Distributed photovoltaic power-carbon coupling grid-connected control method based on machine learning

Abstract

The present invention provides a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning, the method comprising the following steps: S101, collecting meteorological data, and establishing a prediction model based on the meteorological data using a machine learning algorithm; S102, obtaining parameter information of the inverter, and adjusting the predicted power generation according to the parameter information; S103, monitoring the purchase price and transaction price of electricity, and predicting the user's grid-connected power based on the prediction model; S104, calculating the carbon emission reduction according to the energy storage device and distributed photovoltaics, and calculating the carbon emissions consumed; S105, obtaining the user's gas output power and electricity output power, superimposing the gas price to compare the electricity price, and superimposing the gas carbon emissions, electricity carbon emissions and photovoltaic carbon emissions reduction, and inputting them into the prediction model for benefit analysis. The present invention helps enterprises and users choose the most economical and profit-maximizing power generation strategy, and optimizes energy production, storage and utilization through photovoltaic power generation and energy storage devices.

Description

Technical Field

[0001] The present invention relates to the technical field of electricity-carbon coupling grid-connected control, and in particular to a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning. Background Art

[0002] Many countries and regions are actively promoting energy transformation and turning to a greener, more sustainable and low-carbon energy development model. Among them, distributed photovoltaic power generation has received widespread attention. As an important part of energy transformation, the energy grid-connected system has become a hot spot and key area of ​​development in various countries.

[0003] At present, distributed power regulation is generally carried out by setting fixed algorithms or fixed instructions. When energy changes, regulation is carried out according to the change data. However, photovoltaic power generation will be affected by the weather, and the market has subsidy benefits for carbon emission reduction. During the use of fixed algorithms or fixed instructions, due to the fluctuation of weather, carbon emission reduction subsidies and grid electricity prices, the electricity prices from different sources change, and the power regulation operations performed are difficult to achieve good economic efficiency. Due to improper power regulation, carbon emissions will increase, the cost of electricity for enterprises and users will increase, the market subsidy benefits will be limited, and the power revenue cannot be maximized. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a distributed photovoltaic electric-carbon coupled grid-connected control method based on machine learning to solve the problem that the existing technical solutions cannot accurately perform power control operations, thereby leading to increased carbon emissions and increased electricity costs for enterprises and users.

[0005] To achieve the above object, the present invention provides a distributed photovoltaic electric-carbon coupling grid-connected control method based on machine learning, the method comprising the following steps:

[0006] S101, collecting meteorological data, establishing a prediction model based on the meteorological data using a machine learning algorithm, and predicting the power generation of distributed photovoltaics through the prediction model;

[0007] S102, obtaining parameter information of the inverter, adjusting the predicted power generation according to the parameter information, and inputting the adjusted power generation into the prediction model;

[0008] S103, monitoring the purchase price and transaction price of electricity, and predicting the user's off-grid power based on the prediction model, and supplying electricity to the user based on the predicted off-grid power combined with the power grid, energy storage device, and distributed photovoltaics;

[0009] S104. Calculate the carbon emission reduction based on the energy storage device and distributed photovoltaics, then obtain the transaction price of CCER in the carbon trading market, and count the user's off-grid electricity to calculate the carbon emissions consumed;

[0010] S105. Obtain the user's gas output power and electricity output power, superimpose the gas price to compare with the electricity price, and superimpose the gas carbon emissions, electricity carbon emissions and photovoltaic electricity carbon emission reduction, and input them into the prediction model for benefit analysis, and regulate the user's use of gas and electricity according to the output results of the prediction model.

[0011] Furthermore, the step S101 specifically includes the following steps:

[0012] S11, accessing a micro-meteorological station, collecting meteorological data of solar radiation, temperature, humidity, wind speed, wind direction, geographical characteristics, and meteorological environment in the micro-meteorological station, and combining seasonal factors and time factors;

[0013] S12. Construct a prediction model based on a machine learning algorithm, train the prediction model based on step S11, use the prediction model to predict the power consumption and power load of distributed photovoltaics, and construct a distributed photovoltaic output, power consumption, and load prediction model.

