Distributed photovoltaic cluster regulation and control system and method and storage medium

Abstract

The invention relates to the technical field of photovoltaic power generation, and discloses a distributed photovoltaic cluster regulation and control system and method and a storage medium, and the system comprises a feature processing module which is used for generating power supply features according to historical power supply requirements, and obtaining a feature generation rule through a feature extraction technology; the strategy processing module is used for generating a power generation strategy corresponding to the power supply characteristics and obtaining a strategy generation rule by using a characteristic extraction technology; the model construction module is used for constructing an initial cluster regulation and control model and training the initial cluster regulation and control model to obtain a cluster regulation and control model; the model operation module is used for generating a real-time power generation strategy by utilizing a cluster regulation and control model under the guidance of a feature generation rule and a strategy generation rule according to a real-time power supply requirement; and the cluster regulation and control module is used for regulating and controlling the cluster based on the real-time power generation strategy. According to the invention, autonomous real-time power generation strategy generation is realized through the cluster regulation and control model, and optimization of distributed photovoltaic cluster regulation and control is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to a distributed photovoltaic cluster control system, method and storage medium. Background Art

[0002] The power grid company plays the role of energy supply in the energy-saving service providers, load aggregators, power sales agents and carbon trading service providers. Through the inherent coupling relationship, it adopts appropriate methods to provide energy that meets the needs of the corresponding energy-saving service providers, load aggregators, power sales agents and carbon trading service providers. Therefore, it is necessary to explore the inherent coupling relationship based on the role positioning of the power grid company.

[0003] Distributed photovoltaic power generation can provide clean, recyclable and more affordable electricity. Therefore, it is feasible to use photovoltaic power generation to meet the needs of energy-saving service providers, load aggregators, electricity sales agents and carbon trading service providers.

[0004] With the development of machine learning technology, it is necessary to establish a model to adjust the power generation strategy of power grid companies under the "four-in-one" business model to optimize the cluster regulation of distributed photovoltaics.

[0005] CN202411088033.6 discloses a method, system, electronic device and storage medium for dynamic balance control of supply and demand in the power market, including: collecting real-time supply and demand data information in the power market, including the power generation of different types of power generation sides participating in the power market and the power demand of different types of power consumption sides; arranging the collected multiple continuous real-time supply and demand data information in time sequence to obtain the power market supply and demand characteristic matrix; inputting the collected power market supply and demand characteristic matrix into a pre-built power market supply and demand analysis model to obtain the power market supply and demand evaluation characteristic vector, which is used to reflect the supply and demand balance of the power market, and the power market supply and demand evaluation characteristic vector includes a power supply evaluation index and a power consumption evaluation index; obtaining the power market power supply stability influencing factor and the power market power consumption stability influencing factor. However, this invention only regulates from the perspective of supply and demand, and its applicability is poor. Summary of the invention

[0006] The purpose of the present invention is to overcome the problem that the prior art urgently needs to optimize the cluster control of distributed photovoltaics, and to provide a cluster control system, method and storage medium for distributed photovoltaics. This solution provides the necessary technical means for cluster control and realizes the optimization of distributed photovoltaic cluster control.

[0007] In order to achieve the above-mentioned object, the present invention provides a distributed photovoltaic cluster control system on the one hand, which includes a feature processing module, a strategy processing module, a model building module, a model operation module and a cluster control module;

[0008] The feature processing module is used to generate power supply features according to historical power supply requirements using natural language processing technology, and obtain feature generation rules for the power supply features using feature extraction technology;

[0009] The strategy processing module is used to generate a power generation strategy corresponding to the power supply characteristics, and obtain a strategy generation rule for the power generation strategy using a feature extraction technology;

[0010] The model building module is used to build an initial cluster control model using machine learning technology, and train the initial cluster control model using the feature generation rules, strategy generation rules and historical power supply requirements to obtain a cluster control model;

[0011] The model operation module is used to generate a real-time power generation strategy using the cluster control model under the guidance of the feature generation rule and the strategy generation rule according to the real-time power supply requirements;

[0012] The cluster control module is used to control the cluster based on the real-time power generation strategy.

[0013] Preferably, the power supply requirements include quality requirements, metering requirements, response requirements, price requirements, service requirements and carbon trading requirements.

