Demand-side low-carbon response method based on carbon emission factor prediction

Abstract

The invention discloses a demand-side low-carbon response method based on carbon emission factor prediction, and belongs to the technical field of demand-side low-carbon response, and the method comprises the steps: predicting a dynamic carbon emission factor in a period of time in the future; the user side system generates a planned load curve after the intelligent electric equipment is adjusted according to the dynamic carbon emission factor and a preset power utilization strategy, feeds back the planned load curve to the carbon emission factor prediction model, and corrects historical load data; iteratively executing the two steps until the dynamic carbon emission factor prediction result is converged; inputting the dynamic carbon emission factor prediction result into a user side system, and generating a final planned load curve according to the carbon emission factor and a user preset power utilization strategy; and the intelligent electric equipment adjusts the operation power of the intelligent electric equipment according to the final planned load curve so as to realize the low-carbon response of the demand side. According to the method, the synchronous and automatic response of the demand side to the high-time-resolution carbon signal is realized, and the low-carbon transfer of the demand side load is effectively guided while the prediction precision of the carbon emission factor is improved.

Description

Technical Field

[0001] This invention relates to a demand-side low-carbon response method based on carbon emission factor prediction, belonging to the field of demand-side low-carbon response technology. Background Technology

[0002] Achieving the "dual carbon" goals requires the power system to accelerate its transformation towards low carbon emissions. Traditionally, research on carbon emission control and reduction has focused primarily on the generation side. However, the "source follows load" nature of the power system means that demand-side electricity consumption behavior is the root cause of carbon emissions from the generation side. Guiding the demand side to adjust its electricity consumption patterns, shifting load from high-carbon emission periods to low-carbon emission periods, is a key path to reducing the overall carbon emission level of the power system.

[0003] In existing technologies, the main means of guiding demand-side response is electricity price signals, using economic levers such as time-of-use pricing to incentivize users to shift peak loads and fill valleys. However, users' electricity choices are not only influenced by electricity prices but also closely related to their environmental awareness. An emerging approach is to introduce electricity "carbon labels," which provide users with dynamic carbon emission factors of the power system, allowing them to intuitively perceive the carbon emission costs of electricity consumption and thus proactively participate in low-carbon electricity use.

[0004] With the development of forecasting technologies, the time resolution of carbon emission factor predictions has improved from daily to hourly and even minute-level. However, traditional electricity consumption patterns on the user side have inertia and stability, making it difficult to respond synchronously with high-frequency carbon signals at the minute level. Existing demand-side response technologies lack automated and intelligent control methods that can match high-time-resolution carbon forecasts, resulting in the carbon signal's guiding role not being fully realized.

[0005] Therefore, there is an urgent need for a new demand-side response method that can be coordinated with high-precision carbon emission factor prediction to achieve minute-level automatic and accurate response of demand-side load to carbon signals, thereby effectively tapping the carbon emission reduction potential of the demand side. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a demand-side low-carbon response method based on carbon emission factor prediction. This method can receive dynamic carbon emission factor prediction data with high temporal resolution and drive user-side smart electrical devices to respond automatically with the same temporal resolution, forming a closed-loop optimization of "prediction-response-correction". This improves the accuracy of carbon prediction while achieving efficient demand-side low-carbon response.

[0007] The technical solution adopted by this invention to solve its technical problem is as follows: This invention provides a demand-side low-carbon response method based on carbon emission factor prediction, comprising the following steps: Step S1: Obtain future weather conditions, unit power generation status and historical electricity consumption data, and predict the dynamic carbon emission factor for a period of time in the future through the carbon emission factor prediction model. Step S2: The user-side system generates a planned load curve for smart electrical equipment after adjusting it based on the predicted dynamic carbon emission factor and the user's preset electricity consumption strategy. The planned load curve is then fed back to the carbon emission factor prediction model to correct historical load data. Step S3: Iteratively execute steps S1 and S2 until the difference between the predicted dynamic carbon emission factors obtained from two adjacent iterations meets the preset convergence condition. Step S4: Input the dynamic carbon emission factor prediction results that meet the convergence conditions into the user-side system. The user-side system generates the final planned load curve based on the carbon emission factor and the user's preset electricity consumption strategy. Step S5: The intelligent electrical equipment adjusts its operating power according to the final planned load curve to achieve a low-carbon response on the demand side.