[0014] Furthermore, the step S102 specifically includes the following steps:

[0015] S21, obtaining basic parameter information of the inverter, and extracting electrical parameter information from the basic parameter information;

[0016] S22, performing curve analysis on the electrical parameter information in combination with the basic parameter information to obtain a curve analysis result, wherein the curve analysis result includes data on the impact of irradiance and temperature on distributed photovoltaic power generation;

[0017] S23, inputting the impact data on distributed photovoltaic power generation into the prediction model, thereby correcting the prediction data output by the prediction model, as shown below:

[0018] P pv =P ref ×(G / G ref )×[1+α T ×(TT ref )]

[0019] Among them, P pv is the distributed photovoltaic power generation power, P ref is the rated power under standard test conditions STC, G is the current irradiance, G ref is the standard irradiance under STC, α T is the temperature coefficient, T is the current temperature, T refis the standard temperature at STC.

[0020] Furthermore, the step S103 specifically includes the following steps:

[0021] S31, connecting the grid-connected electricity meter and the purchased electricity meter to access the grid, and obtaining the real-time purchased electricity price, the real-time transaction electricity price and the real-time grid-connected electricity price in the grid;

[0022] S32, analyzing the load of the power grid. When the load of the power grid is low, the purchased electricity is used to store and charge the energy storage device through the electricity purchase meter. When the load of the power grid is peak, the energy storage device is used to supply power to users.

[0023] S33. Predict the amount of off-grid electricity required by users based on the prediction model, and supply electricity to users in combination with the power grid, energy storage devices, and distributed photovoltaics.

[0024] Furthermore, the step S33 specifically includes: when the distributed photovoltaics are used for their own purposes, and the user's off-grid electricity consumption is higher than the distributed photovoltaic power generation, and the off-grid electricity consumption is required, regulating and utilizing the energy storage device to supply power to the user; if the energy storage device and the distributed photovoltaic power generation cannot meet the user's off-grid electricity consumption, the power purchasing meter is used to supply power to the user.

[0025] The step S104 specifically includes the following steps:

[0026] S41. Record the grid-connected sales volume of distributed photovoltaic power generation and energy storage devices;

[0027] S42. Calculate carbon emission reduction based on grid-connected electricity sales, obtain the transaction price of CCER in the carbon trading market, and analyze its price fluctuations;

[0028] S43. According to the user's off-grid electricity consumption, the latest emission factor is obtained, and the consumed carbon emissions are calculated according to the grid emission factor.

[0029] Furthermore, the step S105 specifically includes: setting two power generation modes, namely, surplus grid-connected and full grid-connected, obtaining the economic benefits of surplus grid-connected and full grid-connected according to the benefit analysis data, and regulating the grid-connected meters, distributed photovoltaics, energy storage devices, and electricity purchasing meters according to the economic benefits.

[0030] Furthermore, the surplus electricity is connected to the grid to supply electricity to users through distributed photovoltaics, and the surplus electricity is input into the energy storage device. When the energy storage device does not need to be replenished, the surplus electricity is traded for green electricity using the grid-connected electricity meter;

[0031] The full access to the grid will use all the electricity obtained from distributed photovoltaics to trade green electricity using grid-connected meters.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention helps photovoltaic enterprises and users select the most economical and profit-maximizing power generation strategy by analyzing the prediction model and accurately calculating the economic costs and benefits, thereby promoting the realization of low-carbon and sustainable development goals. Through the use of photovoltaic power generation and energy storage systems, the production, storage and utilization of energy are optimized, waste is reduced, and overall energy utilization efficiency is improved. Grid-connected meters, distributed photovoltaics and energy storage devices are rationally used to ensure maximum benefits within the policy framework, reduce operational risks, and achieve joint optimization of economy, carbon emissions, and gas or electricity strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only preferred embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0035] Figure 1 A schematic diagram of a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning provided in an embodiment of the present invention;

[0036] Figure 2 A schematic diagram of the full Internet access provided by an embodiment of the present invention;

[0037] Figure 3 A schematic diagram of the surplus Internet access principle provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0038] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The enumerated embodiments are only used to explain the present invention and are not used to limit the scope of the present invention.

[0039] like Figure 1 As shown, the present invention provides a flow chart of a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning.