[0014] Preferably, the specific operation of generating power supply characteristics according to historical power supply requirements using natural language processing technology includes: parsing the historical power supply requirements using natural language processing technology, capturing the corresponding feature thresholds in the historical power supply requirements, packaging the feature thresholds, and obtaining power supply characteristics.

[0015] Preferably, the specific operation of obtaining the feature generation rule by using the feature extraction technology for the power supply feature includes: extracting the feature of the feature threshold by using the feature extraction technology, and defining the feature of the extracted feature threshold as the feature generation rule;

[0016] The specific operation of extracting the feature of the feature threshold by using the feature extraction technology includes:

[0017] Get the intra-class scatter matrix S ω :

[0018]

[0019] Where, X i represents the sample set of the i-th type of subject; x represents a vector containing multiple feature thresholds; E represents the sample of energy-saving service providers; L represents the sample of load aggregators; S represents the sample of power sales agents; C represents the sample of carbon trading service providers; μ i Represents the mean vector of the i-th class subject calculated based on the feature threshold;

[0020] Get the inter-class scatter matrix S b :

[0021]

[0022] Where n i Indicates the number of samples corresponding to the i-th category subject;

[0023] Solve equation S w -1 S b υ=λυ, where λ represents the eigenvalue; υ represents the eigenvector corresponding to the eigenvalue.

[0024] Preferably, the specific operation of generating a power generation strategy corresponding to the power supply feature includes: sending the power supply feature to the cluster, recording the corresponding action taken by the cluster after receiving the power supply feature, packaging the corresponding action taken, and obtaining the power generation strategy;

[0025] The cluster includes a plurality of distributed photovoltaic power generation groups.

[0026] Preferably, the specific operation of obtaining the strategy generation rule by using the feature extraction technology for the power generation strategy includes: extracting the features of the action by using the feature extraction technology, and defining the extracted features of the action as the strategy generation rule;

[0027] The specific operations of extracting the features of the action using feature extraction technology include:

[0028] The action data of the power generation strategy is represented as a high-dimensional vector set X = {x1, x2, ..., x n}, where x n Indicates the specific action status of the cluster after receiving the power supply characteristics;

[0029] Calculate the mean vector of the action data of the power generation strategy

[0030] Calculate the covariance matrix

[0031] Solve the equation Cω=ρω, where ρ represents the eigenvalue of the action data; ω represents the eigenvector corresponding to the eigenvalue of the action data.

[0032] Preferably, the initial cluster control model is constructed using machine learning technology, and the initial cluster control model is trained using the feature generation rules, strategy generation rules and historical power supply requirements. The specific operations of obtaining the cluster control model include:

[0033] Storing the feature generation rule and the strategy generation rule in an initial cluster control model, and inputting historical power supply requirements into the initial cluster control model;

[0034] Based on historical power supply requirements, under the guidance of the feature generation rules, the initial cluster control model is used to generate the power supply features to be verified;

[0035] Based on the power supply characteristics to be verified, under the guidance of the strategy generation rules, using the initial cluster control model to generate the power generation strategy to be verified;

[0036] Respectively verifying the power supply feature to be verified and the power generation strategy to be verified using the historical power supply feature and the historical power generation strategy corresponding to the historical power supply requirements;

[0037] Repeat the operations of generating the power supply characteristics to be verified to verifying the power supply characteristics to be verified and the power generation strategy to be verified until the generated power supply characteristics to be verified and the power generation strategy to be verified meet the preset verification indicators, thereby obtaining the cluster control model.

[0038] Preferably, according to the real-time power supply requirement, under the guidance of the feature generation rule and the strategy generation rule, the specific operation of generating the real-time power generation strategy using the cluster control model includes:

[0039] Based on the real-time power supply requirements, under the guidance of the strategy generation rules, a cluster control model is used to generate real-time power supply characteristics;

[0040] Based on the real-time power supply characteristics and under the guidance of the strategy generation rules, a cluster control model is used to generate a real-time power generation strategy.

