An artificial intelligence-based carbon platform scheduling optimization system

Abstract

The application discloses an energy-carbon platform scheduling optimization system based on artificial intelligence, which comprises a data acquisition module, a data management module, an energy-carbon modeling module, an artificial intelligence prediction module, a scheduling optimization module, a carbon emission accounting module, an execution control module and a visual interaction module. The system collects energy equipment, energy storage equipment, load equipment, carbon emission monitoring equipment and external market data, and forms a unified energy-carbon data set after cleaning, complementing and standardizing; an energy-carbon coupling model is established based on energy flow, carbon flow and equipment constraints, and an artificial intelligence model is used to predict load, renewable energy output, energy price, carbon trading price and carbon emission trend; further, a scheduling optimization scheme is generated with operation cost and carbon emission as targets, and closed-loop control is realized through execution feedback and rolling optimization. The application can improve the intelligent level of energy-carbon collaborative scheduling, reduce energy cost and carbon emission.

Description

Technical Field

[0001] This invention relates to an energy and carbon platform scheduling and optimization system, and more specifically to an energy and carbon platform scheduling and optimization system based on artificial intelligence. Background Technology

[0002] With the continuous advancement of dual-carbon goals, the operation and management of energy systems has gradually shifted from traditional single energy consumption monitoring and cost control to a comprehensive management model that synergistically optimizes energy consumption, carbon emissions, equipment operation safety, and economic benefits. Energy-consuming entities such as industrial parks, factories, buildings, and data centers typically include multiple energy forms such as electricity, heat, cooling, gas, steam, and water, and are simultaneously connected to energy networks such as photovoltaic, wind power, energy storage, cold and heat source equipment, adjustable loads, and external power grids and gas pipelines. Their energy production, conversion, storage, and consumption processes exhibit significant characteristics of multi-source heterogeneity, multiple time scales, and strong coupling.

[0003] Existing energy management platforms primarily function as data acquisition, energy consumption statistics, report analysis, and anomaly alarms, enabling visualized monitoring of energy usage. However, they lack capabilities in energy load forecasting, renewable energy output forecasting, carbon emission trend forecasting, and adaptive optimization of dispatch strategies. While some energy management systems possess certain equipment control or load regulation functions, they typically rely on manual experience rules or fixed threshold strategies, making it difficult to dynamically adjust dispatch schemes based on real-time load fluctuations, energy price changes, carbon trading price changes, uncertainties in renewable energy output, and changes in equipment operating status.

[0004] Furthermore, existing platforms often remain at the post-hoc statistical stage in carbon emission management, calculating carbon emissions based on cyclical energy consumption data and carbon emission factors, lacking the ability to predict carbon emissions and schedule carbon-constrained operations for future operating cycles. When energy procurement cost targets conflict with carbon emission reduction targets, traditional systems struggle to make comprehensive optimization decisions from multiple dimensions such as economics, low carbon emissions, safety, and feasibility.

[0005] Therefore, there is an urgent need for an energy and carbon platform scheduling optimization system that can integrate multi-source energy carbon data, artificial intelligence prediction, carbon emission accounting, optimized scheduling and closed-loop control, so as to achieve unified modeling of energy flow and carbon flow, coordinated optimization of operating costs and carbon emissions, dynamic updating of scheduling schemes and quantitative evaluation of carbon emission reduction effects, thereby improving the intelligent, low-carbon and refined operation level of integrated energy systems. Summary of the Invention

[0006] The purpose of this invention is to provide an energy and carbon platform scheduling optimization system based on artificial intelligence, in order to solve the problems of insufficient multi-source data fusion capability, disconnect between energy scheduling and carbon emission management, low prediction accuracy, static optimization strategies, difficulty in quantifying carbon emission reduction effects, and lack of closed-loop feedback in control execution of existing energy and carbon management platforms.

[0007] The specific plan is as follows:

[0008] An artificial intelligence-based energy and carbon platform scheduling optimization system includes: a data acquisition module, a data governance module, an energy and carbon modeling module, an artificial intelligence prediction module, a scheduling optimization module, a carbon emission accounting module, an execution control module, and a visualization interaction module. The data acquisition module collects data from energy equipment, energy storage equipment, load equipment, carbon emission monitoring equipment, and external market platforms. The data governance module cleans, verifies, completes, standardizes, and synchronizes the collected data to form a unified energy and carbon dataset. The energy and carbon modeling module establishes a correlation model between energy flow, carbon flow, equipment operating status, energy costs, and carbon emission factors. The artificial intelligence prediction module predicts energy load, renewable energy output, energy prices, carbon trading prices, and carbon emission trends based on the unified energy and carbon dataset. The scheduling optimization module generates scheduling optimization schemes based on the prediction results and correlation models. The carbon emission accounting module calculates the carbon emissions and carbon reductions corresponding to the scheduling optimization schemes. The execution control module converts the scheduling optimization schemes into equipment control commands and issues them for execution.