[0040] This embodiment provides a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning, and the method includes the following steps:

[0041] S101, collecting meteorological data, establishing a prediction model based on the meteorological data using a machine learning algorithm, and predicting the power generation of distributed photovoltaics through the prediction model, specifically including the following steps:

[0042] S11, accessing a micro-meteorological station, collecting meteorological data such as solar radiation, temperature, humidity, wind speed, wind direction, geographical characteristics, meteorological environment, etc. in the micro-meteorological station, and combining factors such as season and time, wherein the season and time factors include time-related characteristics such as hour, date, and season;

[0043] S12. According to step S11, a prediction model is constructed based on a machine learning algorithm. The prediction model is used to predict the power and power load of distributed photovoltaics. A distributed photovoltaic output, power, and load prediction model is constructed to achieve distributed photovoltaic output, power prediction, and power load prediction. Based on the prediction data, the future distributed photovoltaic consumption, power generation and supply balance, and operation risk are calculated and deduced, and auxiliary decision-making information is provided, which is expressed as follows:

[0044] P pv =β0+β1·G+β2·T+β3·S+β4·H

[0045] L=β′0+β′1·T+β′2·S+β′3·H

[0046] Among them, P pv is the predicted distributed photovoltaic power generation, β0 is the baseline value, β1 is the impact of solar radiation data on photovoltaic power generation, β2 is the impact of temperature data on photovoltaic power generation, β3 is the impact of seasonal virtual variables on photovoltaic power generation, β4 is the impact of timestamp-related features on photovoltaic power generation, G is solar radiation data, T is temperature data (which has a greater impact on power load), S is the seasonal virtual variable, H is the timestamp-related feature, L is the predicted grid load, β′0 is the baseline value, β′1 is the impact of temperature data on grid load, β′2 is the impact of seasonal virtual variables on grid load, and β′3 is the impact of timestamp-related features on grid load.

[0047] S102, obtaining parameter information of the inverter, adjusting the predicted power generation according to the parameter information, and inputting the adjusted power generation into the prediction model, specifically comprising the following steps:

[0048] S21, obtaining basic parameter information of the inverter, and extracting electrical parameter information from the basic parameter information, wherein the basic parameter information of the inverter includes electrical parameter information such as DC side voltage and current data, AC voltage, current, active power, frequency, power factor, etc.;

[0049] S22, combining the electrical parameter information with the basic parameter information to perform curve analysis on the impact of irradiance and temperature on distributed photovoltaic power generation;

[0050] S23, inputting the impact data on distributed photovoltaic power generation into the prediction model, thereby correcting the prediction data output by the prediction model, as shown below:

[0051] P pv =P ref ×(G / G ref )×[1+α T ×(TT ref )]

[0052] Among them, P pv is the distributed photovoltaic power generation power, P ref is the rated power under standard test conditions STC, G is the current irradiance, G ref is the standard irradiance under STC, α T is the temperature coefficient, T is the current temperature, T ref is the standard temperature at STC;

[0053] Correction of the prediction model output process: Using the real-time irradiance G and temperature T data, the above distributed photovoltaic power generation model is used to calculate the expected distributed photovoltaic power generation power P pv , the actual measured inverter output power is compared with the expected distributed photovoltaic power generation power. If there is a difference, the parameters of the prediction model are adjusted or the prediction output is corrected according to the difference.

[0054] S103, monitoring the purchase price and transaction price of electricity, and predicting the user's off-grid electricity based on the prediction model, and supplying electricity to the user based on the predicted off-grid electricity combined with the power grid, energy storage device, and distributed photovoltaic power generation, specifically including the following steps:

[0055] S31, by connecting the grid-connected electricity meter and the electricity purchase meter, thereby accessing the grid, obtaining the real-time purchase electricity price, real-time transaction electricity price and real-time grid-connected electricity price in the grid, the real-time grid-connected electricity price includes tiered electricity price, time-of-use electricity price, peak-valley electricity price, competitive transaction electricity price, etc.;

[0056] S32. Analyze the load of the power grid. When the load of the power grid is low, the purchased electricity is used to store and charge the energy storage device through the electricity purchase meter. When the load of the power grid is at a peak, the energy storage device is used to supply power to the user. After comprehensive consideration of the photovoltaic grid-connected electricity price and the green certificate trading price, if the trading price is higher than the grid-connected electricity price of the electricity purchase meter, electricity is sold through the grid-connected meter. Otherwise, photovoltaics charge the energy storage device. When photovoltaics are self-generated and used by the user, the user load is greater and the amount of electricity required to be disconnected from the grid is adjusted and the energy storage device is used to supply power to the user.