[0041] A second aspect of the present invention provides a distributed photovoltaic cluster control method, the method comprising the following steps:

[0042] Using natural language processing technology, power supply features are generated according to historical power supply requirements, and feature extraction technology is used to obtain feature generation rules for power supply features;

[0043] Generate a power generation strategy corresponding to the power supply characteristics, and obtain a strategy generation rule for the power generation strategy using a feature extraction technique;

[0044] An initial cluster control model is constructed by using machine learning technology, and the initial cluster control model is trained by using the feature generation rule, strategy generation rule and historical power supply requirements to obtain a cluster control model;

[0045] According to the real-time power supply requirements, under the guidance of the feature generation rules and the strategy generation rules, the cluster control model is used to generate a real-time power generation strategy;

[0046] The cluster is regulated based on the real-time power generation strategy.

[0047] A third aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0048] The present invention provides a data basis for generating power generation strategies through power supply characteristics, and obtains feature generation rules and strategy generation rules by extracting the characteristics of the corresponding generation process, thereby providing a technical basis for building a cluster control model, and realizing the mining of the inherent coupling relationship between energy-saving service providers, load aggregators, power sales agents and carbon trading service providers based on the role positioning of the power grid company, and realizing autonomous real-time power generation strategy generation through the cluster control model, providing the necessary technical means for cluster control, and realizing the cluster control of optimized distributed photovoltaics. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A structural diagram of a distributed photovoltaic cluster control system provided in Example 1 of the present invention;

[0050] Figure 2 This is a flow chart of a distributed photovoltaic cluster control method provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0051] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation described herein is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0052] The endpoints and any values ​​of the ranges disclosed in this article are not limited to the precise ranges or values, and these ranges or values ​​should be understood to include values ​​close to these ranges or values. For numerical ranges, the endpoint values ​​of each range, the endpoint values ​​of each range and the individual point values, and the individual point values ​​can be combined with each other to obtain one or more new numerical ranges, which should be considered as specifically disclosed in this article.

[0053] Example 1

[0054] like Figure 1 The distributed photovoltaic cluster control system shown includes a feature processing module, a strategy processing module, a model building module, a model operation module and a cluster control module;

[0055] The feature processing module is used to generate power supply features according to historical power supply requirements using natural language processing technology, and obtain feature generation rules for the power supply features using feature extraction technology;

[0056] The strategy processing module is used to generate a power generation strategy corresponding to the power supply characteristics, and obtain a strategy generation rule for the power generation strategy using a feature extraction technology;

[0057] The model building module is used to build an initial cluster control model using machine learning technology, and train the initial cluster control model using the feature generation rules, strategy generation rules and historical power supply requirements to obtain a cluster control model;

[0058] The model operation module is used to generate a real-time power generation strategy using the cluster control model under the guidance of the feature generation rule and the strategy generation rule according to the real-time power supply requirement (the cluster control model is communicated and connected with the system generating the real-time power supply requirement to obtain the real-time power supply requirement);

[0059] The cluster control module is used to control the cluster based on the real-time power generation strategy and execute corresponding actions.

[0060] In a specific implementation manner, the power supply requirements include quality requirements, metering requirements, response requirements, price requirements, service requirements and carbon trading requirements.

[0061] The historical power supply requirements in the present invention are collected using existing big data technologies.

[0062] The quality requirements, metering requirements, response requirements, price requirements, service requirements and carbon trading requirements of this embodiment are obtained based on the actual operational needs of energy-saving service providers, load aggregators, power sales agents and carbon trading service providers. For example, energy-saving service providers need stable electricity with high power quality to improve energy-saving effects, and require accurate electricity metering capabilities to verify energy-saving effects. For example, load aggregators need the power grid to have a higher response capability to improve the degree of aggregation of customers. For example, power sales agents need more favorable electricity prices to improve their business capabilities, and need the power grid company to have a higher satisfaction to improve service quality. For example, carbon trading service providers need the power grid company to accurately calculate its carbon emission quota and actual emissions for accurate carbon trading. In addition, a unified demand survey form can be formulated based on the experience known to technical personnel in this field to provide a high-quality data source for generating power supply characteristics, such as investigating the acceptable voltage fluctuation range, restrictions on harmonic content, the maximum load proportion that can participate in demand response, the acceptable maximum electricity price level, the expected carbon accounting methods and standards, etc.

[0063] Based on this, this embodiment provides a data basis for generating power supply characteristics by acquiring historical power supply requirements, and realizes accurate description of the operational needs of energy-saving service providers, load aggregators, power sales agents and carbon trading service providers.