[0009] The visualization and interactive module is used to display energy operation status, carbon emission status, scheduling results, and carbon performance evaluation results.

[0010] The energy input, energy conversion, energy storage, energy output, and energy consumption processes in an integrated energy system are abstracted into a computable energy flow network. Simultaneously, carbon emissions generated during the production, procurement, conversion, and consumption of different energy types are abstracted into a carbon flow network, enabling the coupling and expression of energy and carbon flows under a unified time scale and unified equipment boundaries. The system acquires real-time operational data and external market data through a data acquisition module, and eliminates data discrepancies caused by different equipment, protocols, sampling periods, and measurement units through a data governance module. This transforms multi-source heterogeneous data into a unified energy-carbon dataset suitable for modeling, prediction, and optimization calculations. Based on this, the energy-carbon modeling module establishes an energy supply-demand balance model, an equipment constraint model, a carbon emission accounting model, and an energy-carbon coupling model, according to energy conservation principles, equipment operating mechanisms, changes in energy storage status, energy price constraints, and carbon emission factors. This provides a mathematical foundation for subsequent optimized scheduling.

[0011] Furthermore, the system performs energy and carbon scheduling optimization according to the following steps:

[0012] S1. Collect carbon data from multiple energy sources: collect data on energy production, energy conversion, energy storage, energy consumption, carbon emissions monitoring, environmental meteorology, and market prices;

[0013] S2. Govern the collected data: perform anomaly identification, missing data completion, unit conversion, time synchronization, and data standardization on the collected data to form a unified energy and carbon dataset;

[0014] S3. Model building: Based on the unified energy and carbon dataset, establish energy flow model, carbon flow model, equipment operation constraint model and energy-carbon coupling model;

[0015] S4. Conduct artificial intelligence forecasting: Utilize artificial intelligence forecasting models to predict energy load, renewable energy output, energy prices, carbon trading prices, and carbon emission trends during future scheduling cycles;

[0016] S5. Constructing the scheduling optimization objective function: Based on the prediction results, the energy-carbon coupling model, and equipment operation constraints, construct a scheduling optimization model with operating costs and carbon emissions as objectives;

[0017] S6. Solve the scheduling optimization scheme: Solve the scheduling optimization model to generate equipment start-up and shutdown plan, equipment output plan, energy storage charging and discharging plan, load adjustment plan and carbon emission control plan;

[0018] S7. Generate control commands according to the scheduling optimization scheme and send them to the corresponding devices for execution;

[0019] S8. Calculate carbon emissions and carbon reductions: Collect execution feedback data, calculate scheduling effectiveness, carbon emissions and carbon reductions, and revise subsequent scheduling strategies based on the calculation results.

[0020] S9. Collect feedback and revise the scheduling plan;

[0021] S10, Perform scrolling optimization.

[0022] Furthermore, in step S3, the energy carbon modeling module establishes an energy balance model based on the energy supply and demand balance relationship. The energy balance model is as follows:

[0023]

[0024] in, Indicates the first Energy-generating equipment during time periods The power supply capacity; Indicates the number of different types of energy production equipment; Indicates the first A single energy storage device during the time period The energy release power; Indicates the number of energy storage devices; Indicates time period Energy power purchased from external energy networks; Indicates time period The load demand power; Indicates the first A single energy storage device during the time period The charging power; Indicates time period Power loss during energy transmission, conversion and distribution.

[0025] Furthermore, in step S4, the artificial intelligence prediction model generates prediction results based on historical operational data, meteorological data, calendar features, production plans, and market price data. The prediction relationship is as follows:

[0026]

[0027] in, Indicates the future number The forecast results for the time period include energy load, renewable energy output, energy prices, carbon trading prices, or carbon emissions; The parameter is Artificial intelligence prediction models; Indicates from to Historical operational data sequences for a given period; Indicates the length of the historical time window; Indicates time period Corresponding meteorological characteristics; Indicates time period Corresponding calendar features; Indicates time period The corresponding production plan or operating condition characteristics; This indicates the prediction step size.