[0008] As one possible implementation of this embodiment, step S1 includes the following steps: To obtain future weather conditions, generator unit power generation status, and users' historical electricity consumption data; The meteorological conditions, generator unit power generation status, and historical electricity consumption data are input into the fully trained carbon emission factor prediction model. The carbon emission factor prediction model is used to calculate and output a dynamic carbon emission factor sequence with a set time resolution for a future period.

[0009] As one possible implementation of this embodiment, the carbon emission factor prediction model is a spatiotemporal neural network model, and its prediction process includes the following stages: In the coding stage, the input historical load data and meteorological sequences are coded to extract their temporal trends and periodic characteristics; In the decoding and calculation stage, combined with future weather conditions and unit state constraints, the power system balance state at each future moment is derived through multiple rounds of spatiotemporal convolution and attention calculation, and the power generation output of each unit is predicted. In the carbon flow calculation stage, based on the predicted power flow distribution of the power grid and the carbon emission intensity of each unit, a carbon flow calculation model based on the proportional sharing principle is adopted to distribute the carbon emissions on the generation side to each load node, thereby calculating the dynamic carbon emission factor.

[0010] As one possible implementation of this embodiment, the carbon flow calculation stage specifically involves: based on the future power generation output prediction results of each unit outputted by the decoding and calculation stage and the known carbon emission intensity of the units, combined with the power flow calculation results, distributing the carbon emissions of the power generation nodes to each line and load node along the power flow direction to obtain the dynamic carbon emission factor sequence.

[0011] As one possible implementation of this embodiment, the time resolution of the dynamic carbon emission factor sequence reaches the minute level, establishing a functional relationship between the carbon emission factor and time.

[0012] As one possible implementation of this embodiment, the future meteorological conditions include at least one parameter among temperature and solar radiation intensity that affects the power generation of the new energy power station, and the forecast data is provided by the meteorological department.

[0013] As one possible implementation of this embodiment, the future generating status of the unit includes at least one parameter that affects the scheduling and output of the generating unit, such as a known maintenance plan and fuel price, and the data is provided by the power plant or grid dispatch center.

[0014] As one possible implementation of this embodiment, the historical electricity consumption data includes the historical operating load data of smart electrical equipment during the same period, and its prediction accuracy is corrected and improved through subsequent iterative steps.

[0015] As one possible implementation of this embodiment, step S2 includes the following steps: The user-side system receives the dynamic carbon emission factor sequence; Based on the dynamic carbon emission factor sequence and the user's preset electricity consumption strategy, an adjusted planned load curve is generated for at least one smart electrical device; the planned load curve represents the time-series distribution of the planned operating power of the smart electrical device over a future period of time. The planned load curve is fed back to the carbon emission factor prediction model to replace or correct the historical load data for the same period used in the next round of prediction.

[0016] As one possible implementation of this embodiment, the user-preset power consumption strategy includes at least one of the following: the strategy of minimizing the total carbon emissions of equipment power consumption, the strategy of prioritizing user comfort, and the user-defined strategy.

[0017] As one possible implementation of this embodiment, the strategy for minimizing the total carbon emissions from electricity consumption aims to minimize the total carbon emissions from electricity consumption. Specifically, this is achieved by controlling the operating load of the intelligent device. E With real-time carbon emission factors C It shows a negative correlation, that is, when C When it rises E Lower, whenC When decrease E Increase, and E The changes are constrained by the user-defined upper limit of the device operating load. E max and lower limit E min between.

[0018] As one possible implementation of this embodiment, the user comfort priority strategy prioritizes user comfort as its primary objective. Specifically, it is implemented by: [details about the operating load of the smart device]. E The default value remains unchanged; only when the predicted carbon emission factor for a future period changes... C Exceeding the user-defined maximum value C max Or below the user-defined minimum value C min Only then can control be achieved within that time period. E Follow C Adjust in the opposite direction to the changes.

[0019] As one possible implementation of this embodiment, the user-defined strategy allows users to set standards for adjusting the power consumption patterns of various smart electrical devices within different time periods or different carbon emission factor ranges.