[0064] In this embodiment, natural language processing technology is used to generate power supply characteristics according to historical power supply requirements. The specific operations include: parsing historical power supply requirements using natural language processing technology, capturing corresponding feature thresholds in historical power supply requirements (capturing corresponding feature thresholds in quality requirements, metering requirements, response requirements, price requirements, service requirements and carbon trading requirements), and packaging feature thresholds to obtain power supply characteristics; therefore, the power supply characteristics of this embodiment are feature threshold sets corresponding to power supply requirements, and the feature threshold set stores feature thresholds that characterize power supply requirements.

[0065] Among them, the natural language processing technology of this embodiment is any one of the natural language processing technologies commonly used in the existing processing, and is intended to achieve the extraction of corresponding feature thresholds. For example, if it is recorded that "acceptable voltage fluctuation range: 220V±5%", the corresponding voltage feature threshold is [209V, 231V]; if it is recorded that "desired carbon accounting method and standard: emission factor method and GB / T 32150-2015", the corresponding carbon trading feature threshold is "emission factor method" and "GB / T 32150-2015"; if it is recorded that "acceptable maximum electricity price: 0.45 yuan / kWh", the corresponding price feature threshold is 0.45, etc.

[0066] Therefore, this embodiment achieves the quantification of the operational needs of energy-saving service providers, load aggregators, power sales agents, and carbon trading service providers by generating power supply characteristics, providing a data basis for generating power generation strategies.

[0067] Furthermore, the specific operations of this embodiment for obtaining feature generation rules using feature extraction technology for power supply characteristics include: using feature extraction technology to extract the characteristics of the above-mentioned feature threshold (that is, the characteristics of the feature threshold captured in the process of generating power supply characteristics, and its characteristics are corresponding characteristics that meet the cooperation needs of energy-saving service providers, load aggregators, electricity sales agents and carbon trading service providers), and defining the characteristics of the extracted feature threshold as feature generation rules.

[0068] The feature generation rule of this embodiment is used to guide the cluster control model to generate power supply features based on power supply requirements.

[0069] Among them, the feature extraction technology of this embodiment is any one of the existing conventional feature extraction technology processing, which aims to realize the extraction of semantic features, grammatical features, etc. corresponding to these feature thresholds when capturing feature thresholds using natural language processing technology in the previous text. For example, keywords such as "voltage stability", "metering accuracy", "response time", "carbon emission accounting", etc. correspond to the feature thresholds in quality requirements, metering requirements, response requirements and carbon trading requirements respectively; if a sentence such as "the acceptable maximum electricity price level is [specific value]" appears, the part of speech of words such as "acceptable", "highest", "electricity price level" can be determined through part-of-speech tagging, thereby extracting specific requirements for electricity prices, that is, the feature threshold of the maximum electricity price level, etc.

[0070] In a preferred embodiment, based on the principle of linear discriminant analysis, the specific operation of extracting the features of the feature threshold using feature extraction technology includes:

[0071] Assume that there are four types of entities: energy-saving service providers, load aggregators, electricity sales agents, and carbon trading service providers. E represents the sample of energy-saving service providers, L represents the sample of load aggregators, S represents the sample of electricity sales agents, and C represents the sample of carbon trading service providers. Set in multiple power supply scenario categories (such as different power load types or time periods, etc.) i Represents the number of samples corresponding to the i-th class subject. For each scene category, its sample mean μ i Each feature threshold can be considered comprehensively to calculate (i.e. μ i represents the mean vector calculated based on the characteristic threshold of the i-th subject). For example, considering the power-related threshold, for energy-saving service provider category E, μ E It can be the average power value of all samples under this category and the adjustment value related to the energy saving target; for load aggregator category L, μ L It can be the average power value under this category and the adjustment value related to the load aggregation strategy; the corresponding overall sample mean μ is the comprehensive average value of all four types of main samples at each feature threshold, μ i and μ reflect the differences of different subject categories in each feature threshold, and obtain the intra-class scatter matrix S ω :

[0072]

[0073] Where, X i represents the sample set of the i-th type of subject; x represents a vector containing multiple feature thresholds, such as x = (voltage value, current value, power value, carbon emission value, ...);

[0074] Get the inter-class scatter matrix S b :

[0075]