[0057] S33. Predict the amount of electricity required by users based on the prediction model, and supply electricity to users in combination with the power grid, energy storage devices, and distributed photovoltaics, specifically including:

[0058] Access the green electricity trading market, obtain the green point transaction price, obtain the green electricity transaction price in real time, calculate the comprehensive income of the current grid-connected electricity price plus the green certificate transaction price, integrate the photovoltaic grid-connected electricity price and the green certificate transaction price, or the comprehensive green electricity transaction price, compare the user's off-grid electricity price, determine the grid-connected or local load consumption, and adjust to the "self-generation and self-use, surplus grid-connected" or "full grid-connected" mode;

[0059] Adjust energy storage devices to participate in the auxiliary market and obtain compensation from the market. The auxiliary service market provides grid stability services such as frequency regulation and capacity reserve. Energy storage devices obtain additional income or compensation by participating in these markets.

[0060] When distributed photovoltaic power generation is for self-use, and the user's grid-connected power consumption is higher than the distributed photovoltaic power generation, and the grid-connected power consumption is needed, the energy storage device is used to regulate and supply power to the user. If the energy storage device and photovoltaic power generation cannot meet the user's grid-connected power consumption, the electricity purchasing meter is used to supply power to the user.

[0061] S104, calculate the carbon emission reduction according to the energy storage device and distributed photovoltaic, then obtain the transaction price of CCER in the carbon trading market, and count the user's off-grid electricity to calculate the consumed carbon emissions, which specifically includes the following steps:

[0062] S41. Record the grid-connected sales of distributed photovoltaic power generation and energy storage devices, and count them up to 1,000 kWh, which means applying for a green certificate. When the statistics reach 1,000 kilowatts (1MWh), the sales electricity can apply for 1 green certificate "Green Electricity Certificate" unit, remind users to apply for green certificates, and reserve access to the public platform to automatically apply for green certificates in the future;

[0063] S42. Calculate carbon emission reductions based on grid-connected electricity sales, apply for record keeping of emission reductions, obtain the transaction price of CCER in the carbon trading market, and analyze its price fluctuations. When the price is at its peak, remind users to sell CCER;

[0064] CCER (National Certified Voluntary Emission Reduction) development, based on the principle of optimal benefits of comprehensive electricity prices, regulates the "self-generation and self-use, surplus grid-connected" or "full grid-connected" mode, connects to the grid to generate electricity, records the distributed photovoltaic grid-connected power generation, obtains the latest emission factors (dynamic power emission factors), calculates carbon emission reductions based on power generation, and outputs verification and certification reports;

[0065] S43. According to the user's off-grid electricity consumption, the latest emission factor is obtained, and the carbon emissions consumed are calculated according to the grid emission factor, including:

[0066] Carbon Credits: refers to the rights obtained by reducing greenhouse gas emissions. For photovoltaic power generation systems, especially those that can prove that they are clean energy sources, carbon credits can be generated;

[0067] Calculation of carbon emission reduction: refers to the amount of electricity supplied to the grid and sold by photovoltaic power generation systems and energy storage devices. It is used to determine the amount of carbon dioxide emissions generated per unit of electricity consumed according to local or international standards. For clean energy systems (such as photovoltaic power generation), this value is usually zero or close to zero;

[0068] Apply for recordation of emission reductions: The electricity sales data recorded by the energy meter is usually used as proof of sales, and the distributed photovoltaic power generation and energy storage devices are proved to be clean energy. After the application for recordation of emission reductions is successful, the corresponding CCER certificate can be obtained;

[0069] Obtain CCER transaction prices: Pay attention to the domestic carbon trading market or carbon market platform to obtain real-time CCER transaction prices. Some energy trading platforms may also provide relevant carbon market information and transaction prices.

[0070] Historical data analysis: Analyze CCER transaction price data over the past period of time to identify seasonal changes and trends in prices;

[0071] Market factor analysis: Analyze the market factors that affect CCER prices, such as policy changes, energy market trends, etc.;

[0072] Technical indicator analysis: Use technical indicators (such as moving averages, relative strength index, etc.) to analyze short-term and long-term price trends;

[0073] Finally, based on the results of the fluctuation analysis, when it is determined that the CCER price is at its peak, a selling reminder will be sent to the user.