[0020] As one possible implementation of this embodiment, the step of feeding the planned load curve back to the carbon emission factor prediction model to replace or correct the historical load data used for the next round of prediction specifically involves the carbon emission factor prediction model replacing the original load input data based on historical smart electricity devices with the planned load curve data during the next round of prediction, thereby improving the accuracy of grid load prediction. This operation is used in iterative prediction to use the user's "planned response" based on the current carbon signal as the new load prediction input, thereby simulating the change in grid load status after demand-side response and thus correcting the carbon flow prediction. This is a "simulation update" or "hypothetical correction" of the prediction model input data, rather than a modification of actual historical records.

[0021] As one possible implementation of this embodiment, step S3 includes the following steps: Steps S1 and S2 are executed cyclically to form a closed-loop iterative process of "prediction-response-correction". After each iteration, the difference measure between the latest predicted dynamic carbon emission factor sequence and the previous predicted dynamic carbon emission factor sequence is calculated; Determine whether the difference metric is less than a preset convergence threshold; If so, the iteration is considered to have reached a stable state, the iteration is stopped, and the dynamic carbon emission factor sequence obtained from the last prediction is taken as the final prediction result; otherwise, the next round of iteration continues.

[0022] As one possible implementation of this embodiment, the difference metric is the relative error value of the two prediction results over all corresponding time periods. ε : , In the formula, ε This is the calculated relative error value; P n For the first n The predicted carbon emission factor for a certain period of time in this forecast; P n+1 For the first n The predicted carbon emission factor for the corresponding time period in the +1 prediction.

[0023] As one possible implementation of this embodiment, the preset convergence threshold is preset according to the specific requirements for prediction accuracy.

[0024] The iterative process simultaneously improves the accuracy of carbon emission factor prediction and the low-carbon optimization effect of load response planning.

[0025] As one possible implementation of this embodiment, step S4 includes the following steps: The final dynamic carbon emission factor prediction result that meets the convergence condition after iterative optimization is input into the user-side system; Based on the final dynamic carbon emission factor prediction results and combined with the user's preset electricity consumption strategy, the user-side system generates the final planned load curves for each smart electrical device for actual execution.

[0026] As one possible implementation of this embodiment, the user-preset power consumption strategy is the same as the strategy adopted in step S2, ensuring the consistency of the optimization goal.

[0027] As one possible implementation of this embodiment, the final planned load curve is generated based on the highest precision carbon signal and user preferences, and serves as the direct control basis for achieving a low-carbon response on the demand side.

[0028] As one possible implementation of this embodiment, step S5 includes the following steps: The final planned load curve is then sent to the corresponding smart electrical devices. Each intelligent electrical device adjusts its operating power or operating mode for the corresponding time period in the future based on the received planned load curve; At the same time, the final dynamic carbon emission factor prediction results are displayed to users.

[0029] As one possible implementation of this embodiment, the smart electrical equipment varies depending on the user type. For industrial users, it mainly includes production equipment with adjustable load; for commercial users, it mainly includes air conditioning systems, energy storage systems, and lighting systems in buildings; and for residential users, it mainly includes smart home devices.

[0030] As one possible implementation of this embodiment, when the smart electrical device is an air conditioner, under the strategy of minimizing the total carbon emissions of the device's electricity consumption, its set temperature changes positively correlated with the carbon emission factor; under the user comfort priority strategy, its set temperature remains unchanged by default, and is only adjusted when the carbon emission factor exceeds the threshold range.

[0031] One of the above technical solutions has the following advantages or beneficial effects: 1. It achieves matching of demand-side response and carbon emission factor prediction in terms of time resolution, and can support minute-level carbon signal guidance and automated load adjustment, with fast response speed and high accuracy.

[0032] 2. A closed-loop iterative optimization mechanism of "prediction-response-correction" was constructed. The user-side response behavior (planned load) feeds back to correct the input data of the prediction model, thereby improving the accuracy of the next round of carbon emission factor prediction. In turn, a more accurate carbon signal can guide a better load response, and the two promote each other.

[0033] 3. It provides flexible and diverse user preset strategies, taking into account both carbon emission reduction targets and personalized user needs (such as comfort), making it highly applicable and easy to be accepted and promoted by different types of users such as industrial, commercial and residential users.

[0034] 4. By concretizing abstract carbon emissions into real-time perceptible carbon signals, it effectively stimulates users' proactive emission reduction behavior based on environmental awareness, and provides a feasible technical path for demand-side participation in achieving the "dual carbon" goal.