[0076] Solve equation S w -1 S b υ=λυ, where λ represents the eigenvalue; υ represents the eigenvector corresponding to the eigenvalue. At this time, υ will reflect the importance of different feature thresholds in distinguishing the four types of entities, namely energy-saving service providers, load aggregators, power sales agents and carbon trading service providers, and these feature vectors will tend to extract features that meet the cooperation needs of all parties. The feature vectors corresponding to the largest k eigenvalues ​​are selected for feature extraction. These feature vectors can be interpreted according to the specific feature thresholds and the cooperation needs of all parties. For example, a certain feature vector may have a large weight on feature thresholds such as voltage, power and carbon emissions, indicating that these features play an important role in meeting the cooperation needs of all parties. The above method can increase interpretability and improve classification performance.

[0077] This embodiment provides a technical basis for the cluster control model by generating feature generation rules, and mines the coupling relationship between power grid companies and energy-saving service providers, load aggregators, power sales agents and carbon trading service providers on supply and demand.

[0078] The cluster in the present invention includes a plurality of distributed photovoltaic power generation groups.

[0079] In the present invention, the specific operations of generating a power generation strategy corresponding to the power supply characteristics include: sending the power supply characteristics to the cluster, recording the corresponding actions taken by the cluster after receiving the power supply characteristics (the action is used to control the combined power generation mode of the cluster, the combined power generation mode includes the respective power generation modes of the photovoltaic power generation groups, that is, the power generation modes of each photovoltaic power generation group together constitute the combined power generation mode, and the power generation mode is used to characterize the use of the electricity generated by the photovoltaic power generation group), packaging the corresponding actions taken, and obtaining the power generation strategy (that is, the collection of actions taken by each photovoltaic power generation group in the cluster after the cluster receives the power supply characteristics).

[0080] It should be noted that the power generation mode of this embodiment is determined based on the output characteristics of the actual photovoltaic power generation group. For example, in the full-grid mode, all the electricity generated by the photovoltaic power generation group is transmitted to the power grid. This mode is usually suitable for photovoltaic power generation groups that have no electricity demand or a small power load; such as the surplus power grid mode, the photovoltaic power generation group transmits the excess electricity to the power grid after meeting its own electricity demand. For example, the photovoltaic system installed in the household generates more electricity during the day and the household electricity consumption is less, and the excess electricity can be transmitted to the power grid. In this way, the combination of power sources and the regulation of power value are achieved.

[0081] Therefore, this embodiment improves the breadth of electricity sources and the operational space of electricity value through the combination of power generation modes, provides energy guarantee to meet the different operating needs of energy-saving service providers, load aggregators, electricity sales agents and carbon trading service providers, and generates power generation strategies that provide a data basis for distributed cluster regulation.

[0082] The strategy generation rules of the present invention are used to guide the cluster control model to generate a power generation strategy based on power supply characteristics.

[0083] In the present invention, the specific operation of using feature extraction technology to obtain strategy generation rules for the power generation strategy includes: using feature extraction technology to extract the characteristics of the above-mentioned actions (that is, the characteristics of the corresponding actions taken by the cluster after receiving the power supply characteristics during the generation process of the power generation strategy), and defining the extracted action characteristics as strategy generation rules.

[0084] Similarly, in the process of obtaining the strategy generation rules, the feature extraction technology used is any one of the existing commonly used feature extraction technology processing, which aims to realize the extraction of the characteristics of the corresponding action taken by the record cluster in the previous article after receiving the power supply characteristics. For example, the customers of the load aggregator are large manufacturing enterprises, which have high electricity consumption during the daytime production period. The power grid company combines photovoltaic power generation groups in full-power grid-connected mode and photovoltaic power generation groups in surplus power grid-connected mode to supply them with power; for example, the customers of the energy-saving service provider require highly stable electricity supply, and the power grid company combines photovoltaic power generation groups in full-power grid-connected mode and equipped with energy storage equipment to provide them with uninterrupted power supply.