[0074] S105. Obtain the user's gas output power and electricity output power, superimpose the gas price to compare with the electricity price, and superimpose the gas carbon emissions, electricity carbon emissions and photovoltaic carbon emissions reduction, and input them into the prediction model for benefit analysis. According to the output results of the prediction model, regulate the user's use of gas and electricity, specifically including:

[0075] Set up two power generation modes, namely surplus grid-connected and full grid-connected. According to the benefit analysis data, obtain the economic benefits of surplus grid-connected and full grid-connected. According to the benefit solution, regulate the grid-connected meters, distributed photovoltaics, energy storage devices, and electricity purchase meters;

[0076] Surplus Internet access: Figure 3 As shown, distributed photovoltaics are used to supply electricity to users, and the excess electricity is input into the energy storage device. When the energy storage device does not need to be replenished, the excess electricity is traded using the grid-connected electricity meter;

[0077] Full payment online: Figure 2 As shown, all the electricity obtained by distributed photovoltaics is traded in green electricity using grid-connected electricity meters;

[0078] The surplus grid connection and full grid connection are expressed as follows:

[0079] C = P grid + P REC

[0080] Among them, C is the comprehensive income, and P grid is the current grid connection price, and P REC is the trading price of green certificates, and P consume is the user's off-grid price;

[0081] When P consume ≥ C, it is recommended to connect the surplus electricity to the grid, input the excess electricity into the energy storage device, and when the energy storage device does not need to be replenished, sell the excess electricity to the power grid;

[0082] When P consume < C, it is recommended to connect all the generated electricity to the grid, that is, sell all the generated electricity to the power grid;

[0083] According to the benefit analysis, when the economic benefit of using gas is high, it is recommended to regulate and control the user to use gas; when the benefit of using electricity is high, it is recommended to regulate and control the user to use electricity. Among them, the electricity is the off-grid electricity and photovoltaic green electricity respectively. Through the prediction model and detailed data analysis, decision-making support is provided to help users and photovoltaic enterprises make power generation strategy decisions based on data and actual situations, and improve the scientificity and reliability of decision-making.

[0084] The input into the prediction model for benefit analysis specifically includes:

[0085] Let the gas output power be P gas , and the electricity output power be P electric , the gas price be C gas , the electricity price be C electric , the gas carbon emission be E gas , the electricity carbon emission be E electric , and the photovoltaic electricity carbon emission reduction be E solar ;

[0086] For each energy method (gas and electricity), calculate its total energy cost according to the power and price. The gas total cost formula is as follows:

[0087] C gas,total = P gas × C gas (yuan)

[0088] Among them, C gas,total is the total gas cost;

[0089] The electricity total cost formula is as follows:

[0090] Celectric,total =P electric ×C electric (Yuan)

[0091] Among them, C electric,total Total electricity cost

[0092] The calculation methods of carbon emissions of different energy sources are similar. The carbon emissions of each energy source are obtained by multiplying the carbon emissions per unit power by the output power. The formula for carbon emissions of gas is as follows:

[0093] E gas,total =P gas ×E gas (kgCO2)

[0094] Among them, E gas,total is the total carbon emission of gas;

[0095] The formula for carbon emissions from electricity consumption is as follows:

[0096] E electric,total =P electric ×E electric (kgCO2)

[0097] Among them, E electric,total is the total carbon emissions from electricity consumption;

[0098] The formula for photovoltaic electricity carbon emission reduction is as follows:

[0099] E solar,total =P electric ×E solar (kgCO2)

[0100] Among them, E solar,total is the total amount of carbon emissions reduction from photovoltaic electricity;

[0101] An important aspect of comprehensive cost analysis is to compare the costs of different energy forms, especially the choice of gas and electricity. The total cost can be obtained by the following formula:

[0102] C total =C gas,total +C electric,total

[0103] Among them, C total is the total cost;

[0104] Calculate total carbon emissions, taking both gas and electricity into account:

[0105] E total =E gas,total +E electric,total -E solar,total

[0106] Among them, Etotal is the total carbon emissions.

[0107] The use of photovoltaic power will reduce carbon emissions caused by electricity use. Therefore, when calculating total carbon emissions, it is necessary to subtract the emission reduction of photovoltaic power. These data are input into the prediction model to evaluate the long-term benefits. The prediction model uses a machine learning model. During the implementation process, it is optimized according to the prediction results. According to the fluctuations in electricity and gas prices in different time periods, energy selection is optimized, and carbon emissions are further reduced by increasing photovoltaic power generation or optimizing equipment efficiency.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A distributed photovoltaic electric-carbon coupling and grid control method based on machine learning, characterized in that: The method comprises the following steps: S101, collecting meteorological data, establishing a prediction model based on the meteorological data using a machine learning algorithm, and predicting the power generation of distributed photovoltaics through the prediction model; S102, obtaining parameter information of the inverter, adjusting the predicted power generation according to the parameter information, and inputting the adjusted power generation into the prediction model; S103, monitoring the purchase price and transaction price of electricity, and predicting the user's off-grid power based on the prediction model, and supplying electricity to the user based on the predicted off-grid power combined with the power grid, energy storage device, and distributed photovoltaics; S104. Calculate the carbon emission reduction based on the energy storage device and distributed photovoltaics, then obtain the transaction price of CCER in the carbon trading market, and count the user's off-grid electricity to calculate the carbon emissions consumed; S105. Obtain the user's gas output power and electricity output power, superimpose the gas price to compare with the electricity price, and superimpose the gas carbon emissions, electricity carbon emissions and photovoltaic electricity carbon emission reduction, and input them into the prediction model for benefit analysis, and regulate the user's use of gas and electricity according to the output results of the prediction model.