[0028] Furthermore, the artificial intelligence prediction model employs Long Short-Term Memory networks, gated recurrent units, temporal convolutional networks, Transformer models, graph neural networks, random forests, gradient boosting trees, or combinations thereof, and updates the model parameters based on the prediction error, wherein the prediction error is:

[0029]

[0030] in, Indicates the mean absolute error of the prediction; Indicates the number of samples; Indicates the first The true value of each sample; Indicates the first The predicted value for each sample.

[0031] Furthermore, in step S5, the scheduling optimization model constructs an objective function with the goals of minimizing overall operating cost and minimizing carbon emissions. The objective function is:

[0032]

[0033] in, This represents the overall optimization target value; Indicates energy procurement costs; Indicates the cost of carbon emissions; This indicates the costs of starting, stopping, maintaining, and operating the equipment; This indicates the cost of comfort loss caused by load adjustment or demand response; , , , These represent the weighting coefficients corresponding to energy procurement costs, carbon emission costs, equipment operating costs, and comfort loss costs, respectively.

[0034] Furthermore, the energy procurement cost and carbon emission cost are calculated according to the following formulas:

[0035]

[0036]

[0037] in, This indicates the energy procurement cost within the scheduling period; Indicates the number of time periods included in the scheduling period; Indicates time period Energy procurement prices; Indicates time period Energy power purchased from external energy networks; Indicates the duration of a single scheduling period; This represents the carbon emission cost within the scheduling cycle; Indicates time period The carbon trading price or carbon emission cost coefficient; Indicates time period The amount of carbon dioxide emissions.

[0038] Furthermore, in step S8, the carbon emission accounting module calculates carbon emissions based on energy consumption and carbon emission factors, and calculates carbon emission reductions based on the carbon emission difference between the baseline scheduling scheme and the actual optimized scheduling scheme; wherein, the carbon emission calculation formula is:

[0039]

[0040] in, Indicates time period Carbon dioxide emissions; Indicates the quantity of energy types; Indicates the first Energy types during time periods Consumption amount; Indicates the first The standard coal equivalent or calorific value conversion factor for this type of energy; Indicates the first The carbon emission factor corresponding to each type of energy; the formula for calculating carbon emission reduction is:

[0041]

[0042] in, Indicates carbon emission reduction; This represents the carbon emissions corresponding to the baseline scheduling scheme; This represents the carbon emissions corresponding to the actual executed scheduling optimization scheme.

[0043] Furthermore, s7 also includes the following sub-steps:

[0044] S701, The scheduling optimization scheme is parsed into equipment start / stop commands, equipment output setting commands, energy storage charging / discharging commands, and adjustable load response commands;

[0045] S702. Verify control commands based on equipment security boundaries, access rules, communication status, and operational constraints;

[0046] S703: The verified control commands are sent to the corresponding devices via industrial communication protocols or energy management interfaces;

[0047] S704. Obtain equipment execution feedback data and determine the deviation between the actual operating status and the scheduling optimization plan;

[0048] S705. When the deviation exceeds the preset threshold, the scheduling optimization module is triggered to regenerate the scheduling optimization scheme.

[0049] Furthermore, the scheduling optimization module updates the scheduling optimization scheme using a rolling optimization method, and the rolling optimization relationship is as follows:

[0050]

[0051] in, Indicates from time period To the time period The optimal scheduling and control sequence; Indicates from time period To the time period Candidate scheduling control sequences; Indicates the length of the rolling optimization prediction time domain; This represents the comprehensive optimization objective function; Indicates time period The system's real-time operating status; Indicates from time period To the time period The prediction result sequence includes load forecast, renewable energy output forecast, energy price forecast and carbon emission forecast.

[0052] Beneficial effects:

[0053] Compared with existing technologies, this invention improves the data foundation quality of the energy and carbon platform by unifying energy and carbon datasets to achieve the fusion processing of multi-source heterogeneous energy data, carbon emission data, and market data; it unifies the expression of energy flow, carbon flow, equipment constraints, and cost factors through energy-carbon coupling modeling, bridging the gap between energy dispatching and carbon emission management; it enhances the foresight of dispatching optimization by using artificial intelligence prediction models to jointly predict load, output, price, and carbon emission trends; it improves the operational efficiency of the integrated energy system by generating dispatching optimization schemes that balance economy, low carbon emissions, safety, and comfort through comprehensive objective functions and constraint models; it achieves quantitative evaluation of dispatching effectiveness through carbon emission accounting and carbon emission reduction calculations; it realizes closed-loop operation of dispatching schemes from calculation results to equipment execution and strategy correction through execution control and feedback correction mechanisms; and it enables the system to adapt to dynamically changing operating environments through rolling optimization mechanisms, thereby improving the intelligence level and low-carbon operation effect of energy and carbon coordinated dispatching. Attached Figure Description