[0085] Further in a preferred embodiment, based on principal component analysis, the specific operation of extracting the features of the action using feature extraction technology includes:

[0086] The action data of the power generation strategy is represented as a high-dimensional vector set X = {x1, x2, ..., x n}, where x n Represents a specific action state of the cluster after receiving a specific power supply feature, for example, information of multiple dimensions such as the adjustment range of power generation, change of power generation time, change of operating parameters of power generation equipment, etc., which corresponds to the power supply feature;

[0087] Calculate the mean vector of the action data of the power generation strategy This mean vector reflects the average state of the cluster’s actions after receiving different power supply characteristics;

[0088] Calculate the covariance matrix This covariance matrix reflects the correlation between different action features and their changes under the influence of different power supply characteristics;

[0089] Solve the equation Cω=ρω, where ρ represents the eigenvalue of the action data; ω represents the eigenvector corresponding to the eigenvalue of the action data, where ω here reflects the importance and change direction of different action features in the process of generating power generation strategies; select the eigenvectors corresponding to the largest k eigenvalues ​​to form the projection matrix V={v1,v2,v3,v k ,…}, project the action data of the power generation strategy to the projection matrix V, and obtain the result after feature extraction where y i That is, the features of the extracted actions, which reflect the key features of the actions taken by the cluster after receiving the power supply features, and can be used to formulate more effective power generation strategies; the above operation can find an optimal projection direction so that different categories of data can be better separated.

[0090] This embodiment provides a technical basis for the cluster control model by generating strategy generation rules, and mines the production coupling relationship between the power grid company and energy-saving service providers, load aggregators, power sales agents and carbon trading service providers.

[0091] Furthermore, in the present invention, an initial cluster control model is first constructed using machine learning technology, and the initial cluster control model is trained to obtain a cluster control model. The cluster control model generates a real-time power generation strategy under the guidance of feature generation rules and strategy generation rules using real-time power supply requirements.

[0092] The initial cluster control model is constructed by using machine learning technology, and the initial cluster control model is trained by using the feature generation rules, strategy generation rules and historical power supply requirements. The specific operations of obtaining the cluster control model include:

[0093] A1. Storing the feature generation rule and the strategy generation rule in an initial cluster control model, and inputting the historical power supply requirements into the initial cluster control model;

[0094] A2. Based on historical power supply requirements and under the guidance of the feature generation rules, the initial cluster control model is used to generate the power supply features to be verified;

[0095] A3. Based on the power supply characteristics to be verified and under the guidance of the strategy generation rules, the initial cluster control model is used to generate the power generation strategy to be verified;

[0096] A4. Use the historical power supply characteristics and historical power generation strategies corresponding to the historical power supply requirements (the historical power supply characteristics and historical power generation strategies corresponding to the historical power supply requirements can be collected using existing big data technology) to verify the power supply characteristics to be verified and the power generation strategy to be verified;

[0097] A5. Repeat steps A2 to A4 (i.e., repeatedly execute the operation from generating the power supply characteristics to be verified to verifying the power supply characteristics to be verified and the power generation strategy to be verified) until the generated power supply characteristics to be verified and the power generation strategy to be verified meet the preset verification indicators, and obtain the cluster control model that meets the verification indicators.

[0098] In this embodiment, the existing machine learning technology is used to realize the construction and training of the cluster control model, aiming to realize the cluster control model to autonomously use the feature generation rules and strategy generation rules to generate the power generation strategy, such as the expert model technology, which uses the feature generation rules and strategy generation rules as expert rules to guide the generation of power supply characteristics and power generation strategies; the verification index of this embodiment is formulated based on the experience known to technicians in this field, such as the preset generation result accuracy of 80%, which is the verification index.

[0099] Therefore, this embodiment provides necessary technical means for cluster control by building and training a cluster control model.

[0100] In a preferred case, according to the real-time power supply requirements, under the guidance of the feature generation rule and the strategy generation rule, the specific operations of using the cluster control model to generate the real-time power generation strategy include:

[0101] B1. Based on the real-time power supply requirements and under the guidance of the strategy generation rules, the cluster control model is used to generate real-time power supply characteristics;

[0102] B2. Based on the real-time power supply characteristics and under the guidance of strategy generation rules, a cluster control model is used to generate a real-time power generation strategy.

[0103] In the above operation, the cluster control model is used to realize the autonomous real-time power generation strategy generation, and the cluster control of distributed photovoltaics is optimized to adapt to the power grid company's adjustment of power generation strategy under the "four-in-one" business model.

