. The multi-modal large model is used to build a multi-state operation control strategy, and the control strategy is optimized in real time.

[0039] To sum up, in this embodiment, in the entire control process, multiple types of data such as edge data, historical data and weather data are comprehensively considered, and the photovoltaic prediction curve and the load prediction curve are first generated through the multimodal large model, and the control strategy is generated based on the two prediction curves through the multimodal large model, and the control strategy that passes the feasibility verification is sent to the edge device to control the edge device, which can improve the flexibility and accuracy of the control strategy; when the feasibility verification of the control strategy fails, the control strategy is corrected by the strategy optimization algorithm model to obtain a corrected control strategy, and the corrected control strategy is sent to the edge device to control the edge device, which further improves the accuracy of the control strategy; by performing consistency judgment on the actual photovoltaic output curve and the photovoltaic prediction curve, the actual load curve and the load prediction curve, the strategy self-evolution mode is entered, and the adaptability of the control strategy is improved to adapt to the complex and changeable power grid environment. Therefore, the cloud-edge collaborative control method based on a large model provided in this embodiment solves the problem that the control strategy of the existing park microgrid is not flexible enough and has low accuracy. Not only that, it also solves the problems of insufficient local absorption capacity, serious reverse overload, and insufficient carrying capacity of low-voltage distribution network lines.

[0040] Figure 8 is a schematic block diagram of a cloud-edge collaborative control device 200 based on a large model provided by an embodiment of the present invention. Figure 8 As shown, corresponding to the above cloud-edge collaborative control method based on a large model, the present invention also provides a cloud-edge collaborative control device 200 based on a large model. The cloud-edge collaborative control device 200 based on a large model includes a unit for executing the above cloud-edge collaborative control method based on a large model, and the device can be configured in a computer device on the cloud. Specifically, please refer to Figure 8 The cloud-edge collaborative control device 200 based on a large model includes a first receiving unit 201, a photovoltaic load prediction unit 202, a strategy generation unit 203 and a control unit 204.

[0041] Among them, the first receiving unit 201 is used to receive edge data uploaded by the collection device, wherein the edge data is data related to the edge device collected by the collection device; the photovoltaic load prediction unit 202 is used to perform photovoltaic prediction and load prediction through a multimodal large model according to the edge data, historical data and weather forecast data, so as to generate a photovoltaic prediction curve and a load prediction curve; the strategy generation unit 203 is used to generate a control strategy through the multimodal large model according to the photovoltaic prediction curve, the load prediction curve and preset equipment information, and to verify the feasibility of the control strategy; the control unit 204 is used to send the control strategy to the edge device via the collection device if the feasibility verification of the control strategy is passed, so as to control the edge device.

[0042] In some embodiments, such as this embodiment, the photovoltaic load prediction unit 202 includes a photovoltaic prediction unit and a load prediction unit.

[0043] Among them, the photovoltaic prediction unit is used to perform photovoltaic prediction through the multimodal large model according to the photovoltaic ledger information, the historical photovoltaic output curve, the historical weather data and the weather forecast data to generate the photovoltaic prediction curve; the load prediction unit is used to perform load prediction through the multimodal large model according to the user industry information, the date attribute information, the historical load curve and the weather forecast data to generate the load prediction curve.

[0044] In some embodiments, such as the present embodiment, the photovoltaic prediction unit includes a first cleaning extraction unit, a first vector encoding unit, a second vector encoding unit, a first generation encoding unit, and a first correction generation unit.

[0045] Among them, the first cleaning and extraction unit is used to perform data cleaning and feature extraction on the historical photovoltaic output curve in sequence to obtain historical photovoltaic output characteristics; the first vector encoding unit is used to vectorize the photovoltaic ledger information to generate a photovoltaic ledger encoding vector; the second vector encoding unit is used to vectorize the weather forecast data to generate a weather forecast encoding vector; the first generation encoding unit is used to generate a similar daily output curve according to the historical photovoltaic output curve, the historical weather data and the weather forecast data, and vector encode the similar daily output curve to obtain a daily output curve encoding vector; the first correction generation unit is used to correct the daily output curve encoding vector according to the historical photovoltaic output characteristics, the photovoltaic ledger encoding vector and the weather forecast encoding vector through the multimodal large model to generate the photovoltaic prediction curve.

[0046] In some embodiments, such as the present embodiment, the first generation coding unit includes a first retrieval generation unit or a modified retrieval unit.

[0047] Among them, the first retrieval generation unit is used to perform a similarity search on the weather forecast data in the weather database where the historical weather data is located to obtain the weather daily curve data of the corresponding day, and perform a similarity search in the historical edge database where the historical photovoltaic output curve is located based on the weather daily curve data of the corresponding day to generate the output curve for the similar day; or the correction retrieval unit is used to correct the weather forecast data according to the real environment observation data to obtain the corrected weather forecast data, and perform a similarity search on the corrected weather forecast data in the weather database where the historical weather data is located to obtain the weather daily curve data of the corresponding day, and execute the step of performing a similarity search in the historical edge database where the historical photovoltaic output curve is located based on the weather daily curve data of the corresponding day to generate the output curve for the similar day.