[0035] This invention receives dynamic carbon emission factor prediction data with high temporal resolution and drives user-side smart electrical devices to respond automatically with the same temporal resolution, forming a closed-loop optimization of "prediction-response-correction." This improves carbon prediction accuracy while achieving efficient demand-side low-carbon response. (See attached figures.) Figure 1 This is a flowchart illustrating a demand-side low-carbon response method based on carbon emission factor prediction, according to an exemplary embodiment. Figure 2 This is a flowchart illustrating a specific implementation of a demand-side low-carbon response based on carbon emission factor prediction, according to an exemplary embodiment. Detailed Implementation

[0036] To more clearly illustrate the technical features of the present invention, the present invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings.

[0037] like Figure 1 As shown in the figure, an embodiment of the present invention provides a demand-side low-carbon response method based on carbon emission factor prediction, which includes the following steps: Step S1: Obtain future meteorological conditions, unit power generation status and historical electricity consumption data, and predict the dynamic carbon emission factor for a future period of time through the carbon emission factor prediction model.

[0038] Specifically, step S1 includes the following steps: To obtain future weather conditions, generator unit power generation status, and users' historical electricity consumption data; The meteorological conditions, generator unit power generation status, and historical electricity consumption data are input into a fully trained carbon emission factor prediction model. The future meteorological conditions include at least one parameter, such as temperature and solar radiation intensity, that affects the power generation capacity of new energy power plants, and the prediction data is provided by the meteorological department. The future generator unit power generation status includes at least one parameter, such as known maintenance plans and fuel prices, that affects generator unit scheduling and output, and the data is provided by the power plant or grid dispatch center. The historical electricity consumption data includes historical concurrent operating load data of smart electrical equipment, and the prediction accuracy is corrected and improved through subsequent iteration steps.

[0039] The carbon emission factor prediction model is used to calculate and output a dynamic carbon emission factor sequence with a set time resolution for a future period.

[0040] As one possible implementation of this embodiment, the carbon emission factor prediction model is a spatiotemporal neural network model, and its prediction process includes the following stages: In the coding stage, the input historical load data and meteorological sequences are coded to extract their temporal trends and periodic characteristics; In the decoding and calculation stage, combined with constraints such as future weather conditions and unit status, the power system balance state at each future moment is derived through multiple rounds of spatiotemporal convolution and attention calculation, and the power generation output of each unit is predicted. In the carbon flow calculation stage, based on the predicted power flow distribution of the power grid and the carbon emission intensity of each unit, a carbon flow calculation model based on the proportional sharing principle is adopted to distribute the carbon emissions on the generation side to each load node, thereby calculating the dynamic carbon emission factor.

[0041] The carbon flow calculation stage specifically involves: based on the future power generation output predictions of each generating unit output from the decoding and calculation stage and the known carbon emission intensity of the generating units, combined with the power flow calculation results, distributing the carbon emissions of the generating nodes to each line and load node along the power flow direction to obtain the dynamic carbon emission factor sequence. The time resolution of the dynamic carbon emission factor sequence reaches the minute level, establishing a functional relationship between the carbon emission factor and time.

[0042] In step S2, the user-side system generates a planned load curve for smart electrical equipment based on the predicted dynamic carbon emission factor and the user's preset electricity consumption strategy. The planned load curve is then fed back to the carbon emission factor prediction model to correct historical load data.

[0043] Specifically, step S2 includes the following steps: The user-side system receives the dynamic carbon emission factor sequence; Based on the dynamic carbon emission factor sequence and the user's preset electricity consumption strategy, an adjusted planned load curve is generated for at least one smart electrical device; the planned load curve represents the time-series distribution of the planned operating power of the smart electrical device over a future period of time. The planned load curve is fed back to the carbon emission factor prediction model to replace or correct the historical load data for the same period used in the next round of prediction.

[0044] As one possible implementation of this embodiment, the user-preset electricity consumption strategy includes at least one of the following: a strategy to minimize the total carbon emissions of equipment electricity consumption, a user comfort-first strategy, and a user-defined strategy. The strategy to minimize the total carbon emissions of equipment electricity consumption aims to minimize total carbon emissions from electricity consumption, and is specifically implemented by controlling the operating load of the smart device. E With real-time carbon emission factors C It shows a negative correlation, that is, when C When it rises E Lower, when C When decrease E Increase, and E The changes are constrained by the user-defined upper limit of the device operating load. E max and lower limit E min Between. The user comfort priority strategy prioritizes user comfort as its primary goal, and is specifically implemented by: the operating load of the smart device. E The default value remains unchanged; only when the predicted carbon emission factor for a future period changes... C Exceeding the user-defined maximum value C max Or below the user-defined minimum value Cmin Only then can control be achieved within that time period. E Follow C The user-defined strategy allows users to set standards for adjusting the power consumption patterns of various smart devices within different time periods or carbon emission factor ranges.