2. According to a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 1, it is characterized in that: The step S101 specifically includes the following steps: S11, accessing a micro-meteorological station, collecting meteorological data of solar radiation, temperature, humidity, wind speed, wind direction, geographical characteristics, and meteorological environment in the micro-meteorological station, and combining seasonal factors and time factors; S12. Construct a prediction model based on a machine learning algorithm, train the prediction model based on step S11, use the prediction model to predict the power consumption and power load of distributed photovoltaics, and construct a distributed photovoltaic output, power consumption, and load prediction model.

3. According to a distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning in claim 1, it is characterized in that: The step S102 specifically includes the following steps: S21, obtaining basic parameter information of the inverter, and extracting electrical parameter information from the basic parameter information; S22, performing curve analysis on the electrical parameter information in combination with the basic parameter information to obtain a curve analysis result, wherein the curve analysis result includes data on the impact of irradiance and temperature on distributed photovoltaic power generation; S23, input the impact data into the prediction model, so as to correct the prediction data output by the prediction model, as shown below: P pv =P ref ×(G / G ref )×[1+α T ×(T-T ref )] Among them, P pv is the distributed photovoltaic power generation power, P ref is the rated power under standard test conditions STC, G is the current irradiance, G ref is the standard irradiance under STC, α T is the temperature coefficient, T is the current temperature, T ref is the standard temperature at STC.

4. The distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 1 is characterized in that: The step S103 specifically includes the following steps: S31, connecting the grid-connected electricity meter and the purchased electricity meter to access the grid, and obtaining the real-time purchased electricity price, the real-time transaction electricity price and the real-time grid-connected electricity price in the grid; S32, analyzing the load of the power grid. When the load of the power grid is low, the purchased electricity is used to store and charge the energy storage device through the electricity purchase meter. When the load of the power grid is peak, the energy storage device is used to supply power to users. S33. Predict the amount of off-grid electricity required by users based on the prediction model, and supply electricity to users in combination with the power grid, energy storage devices, and distributed photovoltaics.

5. The distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 3 is characterized in that: The step S33 specifically includes: when the distributed photovoltaic power generation is for self-use, and the user's off-grid power consumption is higher than the distributed photovoltaic power generation, and the off-grid power consumption is required, regulating and utilizing the energy storage device to supply power to the user; if the energy storage device and the distributed photovoltaic power generation cannot meet the user's off-grid power consumption, the power purchasing meter is used to supply power to the user.

6. The distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 1 is characterized in that: The step S104 specifically includes the following steps: S41. Record the grid-connected sales volume of distributed photovoltaic power generation and energy storage devices; S42. Calculate carbon emission reduction based on grid-connected electricity sales, obtain the transaction price of CCER in the carbon trading market, and analyze its price fluctuations; S43. According to the user's off-grid electricity consumption, the latest emission factor is obtained, and the consumed carbon emissions are calculated according to the grid emission factor.

7. The distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 1 is characterized in that: The step S105 specifically includes: setting two power generation modes, namely, surplus grid-connected and full grid-connected, obtaining the economic benefits of surplus grid-connected and full grid-connected according to the benefit analysis data, and regulating the grid-connected meters, distributed photovoltaics, energy storage devices, and electricity purchase meters according to the economic benefits.

8. The distributed photovoltaic electricity-carbon coupling grid-connected control method based on machine learning according to claim 6 is characterized in that: The surplus power is connected to the grid to supply power to users through distributed photovoltaics, and the surplus power is input into the energy storage device. When the energy storage device does not need to be replenished, the surplus power is traded for green electricity using the grid-connected electricity meter; The full access to the grid will use all the electricity obtained from distributed photovoltaics to trade green electricity using grid-connected meters.

Images