[0054] Figure 1 This is a flowchart of an energy and carbon platform scheduling optimization based on artificial intelligence;

[0055] Figure 2 This is a partial prediction result obtained from artificial intelligence prediction in step S4 of the embodiment;

[0056] Figure 3 This is a partial scheduling scheme obtained from step S6 of the embodiment. Detailed Implementation

[0057] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0058] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0059] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0060] Example:

[0061] This embodiment uses a comprehensive energy system in an industrial park as an example. The park includes a photovoltaic power generation system, an external power grid connection system, gas-fired boilers, chiller / heater units, energy storage batteries, adjustable loads, and carbon emission monitoring equipment. The photovoltaic installed capacity is 800kW, the rated capacity of the energy storage batteries is 1000kWh, the maximum charging power of the energy storage is 300kW, the maximum discharging power is 300kW, the allowable state of charge range for the energy storage is 20% to 90%, and the initial state of charge is 50%. The maximum electrical load of the park is approximately 1200kW, and the average daily electrical load is approximately 760kW. The dispatch cycle is set to 24 hours, with each dispatch period lasting 1 hour, meaning the dispatch cycle comprises 24 time periods.

[0062] External electricity purchases are subject to time-of-use pricing: RMB 0.35 / kWh during off-peak hours, RMB 0.68 / kWh during normal hours, and RMB 1.10 / kWh during peak hours. The carbon trading price is set at RMB 80 / tCO2. The carbon emission factor for grid-purchased electricity is 0.570 kgCO2 / kWh, and the carbon emission factor for natural gas is 2.15 kgCO2 / m³. The park's control objective is to reduce overall operating costs and carbon emissions while meeting load demand and equipment safety constraints.

[0063] In this embodiment, the energy and carbon platform scheduling optimization system based on artificial intelligence follows the following... Figure 1 Perform the steps shown.

[0064] S1. Collect carbon data from multiple energy sources: collect data on energy production, energy conversion, energy storage, energy consumption, carbon emissions monitoring, environmental meteorology, and market prices;

[0065] S2. Govern the collected data: perform anomaly identification, missing data completion, unit conversion, time synchronization, and data standardization on the collected data to form a unified energy and carbon dataset;

[0066] S3. Model building: Based on the unified energy and carbon dataset, establish energy flow model, carbon flow model, equipment operation constraint model and energy-carbon coupling model;

[0067] S4. Conduct artificial intelligence forecasting: Utilize artificial intelligence forecasting models to predict energy load, renewable energy output, energy prices, carbon trading prices, and carbon emission trends during future scheduling cycles;

[0068] S5. Constructing the scheduling optimization objective function: Based on the prediction results, the energy-carbon coupling model, and equipment operation constraints, construct a scheduling optimization model with operating costs and carbon emissions as objectives;

[0069] S6. Solve the scheduling optimization scheme: Solve the scheduling optimization model to generate equipment start-up and shutdown plan, equipment output plan, energy storage charging and discharging plan, load adjustment plan and carbon emission control plan;

[0070] S7. Generate control commands according to the scheduling optimization scheme and send them to the corresponding devices for execution;

[0071] S8. Calculate carbon emissions and carbon reductions: Collect execution feedback data, calculate scheduling effectiveness, carbon emissions and carbon reductions, and revise subsequent scheduling strategies based on the calculation results.

[0072] S9. Collect feedback and revise the scheduling plan;

[0073] S10, Perform scrolling optimization.

[0074] Step S1: Collect carbon data from multiple energy sources.

[0075] The data acquisition module collects operational data through smart meters, photovoltaic inverters, energy storage converters, gas meters, cold and heat source controllers, carbon emission monitoring equipment, and external market interfaces. The collected data includes the park's electricity load, photovoltaic power generation, energy storage status of charge, energy storage charging and discharging power, externally purchased electricity, natural gas consumption, ambient temperature, solar irradiance, humidity, electricity price, and carbon trading price.

[0076] In this embodiment, the data sampling period is 5 minutes, and the platform aggregates the sampled data into 1 hour of scheduling data. For example, during the period from 8:00 to 9:00, the average electrical load of the park is 820kW, the average photovoltaic output is 210kW, the energy storage state of charge is 52%, the external power purchase is 610kW, the ambient temperature is 27℃, and the solar irradiance is 420W / m².

[0077] Step S2: Process the collected data.