[0104] In summary, this embodiment provides a data basis for generating power supply characteristics through power supply requirements, provides a data basis for generating power generation strategies through power supply characteristics, and obtains feature generation rules and strategy generation rules by extracting the features of the corresponding generation process, thereby providing a technical basis for building a cluster control model, and realizing the mining of the inherent coupling relationship between energy-saving service providers, load aggregators, power sales agents and carbon trading service providers based on the role positioning of the power grid company, and realizing autonomous real-time power generation strategy generation through the cluster control model, providing the necessary technical means for cluster control, and realizing the optimization of distributed photovoltaic cluster control.

[0105] Example 2

[0106] like Figure 2The distributed photovoltaic cluster control method shown includes the following steps:

[0107] S1. Generate power supply features based on historical power supply requirements using natural language processing technology, and obtain feature generation rules using feature extraction technology for the power supply features;

[0108] S2, generating a power generation strategy corresponding to the power supply characteristics, and obtaining a strategy generation rule for the power generation strategy using a feature extraction technique;

[0109] S3. Use machine learning technology to build an initial cluster control model, and use the feature generation rules, strategy generation rules and historical power supply requirements to train the initial cluster control model to obtain a cluster control model;

[0110] S4. According to the real-time power supply requirements, under the guidance of the feature generation rules and the strategy generation rules, using the cluster control model to generate a real-time power generation strategy;

[0111] S5. Regulating the cluster based on the real-time power generation strategy.

[0112] Example 3

[0113] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method in Example 2 are implemented.

[0114] In the above-mentioned embodiments of the present invention, the embodiments described herein can be implemented in hardware, software, firmware, middleware, code or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described herein or their combination. For software implementation, part or all of the process of the embodiment can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein the communication medium includes any medium that is convenient for transmitting a computer program from one place to another. The storage medium can be any available medium that can be accessed by a computer. Computer-readable storage media can include but are not limited to RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0115] The distributed photovoltaic cluster control system, method and storage medium of the present invention provide a data basis for generating power generation strategies through power supply characteristics, and obtain feature generation rules and strategy generation rules by extracting the characteristics of the corresponding generation process, thereby providing a technical basis for building a cluster control model, and realizing the mining of the inherent coupling relationship with energy-saving service providers, load aggregators, power sales agents and carbon trading service providers based on the role positioning of the power grid company, and realizing autonomous real-time power generation strategy generation through the cluster control model, providing the necessary technical means for cluster control, and realizing the optimization of distributed photovoltaic cluster control.

[0116] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0117] The preferred embodiments of the present invention are described in detail above, but the present invention is not limited thereto. Within the technical concept of the present invention, the technical solution of the present invention can be subjected to a variety of simple modifications, including the combination of various technical features in any other suitable manner, and these simple modifications and combinations should also be regarded as the contents disclosed by the present invention and belong to the protection scope of the present invention.

Claims

1. A distributed photovoltaic cluster control system, characterized in that: The system includes a feature processing module, a strategy processing module, a model building module, a model operation module and a cluster control module; The feature processing module is used to generate power supply features according to historical power supply requirements using natural language processing technology, and obtain feature generation rules for the power supply features using feature extraction technology; The strategy processing module is used to generate a power generation strategy corresponding to the power supply characteristics, and obtain a strategy generation rule for the power generation strategy using a feature extraction technology; The model building module is used to build an initial cluster control model using machine learning technology, and train the initial cluster control model using the feature generation rules, strategy generation rules and historical power supply requirements to obtain a cluster control model; The model operation module is used to generate a real-time power generation strategy using the cluster control model under the guidance of the feature generation rule and the strategy generation rule according to the real-time power supply requirements; The cluster control module is used to control the cluster based on the real-time power generation strategy.

2. The distributed photovoltaic cluster control system according to claim 1 is characterized in that: Power supply requirements include quality requirements, metering requirements, response requirements, price requirements, service requirements and carbon trading requirements.

3. The distributed photovoltaic cluster control system according to claim 1 or 2, characterized in that: The specific operation of generating power supply characteristics according to historical power supply requirements using natural language processing technology includes: parsing the historical power supply requirements using natural language processing technology, capturing the corresponding feature thresholds in the historical power supply requirements, packaging the feature thresholds, and obtaining power supply characteristics.