[0045] As one possible implementation of this embodiment, feeding the planned load curve back to the carbon emission factor prediction model to replace or correct the historical load data for the next round of prediction specifically means that the carbon emission factor prediction model replaces the original load input data based on the historical smart electrical equipment with the planned load curve data in the next round of prediction, so as to improve the accuracy of power grid load prediction.

[0046] Step S3: Iterate through steps S1 and S2 until the difference between the predicted dynamic carbon emission factors obtained from two adjacent iterations meets the preset convergence condition.

[0047] Specifically, step S3 includes the following steps: Steps S1 and S2 are executed cyclically to form a closed-loop iterative process of "prediction-response-correction". After each iteration, the latest predicted dynamic carbon emission factor sequence is calculated. P n+1 Compared with the dynamic carbon emission factor sequence obtained in the previous prediction P n The difference measure between them; Determine whether the difference metric is less than a preset convergence threshold; If so, the iteration is considered to have reached a stable state, the iteration is stopped, and the dynamic carbon emission factor sequence obtained from the last prediction is taken as the final prediction result; otherwise, the next round of iteration continues.

[0048] As one possible implementation of this embodiment, the difference metric is the relative error value of the two prediction results over all corresponding time periods. ε : , In the formula, ε This is the calculated relative error value; P n For the first n The predicted carbon emission factor for a certain period of time in this forecast; P n+1 For the first n The predicted carbon emission factor for the corresponding time period in the +1 prediction.

[0049] As one possible implementation of this embodiment, the preset convergence threshold is preset according to the specific requirements for prediction accuracy.

[0050] The iterative process simultaneously improves the accuracy of carbon emission factor prediction and the low-carbon optimization effect of load response planning.

[0051] Step S4: Input the dynamic carbon emission factor prediction results that meet the convergence conditions into the user-side system. The user-side system generates the final planned load curve based on the carbon emission factor and the user's preset electricity consumption strategy.

[0052] Specifically, step S4 includes the following steps: The final dynamic carbon emission factor prediction result that meets the convergence condition after iterative optimization is input into the user-side system; Based on the final dynamic carbon emission factor prediction results and combined with the user's preset electricity consumption strategy, the user-side system generates the final planned load curves for each smart electrical device for actual execution.

[0053] As one possible implementation of this embodiment, the user-preset power consumption strategy is the same as the strategy adopted in step S2, ensuring the consistency of the optimization goal.

[0054] As one possible implementation of this embodiment, the final planned load curve is generated based on the highest precision carbon signal and user preferences, and serves as the direct control basis for achieving a low-carbon response on the demand side.

[0055] Step S5: The intelligent electrical equipment adjusts its operating power according to the final planned load curve to achieve a low-carbon response on the demand side.

[0056] Specifically, step S5 includes the following steps: The final planned load curve is then sent to the corresponding smart electrical devices. Each intelligent electrical device adjusts its operating power or operating mode for the corresponding time period in the future based on the received planned load curve; At the same time, the system displays the final dynamic carbon emission factor prediction results to users for reference and to allow them to adjust the use of non-smart controlled electrical equipment themselves.

[0057] As one possible implementation of this embodiment, the smart electrical equipment varies depending on the user type. For industrial users, it mainly includes production equipment with adjustable load; for commercial users, it mainly includes air conditioning systems, energy storage systems, and lighting systems in buildings; and for residential users, it mainly includes smart home devices.

[0058] As one possible implementation of this embodiment, when the smart electrical device is an air conditioner, under the strategy of minimizing the total carbon emissions of the device's electricity consumption, its set temperature changes positively correlated with the carbon emission factor; under the user comfort priority strategy, its set temperature remains unchanged by default, and is only adjusted when the carbon emission factor exceeds the threshold range.

[0059] Step S5 optimizes the allocation of demand-side electricity load over time, shifting electricity consumption behavior from predicted high-carbon emission periods to low-carbon emission periods, thereby reducing the overall carbon emissions of the system caused by user-side electricity consumption behavior.