[0078] The data governance module performs anomaly identification, missing data completion, unit conversion, time synchronization, and standardization on the collected data. For electrical load data, if the rate of change between two adjacent sampling points exceeds 40% and the equipment operating status remains unchanged, it is identified as an outlier and replaced with the linear interpolation result of the adjacent valid sampling points. For photovoltaic output data, if the missing time of a sampling point is no more than 15 minutes, it is completed using interpolation of adjacent time periods; if the missing time exceeds 15 minutes, it is completed by combining solar irradiance and historical similar day data.

[0079] For example, if the photovoltaic output data collected at 9:20 is 0kW, while the solar irradiance at the same time is 610W / m², and the photovoltaic output at the sampling points before and after the sampling is 355kW and 368kW respectively, the data governance module will determine that the 0kW data is an outlier and correct it to 361.5kW. After governance, a unified energy and carbon dataset is formed, with a unified data time granularity of 1h, a unified power unit of kW, a unified electricity unit of kWh, and a unified carbon emission unit of kgCO2.

[0080] Step S3: Establish an energy-carbon coupling model.

[0081] The energy and carbon modeling module establishes an energy balance model based on the energy supply and demand relationship in the park:

[0082]

[0083] In this embodiment, energy transmission loss is calculated as 2% of load demand. Taking the period from 12:00 to 13:00 as an example, the predicted load is 900kW, the predicted photovoltaic output is 620kW, and if the energy storage discharge power is 150kW, then the energy loss is 18kW, and the required external power purchase is:

[0084]

[0085] Therefore, by prioritizing the absorption of photovoltaic power and energy storage discharge during this period, the external power purchase can be reduced to 148kW.

[0086] Meanwhile, the state of charge of the energy storage device is updated according to the following formula:

[0087]

[0088] in, Indicates the state of charge of the energy storage in the next time period; Indicates the current state of charge of the energy storage system; This represents the energy storage charging efficiency, which is 0.95 in this embodiment. This represents the energy storage discharge efficiency, which is 0.94 in this embodiment; This indicates the energy storage charging power, measured in kW. This indicates the energy storage discharge power, measured in kW. This indicates the length of a single scheduling period; in this embodiment, it is taken as 1 hour. This indicates the rated capacity of the energy storage battery; in this embodiment, it is taken as 1000 kWh.

[0089] For example, if the electricity price is low between 1:00 and 2:00, and the energy storage is charging at a power of 250kW, with the current state of charge (SOC) at 50%, then the SOC for the next period will be:

[0090]

[0091] This means that the energy storage state of charge has increased from 50% to 73.75%.

[0092] Step S4: Perform artificial intelligence prediction.

[0093] The AI ​​prediction module uses load data, photovoltaic output data, meteorological data, calendar data, production plans, and time-of-use electricity price data from the past 30 days as training samples to build a prediction model. The prediction model adopts a combination of Transformer and Gradient Boosting Tree, where Transformer is used to extract the time-series features of load and photovoltaic output, and Gradient Boosting Tree is used to integrate non-time-series features such as meteorological, calendar, and production plans.

[0094] The predictive relationship is:

[0095]

[0096] in, Indicates the future number Prediction results for the time period; The parameter is Artificial intelligence prediction models; Indicates from to Historical operational data sequences for a given period; This indicates the length of the historical time window; in this embodiment, it is set to 168 hours. It indicates meteorological characteristics, including temperature, humidity, and solar irradiance; It indicates calendar features, including weekdays, holidays, and hourly serial numbers; Indicates the characteristics of production plans or operating conditions; This indicates the prediction step size.

[0097] In this embodiment, the model outputs forecasts for the load and photovoltaic output for the next 24 hours. Some of the forecast results are shown below. Figure 2 As shown.

[0098] The model prediction error is calculated using the mean absolute error:

[0099]

[0100] in, Indicates the mean absolute error; Indicates the number of samples; Indicates the first The true value of each sample; Indicates the first The sample predicted values ​​were verified. The average absolute error of load prediction in this embodiment was 32kW, and the average absolute error of photovoltaic output prediction was 28kW, meeting the scheduling optimization requirements.

[0101] Step S5: Construct the scheduling optimization objective function.

[0102] The scheduling optimization module aims to minimize overall operating costs, while also considering carbon emission costs, equipment operating costs, and comfort loss costs. The objective function is:

[0103]

[0104] In this embodiment, take , , , The formula for calculating energy procurement costs is as follows:

[0105]

[0106] in, This represents the cost of electricity purchases within the dispatch period; This indicates the number of time periods in the scheduling cycle; in this example, it is 24. Indicates time period The electricity purchase price; Indicates time period External power purchase capacity; This indicates the length of the time period; in this example, it is 1 hour.