4. The distributed photovoltaic cluster control system according to claim 3 is characterized in that: The specific operation of obtaining the feature generation rule by using the feature extraction technology for the power supply feature includes: extracting the feature of the feature threshold by using the feature extraction technology, and defining the feature of the extracted feature threshold as the feature generation rule; The specific operation of extracting the feature of the feature threshold by using the feature extraction technology includes: Get the intra-class scatter matrix S ω : In the formula, X i represents the sample set of the i-th type of subject; x represents a vector containing multiple feature thresholds; E represents the sample of energy-saving service providers; L represents the sample of load aggregators; S represents the sample of power sales agents; C represents the sample of carbon trading service providers; μ i represents the mean vector of the i-th class subject calculated based on the feature threshold; Get the inter-class scatter matrix S b : Where n i Indicates the number of samples corresponding to the i-th category subject; Solve equation S w -1 S b υ=λυ, where λ represents the eigenvalue; υ represents the eigenvector corresponding to the eigenvalue.

5. The distributed photovoltaic cluster control system according to claim 1 is characterized in that: The specific operation of generating a power generation strategy corresponding to the power supply feature includes: sending the power supply feature to a cluster, recording a corresponding action taken by the cluster after receiving the power supply feature, packaging the corresponding actions taken, and obtaining a power generation strategy; The cluster includes a plurality of distributed photovoltaic power generation groups.

6. The distributed photovoltaic cluster control system according to claim 5, characterized in that: The specific operation of obtaining the strategy generation rule by using the feature extraction technology for the power generation strategy includes: extracting the features of the action by using the feature extraction technology, and defining the extracted features of the action as the strategy generation rule; The specific operations of extracting the features of the action using feature extraction technology include: The action data of the power generation strategy is represented as a high-dimensional vector set X = {x1, x2, ..., x n }, where x n Indicates the specific action status of the cluster after receiving the power supply characteristics; Calculate the mean vector of the action data of the power generation strategy Calculate the covariance matrix Solve the equation Cω=ρω, where ρ represents the eigenvalue of the action data; ω represents the eigenvector corresponding to the eigenvalue of the action data.

7. The distributed photovoltaic cluster control system according to claim 1, characterized in that: The initial cluster control model is constructed by using machine learning technology, and the initial cluster control model is trained by using the feature generation rules, strategy generation rules and historical power supply requirements. The specific operations of obtaining the cluster control model include: Storing the feature generation rule and the strategy generation rule in an initial cluster control model, and inputting historical power supply requirements into the initial cluster control model; Based on historical power supply requirements, under the guidance of the feature generation rules, the initial cluster control model is used to generate the power supply features to be verified; Based on the power supply characteristics to be verified, under the guidance of the strategy generation rules, using the initial cluster control model to generate the power generation strategy to be verified; Respectively verifying the power supply feature to be verified and the power generation strategy to be verified using the historical power supply feature and the historical power generation strategy corresponding to the historical power supply requirements; Repeat the operations of generating the power supply characteristics to be verified to verifying the power supply characteristics to be verified and the power generation strategy to be verified until the generated power supply characteristics to be verified and the power generation strategy to be verified meet the preset verification indicators, thereby obtaining the cluster control model.

8. The distributed photovoltaic cluster control system according to claim 1, characterized in that: According to the real-time power supply requirements, under the guidance of the feature generation rules and the strategy generation rules, the specific operations of using the cluster control model to generate the real-time power generation strategy include: Based on the real-time power supply requirements, under the guidance of the strategy generation rules, a cluster control model is used to generate real-time power supply characteristics; Based on the real-time power supply characteristics and under the guidance of the strategy generation rules, a cluster control model is used to generate a real-time power generation strategy.

9. A distributed photovoltaic cluster control method, characterized in that: The method comprises the following steps: Using natural language processing technology, power supply features are generated according to historical power supply requirements, and feature extraction technology is used to obtain feature generation rules for power supply features; Generate a power generation strategy corresponding to the power supply characteristics, and obtain a strategy generation rule for the power generation strategy using a feature extraction technique; An initial cluster control model is constructed by using machine learning technology, and the initial cluster control model is trained by using the feature generation rule, strategy generation rule and historical power supply requirements to obtain a cluster control model; According to the real-time power supply requirements, under the guidance of the feature generation rules and the strategy generation rules, the cluster control model is used to generate a real-time power generation strategy; The cluster is regulated based on the real-time power generation strategy.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 9 are implemented.

Images