[0048] In some embodiments, such as the present embodiment, the load prediction unit includes a second cleaning and extraction unit, a third vector encoding unit, a fourth vector encoding unit, a second retrieval generation unit, and a second correction generation unit.

[0049] Among them, the second cleaning and extraction unit is used to perform data cleaning and feature extraction on the historical load curve in sequence to obtain historical load characteristics; the third vector encoding unit is used to vectorize the user industry information to obtain a user industry coding vector; the fourth vector encoding unit is used to vectorize the weather forecast data to generate a weather forecast coding vector; the second retrieval generation unit is used to perform similarity retrieval in a constructed user energy consumption portrait database according to the user industry information, the date attribute information, the historical load curve and the weather forecast data to obtain a similar daily load curve, and vectorize the similar daily load curve to generate a daily load curve coding vector; the second correction generation unit is used to correct the daily load curve coding vector through the multimodal large model according to the historical load characteristics, the user industry coding vector and the weather forecast coding vector to generate the load forecast curve.

[0050] In certain embodiments, such as the present embodiment, the cloud-edge collaborative control device 200 based on a large model also includes a correction sending unit, a second receiving unit, a first judgment unit, a second judgment unit, and a training evaluation deployment unit.

[0051] Among them, the correction and sending unit is used to correct the control strategy through the strategy optimization algorithm model to obtain a corrected control strategy if the feasibility verification of the control strategy fails, and send the corrected control strategy to the edge device through the acquisition device to control the edge device; the second receiving unit is used to receive the real photovoltaic output curve and the real load curve uploaded by the acquisition device; the first discrimination unit is used to perform consistency judgment on the real photovoltaic output curve and the photovoltaic prediction curve to obtain a photovoltaic discrimination result; the second discrimination unit is used to perform consistency judgment on the real load curve and the load prediction curve to obtain a load discrimination result; the training, evaluation and deployment unit is used to enter the strategy self-evolution mode if the accuracy of the photovoltaic discrimination result / the load discrimination result is lower than a preset threshold, so as to use recent historical data to train the multimodal large model to obtain an updated multimodal large model, and perform performance evaluation on the updated multimodal large model, and deploy the updated multimodal large model after the performance evaluation is qualified on the cloud.

[0052] The specific implementation method of the cloud-edge collaborative control device 200 based on a large model in an embodiment of the present invention corresponds to the above-mentioned cloud-edge collaborative control method based on a large model, and will not be repeated here.

[0053] The cloud-edge collaborative control device based on the large model can be implemented in the form of a computer program. The computer program can be used in Fig. 9 Runs on the computer device shown.

[0054] See also Fig. 9 , Fig. 9 300 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device 300 is a device having a cloud-edge collaborative control function based on a large model. See also Fig. 9 The computer device 300 includes a processor 302 , a memory and a network interface 305 connected via a system bus 301 , wherein the memory may include a storage medium 303 and an internal memory 304 .

[0055] The storage medium 303 may store an operating system 3031 and a computer program 3032. When the computer program 3032 is executed, the processor 302 may execute a cloud-edge collaborative control method based on a large model.

[0056] The processor 302 is used to provide computing and control capabilities to support the operation of the entire computer device 300 .

[0057] The internal memory 304 provides an environment for the operation of the computer program 3032 in the storage medium 303. When the computer program 3032 is executed by the processor 302, the processor 302 can execute a cloud-edge collaborative control method based on a large model.

[0058] The network interface 305 is used to communicate with other devices over the network. Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 300 to which the solution of the present application is applied. The specific computer device 300 may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0059] Among them, the processor 302 is used to run the computer program 3032 stored in the memory to implement any embodiment of the above-mentioned cloud-edge collaborative control method based on the big model.

[0060] It should be understood that in the embodiment of the present application, the processor 302 may be a central processing unit (CPU), and the processor 302 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0061] It is understood by those skilled in the art that all or part of the processes in the method for implementing the above embodiment can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiment of the above method.

[0062] Therefore, the present invention also provides a storage medium. The storage medium may be a computer-readable storage medium. The storage medium stores a computer program. When the computer program is executed by a processor, the processor executes any embodiment of the cloud-edge collaborative control method based on a large model.

[0063] The storage medium may be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk, etc., which are computer-readable storage media that can store program codes.

[0064] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0065] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of each unit is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some feature diagrams can be ignored or not executed.

[0066] The steps in the method of the embodiment of the present invention can be adjusted in order, combined and deleted according to actual needs. The units in the device of the embodiment of the present invention can be combined, divided and deleted according to actual needs. In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0067] 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 storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, terminal, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention.

[0068] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0069] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

[0070] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A cloud-edge collaborative control method based on a large model, applied to the cloud, characterized in that: include: Receiving edge data uploaded by a collection device, wherein the edge data is data collected by the collection device and related to the edge device; Performing photovoltaic prediction and load prediction through a multimodal large model according to the edge data, historical data and weather forecast data to generate a photovoltaic prediction curve and a load prediction curve; Generate a control strategy through the multi-modal large model according to the photovoltaic prediction curve, the load prediction curve and the preset equipment information, and verify the feasibility of the control strategy; If the feasibility verification of the control strategy is passed, the control strategy is sent to the edge device via the acquisition device to control the edge device.