[0107] The formula for calculating carbon emission costs is:

[0108]

[0109] in, This represents the cost of carbon emissions during the scheduling period; This indicates the carbon trading price, which in this example is 80 yuan / tCO2; Indicates time period The amount of carbon dioxide emissions.

[0110] Step S6: Solve for the scheduling optimization scheme.

[0111] The scheduling optimization module inputs energy balance constraints, energy storage state of charge constraints, energy storage charging and discharging power constraints, external power purchase constraints, adjustable load constraints, and carbon emission constraints into the optimization solver. The solution method employs mixed-integer linear programming. Constraints include: energy storage charging power not exceeding 300kW, energy storage discharging power not exceeding 300kW, energy storage state of charge not lower than 20% and not higher than 90%, maximum adjustable load reduction ratio not exceeding 8%, and maximum power reduction in a single time period not exceeding 80kW.

[0112] After solving, the system generates the following partial scheduling scheme: Figure 3 As shown

[0113] During the period from 1:00 to 2:00, the electricity price is 0.35 yuan / kWh, and the system controls the energy storage to charge at 250kW; during the period from 18:00 to 19:00, the electricity price is 1.10 yuan / kWh and the load is high, so the system controls the energy storage to discharge at 280kW to reduce the power purchased from external sources during peak hours.

[0114] S7. Generate control commands according to the scheduling optimization scheme and send them to the corresponding devices for execution;

[0115] S8. Calculate carbon emissions and carbon reductions: Collect execution feedback data, calculate scheduling effectiveness, carbon emissions and carbon reductions, and revise subsequent scheduling strategies based on the calculation results.

[0116] The carbon emission accounting module calculates carbon emissions based on purchased electricity and carbon emission factors:

[0117]

[0118] For purchased electricity Take 1, Let's take 0.570 kg CO2 / kWh. Taking the period from 18:00 to 19:00 as an example, the optimized purchased electricity volume is 813 kWh. Therefore, the carbon emissions from electricity purchases during this period are:

[0119]

[0120] If optimized scheduling is not adopted, the purchased power during the period from 18:00 to 19:00 will be:

[0121]

[0122] The 23kW figure represents losses calculated based on 2% of the load. Therefore, the unoptimized carbon emissions are:

[0123]

[0124] The carbon emission reduction during this period was:

[0125]

[0126] Based on daily statistics, the baseline dispatch plan resulted in 15,180 kWh of purchased electricity, corresponding to 8,652.6 kg CO2 emissions. After optimizing the dispatch using this invention, the total purchased electricity volume for the day was 13,940 kWh, corresponding to 7,945.8 kg CO2 emissions. The total daily carbon emission reduction was:

[0127]

[0128] Based on a carbon trading price of 80 yuan / tCO2, the daily carbon emission reduction revenue is:

[0129]

[0130] The execution control module translates the scheduling optimization scheme into equipment control commands. For example, at 1:00, the system generates an energy storage charging command: the target device is the energy storage converter PCS-01, the control mode is constant power charging, the target power is 250kW, the duration is 1 hour, and the upper limit of the state of charge is 90%. At 18:00, the system generates an energy storage discharging command: the target device is PCS-01, the control mode is constant power discharging, the target power is 280kW, the duration is 1 hour, and the lower limit of the state of charge is 20%.

[0131] Before the command is issued, the command verification unit performs a security check. Taking a discharge command from 18:00 to 19:00 as an example, if the current state of charge of the energy storage is 82%, the energy storage capacity is 1000kWh, and the discharge efficiency is 0.94, then the state of charge after discharging 280kW continuously for 1 hour will be:

[0132]

[0133] That is, the state of charge after discharge is 52.21%, which is higher than the safety lower limit of 20%, so the instruction passes the verification and is issued for execution.

[0134] Step S9: Collect feedback and revise the scheduling plan.

[0135] After the control command is executed, the data acquisition module continues to collect actual operating data. If there is a deviation between the actual photovoltaic output, load, or equipment power output and the predicted value, the feedback correction unit determines whether the deviation exceeds a preset threshold. In this embodiment, the load deviation threshold is 5%, the photovoltaic output deviation threshold is 10%, and the energy storage power deviation threshold is 5%.

[0136] For example, during the period from 14:00 to 15:00, the predicted photovoltaic output was 580kW, but due to cloud cover, the actual photovoltaic output dropped to 430kW, with the deviation being:

[0137]

[0138] When the deviation exceeds the 10% threshold, the system triggers rolling optimization. The scheduling optimization module reacquires current load, energy storage state of charge, actual photovoltaic output, and future weather forecast data, and recalculates the scheduling plan from 15:00 to 24:00. After re-optimization, the system adjusts the energy storage discharge power from 15:00 to 16:00 from the originally planned 80kW to 160kW, and sets the adjustable load reduction power at 30kW to compensate for the power supply gap caused by the decrease in photovoltaic output.