[0060] This embodiment constructs a closed-loop iterative system of "prediction-response-correction" to achieve synchronous and automatic response of the demand side to high time resolution carbon signals. While improving the prediction accuracy of carbon emission factors, it effectively guides the low-carbon shift of demand-side loads and provides an innovative technical means for reducing carbon emissions in the power system.

[0061] Figure 2 This is a flowchart illustrating a specific implementation of a demand-side low-carbon response based on carbon emission factor prediction, according to an embodiment of the present invention. Figure 2 As shown, the present invention describes the specific process of demand-side low-carbon response based on carbon emission factor prediction. It mainly involves three parts: carbon emission factor prediction system, user-side system, and smart electrical equipment. Based on the prior acquisition of relevant data, the specific implementation steps are as follows.

[0062] S1.1 inputs future meteorological conditions, generator status, and historical user electricity consumption data into a fully trained spatiotemporal neural network prediction model. Future meteorological conditions, including temperature and solar radiation intensity, affect the power generation of new energy power plants; these forecasts are provided by the meteorological bureau and have high accuracy. Future generator status, including known maintenance plans and fuel prices, affects generator scheduling and output; these are provided by the power plants and also have high accuracy. However, historical user electricity consumption data is significantly influenced by user behavior and has relatively low accuracy. The low accuracy of historical user electricity consumption data affects the prediction of grid load, which in turn affects the prediction of generator scheduling and significantly impacts the prediction of dynamic carbon emission factors, requiring correction to improve prediction accuracy.

[0063] In S1.2, the encoder in the model encodes the input historical load and meteorological sequences, extracting their temporal trends and periodic features. The decoder, combining future weather and unit status conditions, derives the power balance state at each future moment through multiple rounds of spatiotemporal convolution and attention calculations under complex network constraints, power balance constraints, and reserve constraints. The model ultimately outputs predictions of the power generation output of future nodes and each generating unit, and, combined with grid parameters, obtains the active power of each branch and the injected power of each node through power flow calculations. Simultaneously, based on the predicted unit output and the known carbon emission intensity of thermal power units, new energy units, and energy storage units, the model obtains and outputs the carbon emissions of each generating unit. S1.3, based on the proportional sharing principle of carbon emission flow theory, and combined with the active power of each branch, the injected power of each node, the output of each generating unit and the carbon emissions of each generating unit obtained in S1.2, which are synchronized in time, a functional relationship between the carbon emission factor and the time period is established. y = f ( x The time resolution of this carbon emission factor reaches the minute level, and the prediction result of this dynamic carbon emission factor is input into the user-side system. In this embodiment, there are three different user groups on the demand side: a, b, and c, each employing different preset strategies. S2.1, user-side systems a, b, and c respectively receive the predicted dynamic carbon emission factor information for each time period; S2.2, the user-side systems of types a, b, and c, based on the received future dynamic carbon emission factor information and in conjunction with the different electricity consumption strategies preset by users (types a, b, and c), plan and adjust the operation of smart electrical equipment. These user-preset strategies include various types such as minimizing total carbon emissions from equipment electricity consumption, prioritizing user comfort, and user-defined strategies. The strategy of minimizing total carbon emissions from equipment electricity consumption prioritizes minimizing total carbon emissions, i.e., minimizing the operating load of smart equipment. E Will respond to carbon emission factors in real time C The changes. When C When it rises E Reduce appropriately. C When decrease E Increase appropriately. All adjustments are strictly limited to the user-defined upper limit of the equipment operating load. E max and lower limit E min Within this scope. The user comfort-first strategy prioritizes user comfort, meaning it considers the operational load of smart devices. E The default value remains unchanged, and the predicted carbon emission factor is only maintained at a certain future time. C Exceeding the user-defined maximum value C max Or below the user-defined minimum valueC min At that time, the operating load of smart devices E Only during this period of time C The increase should be appropriately reduced, with C The power consumption of smart devices will be adjusted accordingly, with a reduction in the predicted carbon emission factor and a corresponding increase in the power consumption level. User-defined strategies allow users to set standards to adjust the power consumption patterns of different smart devices at different times. Specifically, Category A users adopt a strategy that minimizes the total carbon emissions from device power consumption, Category B users adopt a user comfort-first strategy, and Category C users adopt a custom strategy. The custom strategy for Category C users is to prioritize user comfort when the predicted carbon emission factor is less than or equal to their set value, and to minimize the total carbon emissions from device power consumption when the predicted carbon emission factor is greater than their set value. S2.3, the user-side systems of a, b, and c output the planned load curves of each device after adjustment according to a preset strategy, and input these load curves into the carbon emission factor prediction model. The planned load curve represents the time-series distribution of the planned operating power of the smart electrical equipment within a specific future time range. The carbon emission factor prediction model replaces the original load input data based on the smart electrical equipment of the same historical period with the planned load curve data of the future time period to increase the accuracy of grid load prediction. Other inputs and the prediction model remain unchanged, and carbon emission factor prediction is performed again. S3, repeat S1.1, S1.2, S1.3, S2.1, S2.2, and S2.3. If the maximum relative error between a current dynamic carbon emission factor prediction and the previous prediction for each corresponding time period is greater than or equal to a set value, continue iterating. The set value is used to determine whether the iteration has reached a stable state and is preset according to specific accuracy requirements. The relative error value is used to measure the relative magnitude of change between two adjacent iteration results, and its calculation method is shown in the following formula:

[0064] In the formula: ε This is the calculated relative error value; P n For the first n The predicted carbon emission factor for a certain period of time in this forecast; P n+1 For the first n The predicted carbon emission factor for the corresponding time period in the +1 prediction.

[0065] If the maximum relative error between the prediction result of a certain dynamic carbon emission factor and the result of the previous prediction in each corresponding time period is less than the set value, the iteration process stops and proceeds to S4. S4 inputs the dynamic carbon emission factor prediction results that meet the stopping iteration conditions into the user-side systems of a, b, and c. The user-side systems adjust the operation of smart electrical equipment in combination with the preset power consumption strategies of users of types a, b, and c, and output the adjusted planned operating load curve to the smart electrical equipment. S5, controlled smart electrical devices, including adjustable-load industrial production equipment, smart homes, and building air conditioning and lighting systems, operate according to the adjusted load. Users can also adjust the usage of other non-smart electrical devices based on dynamic carbon emission factor data. Taking air conditioning operation in cooling mode as an example, for Class A users, the air conditioner's set temperature is positively correlated with the carbon emission factor; that is, the set temperature increases when the carbon emission factor increases and decreases when the carbon emission factor decreases. The air conditioner's set temperature adjustment range is strictly limited to the user's preset upper and lower limits. For Class B users, the air conditioner's set temperature remains unchanged. Only when the predicted carbon emission factor value exceeds the user's set maximum value or falls below the user's set minimum value does the set temperature show a positive correlation with the carbon emission factor; that is, the set temperature increases when the carbon emission factor increases and decreases when the carbon emission factor decreases. For Class C users, when the carbon emission factor is greater than the set value, the air conditioner operates as described for Class A users; when the carbon emission factor is less than or equal to the set value, the air conditioner operates as described for Class B users.

[0066] Through the above steps, a low-carbon response on the demand side was achieved.

[0067] This invention allows users to perceive the carbon emissions from future electricity consumption by inputting carbon emission factors to the user side, guiding them to implement low-carbon operation strategies for smart electrical devices. It also adapts to high-time-resolution carbon emission factor prediction. The invention is rationally designed, with simple and easily promoted methods, strong applicability, and the ability to scientifically and intelligently adjust the operation of demand-side smart electrical devices, guiding them to adopt low-carbon electricity consumption behaviors and reduce carbon emissions. Furthermore, it improves the prediction accuracy of carbon emission factors, ensuring the effective operation of the invention.

[0068] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features, and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A demand-side low-carbon response method based on carbon emission factor prediction, characterized in that, Includes the following steps: Step S1: Obtain future weather conditions, unit power generation status and historical electricity consumption data, and predict the dynamic carbon emission factor for a period of time in the future through the carbon emission factor prediction model. Step S2: The user-side system generates a planned load curve for smart electrical equipment after adjusting it based on the predicted dynamic carbon emission factor and the user's preset electricity consumption strategy. The planned load curve is then fed back to the carbon emission factor prediction model to correct historical load data. Step S3: Iteratively execute steps S1 and S2 until the difference between the predicted dynamic carbon emission factors obtained from two adjacent iterations meets the preset convergence condition. Step S4: Input the dynamic carbon emission factor prediction results that meet the convergence conditions into the user-side system. The user-side system generates the final planned load curve based on the carbon emission factor and the user's preset electricity consumption strategy. Step S5: The intelligent electrical equipment adjusts its operating power according to the final planned load curve to achieve a low-carbon response on the demand side.

2. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 1, characterized in that, Step S1 includes the following steps: To obtain future weather conditions, generator unit power generation status, and users' historical electricity consumption data; The meteorological conditions, generator unit power generation status, and historical electricity consumption data are input into the fully trained carbon emission factor prediction model. The carbon emission factor prediction model is used to calculate and output a dynamic carbon emission factor sequence with a set time resolution for a future period.

3. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 1, characterized in that, The carbon emission factor prediction model is a spatiotemporal neural network model, and its prediction process... Includes the following stages: In the coding stage, the input historical load data and meteorological sequences are coded to extract their temporal trends and periodic characteristics; In the decoding and calculation stage, combined with future meteorological conditions and unit state constraints, the power system balance state at each future moment is derived through multiple rounds of spatiotemporal convolution and attention calculation, and the power generation output of each unit is predicted. In the carbon flow calculation stage, based on the predicted power flow distribution of the power grid and the carbon emission intensity of each unit, a carbon flow calculation model based on the proportional sharing principle is adopted to distribute the carbon emissions on the generation side to each load node, thereby calculating the dynamic carbon emission factor.

4. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 1, characterized in that, Step S2 includes the following steps: The user-side system receives the dynamic carbon emission factor sequence; Based on the dynamic carbon emission factor sequence and the user's preset electricity consumption strategy, an adjusted planned load curve is generated for at least one smart electrical device; the planned load curve represents the time-series distribution of the planned operating power of the smart electrical device over a future period of time. The planned load curve is fed back to the carbon emission factor prediction model to replace or correct the historical load data for the same period used in the next round of prediction.

5. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 4, characterized in that, The strategy for minimizing total carbon emissions from electricity consumption aims to minimize total carbon emissions from electricity consumption. Specifically, it involves controlling the operating load of the smart devices. E With real-time carbon emission factors C It shows a negative correlation, that is, when C When it rises E Lower, when C When decrease E Increase, and E The changes are constrained by the user-defined upper limit of the device operating load. E max and lower limit E min between.

6. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 5, characterized in that, The user comfort-first strategy prioritizes user comfort as its primary objective. Specifically, it is implemented by considering the operating load of smart devices. E The default value remains unchanged; only when the predicted carbon emission factor for a future period changes... C Exceeding the user-defined maximum value C max Or below the user-defined minimum value C min Only then can control be achieved within that time period. E Follow C Adjust in the opposite direction to the changes.

7. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 1, characterized in that, Step S3 includes the following steps: Steps S1 and S2 are executed cyclically to form a closed-loop iterative process of "prediction-response-correction". After each iteration, the difference measure between the latest predicted dynamic carbon emission factor sequence and the previous predicted dynamic carbon emission factor sequence is calculated; Determine whether the difference metric is less than a preset convergence threshold; If so, the iteration is considered to have reached a stable state, the iteration is stopped, and the dynamic carbon emission factor sequence obtained from the last prediction is taken as the final prediction result; otherwise, the next round of iteration continues.

8. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 7, characterized in that, The difference metric is the relative error value between the two prediction results across all corresponding time periods. ε : ， In the formula, ε This is the calculated relative error value; P n For the first n The predicted carbon emission factor for a certain period of time in this forecast; P n+1 For the first n The predicted carbon emission factor for the corresponding time period in the +1 prediction.

9. The demand-side low-carbon response method based on carbon emission factor prediction according to claim 1, characterized in that, Step S4 includes the following steps: The final dynamic carbon emission factor prediction result that meets the convergence condition after iterative optimization is input into the user-side system; Based on the final dynamic carbon emission factor prediction results and combined with the user's preset electricity consumption strategy, the user-side system generates the final planned load curves for each smart electrical device for actual execution.

10. The demand-side low-carbon response method based on carbon emission factor prediction according to any one of claims 1-9, characterized in that, Step S5 includes the following steps: The final planned load curve is then sent to the corresponding smart electrical devices. Each intelligent electrical device adjusts its operating power or operating mode for the corresponding time period in the future based on the received planned load curve; At the same time, the final dynamic carbon emission factor prediction results are displayed to users.

Images