2. The method according to claim 1, characterized in that The edge data includes photovoltaic ledger information, user industry information and date attribute information, the historical data includes historical photovoltaic output curve, historical load curve and historical weather data, and the photovoltaic prediction and load prediction are performed through a multimodal large model based on the edge data, historical data and weather forecast data to generate photovoltaic prediction curves and load prediction curves, including: Performing photovoltaic prediction through the multimodal large model according to the photovoltaic ledger information, the historical photovoltaic output curve, the historical weather data and the weather forecast data to generate the photovoltaic prediction curve; The load forecasting curve is generated by performing load forecasting through the multimodal large model according to the user industry information, the date attribute information, the historical load curve and the weather forecast data.

3. The method according to claim 2, characterized in that The photovoltaic prediction is performed by the multimodal large model according to the photovoltaic ledger information, the historical photovoltaic output curve, the historical weather data and the weather forecast data to generate the photovoltaic prediction curve, including: Performing data cleaning and feature extraction on the historical photovoltaic output curve in turn to obtain historical photovoltaic output features; Vectorizing and encoding the photovoltaic ledger information to generate a photovoltaic ledger encoding vector; Performing vectorized encoding on the weather forecast data to generate a weather forecast encoding vector; Generate a similar day output curve according to the historical photovoltaic output curve, the historical weather data and the weather forecast data, and perform vector encoding on the similar day output curve to obtain a day output curve encoding vector; The daily output curve coding vector is corrected through the multimodal large model according to the historical photovoltaic output characteristics, the photovoltaic ledger coding vector, and the weather forecast coding vector to generate the photovoltaic prediction curve.

4. The method according to claim 3, characterized in that The generating a similar day output curve according to the historical photovoltaic output curve, the historical weather data and the weather forecast data comprises: Performing a similarity search on the weather forecast data in the weather database where the historical weather data is located to obtain weather daily curve data of the corresponding day, and performing a similarity search on the historical edge database where the historical photovoltaic output curve is located based on the weather daily curve data of the corresponding day to generate the similar day output curve; or The weather forecast data is corrected according to the real environment observation data to obtain the corrected weather forecast data, and a similarity search is performed on the corrected weather forecast data in the weather database where the historical weather data is located to obtain the weather daily curve data of the corresponding day, and the step of performing a similarity search in the historical edge database where the historical photovoltaic output curve is located according to the weather daily curve data of the corresponding day to generate the similar day output curve is performed.

5. The method according to claim 2, characterized in that: The performing load forecasting through the multimodal large model according to the user industry information, the date attribute information, the historical load curve and the weather forecast data to generate the load forecast curve includes: Performing data cleaning and feature extraction on the historical load curve in sequence to obtain historical load features; Vectorizing and encoding the user industry information to obtain a user industry encoding vector; Performing vectorized encoding on the weather forecast data to generate a weather forecast encoding vector; Perform similarity search in a constructed user energy consumption portrait database according to the user industry information, the date attribute information, the historical load curve and the weather forecast data to obtain a similar daily load curve, and vectorize and encode the similar daily load curve to generate a daily load curve encoding vector; The daily load curve coding vector is modified through the multimodal large model according to the historical load characteristics, the user industry coding vector and the weather forecast coding vector to generate the load forecast curve.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: If the feasibility verification of the control strategy fails, the control strategy is modified through the strategy optimization algorithm model to obtain a modified control strategy, and the modified control strategy is sent to the edge device through the acquisition device to control the edge device.

7. The method according to claim 6, characterized in that The method further comprises: Receiving the real photovoltaic output curve and the real load curve uploaded by the acquisition device; Performing consistency judgment on the actual photovoltaic output curve and the photovoltaic prediction curve to obtain a photovoltaic judgment result; Performing consistency judgment on the real load curve and the load forecast curve to obtain a load judgment result; If the accuracy of the photovoltaic discrimination result / the load discrimination result is lower than a preset threshold, the strategy self-evolution mode is entered to train the multimodal large model with recent historical data to obtain an updated multimodal large model, and the performance of the updated multimodal large model is evaluated. The updated multimodal large model that passes the performance evaluation is deployed on the cloud.

8. A cloud-edge collaborative control device based on a large model, applied to the cloud, characterized in that: include: A first receiving unit is used to receive edge data uploaded by a collection device, wherein the edge data is data collected by the collection device and related to the edge device; A photovoltaic load prediction unit, used to perform photovoltaic prediction and load prediction through a multi-modal large model according to the edge data, historical data and weather forecast data, so as to generate a photovoltaic prediction curve and a load prediction curve; A strategy generating unit, configured to generate a control strategy through the multi-modal large model according to the photovoltaic prediction curve, the load prediction curve and the preset equipment information, and to verify the feasibility of the control strategy; The control unit is used to send the control strategy to the edge device via the acquisition device if the feasibility verification of the control strategy is passed, so as to control the edge device.

9. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images