[0139] Step S10: Perform scrolling optimization.

[0140] In this embodiment, the rolling optimization period is 1 hour, and the prediction time domain length is 6 hours. At the beginning of each new scheduling period, the system executes the following rolling optimization relationship:

[0141]

[0142] in, Indicates from time period To the time period The optimal scheduling and control sequence; Indicates the candidate scheduling control sequence; This indicates the length of the rolling optimization prediction time domain; in this embodiment, it is taken as 6h. This represents the comprehensive optimization objective function; It indicates the current real-time operating status of the system, including the current load, energy storage charge status, equipment operating status, and external power purchase. This represents a sequence of forecast results for the next 6 hours, including load forecast, photovoltaic output forecast, electricity price forecast, and carbon emission forecast.

[0143] The system executes control commands for the current hour only at a time, and re-optimizes based on the latest data in the next time period, thereby avoiding scheduling failures due to prediction errors and operational disturbances.

[0144] After adopting the AI-based energy and carbon platform scheduling optimization system of this embodiment, the daily electricity purchase volume of the park decreased from 15180 kWh to 13940 kWh, a reduction of 1240 kWh; the daily electricity purchase cost decreased from approximately 10486 yuan to approximately 9365 yuan, saving approximately 1121 yuan; the daily carbon emissions decreased from 8652.6 kg CO2 to 7945.8 kg CO2, a reduction of 706.8 kg CO2; and the maximum external power purchase during peak hours decreased from 1093 kW to 813 kW, a reduction of 280 kW. Therefore, this invention can achieve energy cost reduction, carbon emission reduction, photovoltaic absorption improvement, and energy storage operation optimization while meeting the constraints of energy supply and demand balance and equipment safe operation.

[0145] As a further improvement, 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 principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An energy and carbon platform scheduling and optimization system based on artificial intelligence, characterized in that, include: The system comprises a data acquisition module, a data governance module, an energy and carbon modeling module, an artificial intelligence prediction module, a scheduling optimization module, a carbon emission accounting module, an execution control module, and a visualization interaction module. The data acquisition module collects data from energy equipment, energy storage equipment, load equipment, carbon emission monitoring equipment, and external market platforms. The data governance module cleans, verifies, completes, standardizes, and synchronizes the collected data to form a unified energy and carbon dataset. The energy and carbon modeling module establishes a correlation model between energy flow, carbon flow, equipment operating status, energy costs, and carbon emission factors. The artificial intelligence prediction module predicts energy load, renewable energy output, energy prices, carbon trading prices, and carbon emission trends based on the unified energy and carbon dataset. The scheduling optimization module is used to generate a scheduling optimization scheme based on the prediction results and the correlation model; The carbon emission accounting module is used to calculate the carbon emissions and carbon emission reductions corresponding to the scheduling optimization scheme; the execution control module is used to convert the scheduling optimization scheme into equipment control commands and issue them for execution. The visualization and interactive module is used to display energy operation status, carbon emission status, scheduling results, and carbon performance evaluation results.

2. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 1, characterized in that, The system performs energy and carbon scheduling optimization according to the following steps: S1. Collect carbon data from multiple energy sources: collect data on energy production, energy conversion, energy storage, energy consumption, carbon emissions monitoring, environmental meteorology, and market prices; S2. Govern the collected data: perform anomaly identification, missing data completion, unit conversion, time synchronization, and data standardization on the collected data to form a unified energy and carbon dataset; S3. Model building: Based on the unified energy and carbon dataset, establish energy flow model, carbon flow model, equipment operation constraint model and energy-carbon coupling model; S4. Conduct artificial intelligence forecasting: Utilize artificial intelligence forecasting models to predict energy load, renewable energy output, energy prices, carbon trading prices, and carbon emission trends during future scheduling cycles; S5. Constructing the scheduling optimization objective function: Based on the prediction results, the energy-carbon coupling model, and equipment operation constraints, construct a scheduling optimization model with operating costs and carbon emissions as objectives; S6. Solve the scheduling optimization scheme: Solve the scheduling optimization model to generate equipment start-up and shutdown plan, equipment output plan, energy storage charging and discharging plan, load adjustment plan and carbon emission control plan; S7. Generate control commands according to the scheduling optimization scheme and send them to the corresponding devices for execution; S8. Calculate carbon emissions and carbon reductions: Collect execution feedback data, calculate scheduling effectiveness, carbon emissions and carbon reductions, and revise subsequent scheduling strategies based on the calculation results. S9. Collect feedback and revise the scheduling plan; S10, Perform scrolling optimization.

3. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 2, characterized in that, In step S3, the energy carbon modeling module establishes an energy balance model based on the energy supply and demand balance relationship. The energy balance model is as follows: in, Indicates the first Energy-generating equipment during time periods The power supply capacity; Indicates the number of different types of energy production equipment; Indicates the first A single energy storage device during the time period The energy release power; Indicates the number of energy storage devices; Indicates time period Energy power purchased from external energy networks; Indicates time period The load demand power; Indicates the first A single energy storage device during the time period The charging power; Indicates time period Power loss during energy transmission, conversion and distribution.

4. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 2, characterized in that, In step S4, the artificial intelligence prediction model generates prediction results based on historical operational data, meteorological data, calendar features, production plans, and market price data. The prediction relationship is as follows: in, Indicates the future number The forecast results for the time period include energy load, renewable energy output, energy prices, carbon trading prices, or carbon emissions; The parameter is Artificial intelligence prediction models; Indicates from to Historical operational data sequences for a given period; Indicates the length of the historical time window; Indicates time period Corresponding meteorological characteristics; Indicates time period Corresponding calendar features; Indicates time period The corresponding production plan or operating condition characteristics; This indicates the prediction step size.

5. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 4, characterized in that, The artificial intelligence prediction model employs Long Short-Term Memory (LSTM) networks, gated recurrent units (GNUs), temporal convolutional networks (TRNs), Transformer models, graph neural networks, random forests, gradient boosting trees, or combinations thereof, and updates the model parameters based on the prediction error. The prediction error is: in, Indicates the mean absolute error of the prediction; Indicates the number of samples; Indicates the first The true value of each sample; Indicates the first The predicted value for each sample.

6. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 2, characterized in that, In step S5, the scheduling optimization model constructs an objective function with the goals of minimizing overall operating cost and minimizing carbon emissions. The objective function is: in, This represents the overall optimization target value; Indicates energy procurement costs; Indicates the cost of carbon emissions; This indicates the costs of starting, stopping, maintaining, and operating the equipment; This indicates the cost of comfort loss caused by load adjustment or demand response; , , , These represent the weighting coefficients corresponding to energy procurement costs, carbon emission costs, equipment operating costs, and comfort loss costs, respectively.

7. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 6, characterized in that, The energy procurement cost and carbon emission cost are calculated according to the following formulas: in, This indicates the energy procurement cost within the scheduling period; Indicates the number of time periods included in the scheduling period; Indicates time period Energy procurement prices; Indicates time period Energy power purchased from external energy networks; Indicates the duration of a single scheduling period; This represents the carbon emission cost within the scheduling cycle; Indicates time period The carbon trading price or carbon emission cost coefficient; Indicates time period The amount of carbon dioxide emissions.

8. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 2, characterized in that, In step S8, the carbon emission accounting module calculates carbon emissions based on energy consumption and carbon emission factors, and calculates carbon emission reductions based on the carbon emission difference between the baseline scheduling scheme and the actual optimized scheduling scheme; wherein, the carbon emission calculation formula is: in, Indicates time period Carbon dioxide emissions; Indicates the quantity of energy types; Indicates the first Energy types during time periods Consumption amount; Indicates the first The standard coal equivalent or calorific value conversion factor for this type of energy; Indicates the first The carbon emission factor corresponding to each type of energy; the formula for calculating carbon emission reduction is: in, Indicates carbon emission reduction; This represents the carbon emissions corresponding to the baseline scheduling scheme; This represents the carbon emissions corresponding to the actual executed scheduling optimization scheme.

9. The energy and carbon platform scheduling and optimization system based on artificial intelligence according to claim 2, characterized in that, s7 also includes the following sub-steps: S701, The scheduling optimization scheme is parsed into equipment start / stop commands, equipment output setting commands, energy storage charging / discharging commands, and adjustable load response commands; S702. Verify control commands based on equipment security boundaries, access rules, communication status, and operational constraints; S703: The verified control commands are sent to the corresponding devices via industrial communication protocols or energy management interfaces; S704. Obtain equipment execution feedback data and determine the deviation between the actual operating status and the scheduling optimization plan; S705. When the deviation exceeds the preset threshold, the scheduling optimization module is triggered to regenerate the scheduling optimization scheme.

Images