A method for detecting carbon emissions of power users

Abstract

The present application discloses a method for detecting carbon emissions of power users, which relates to the field of carbon emission technology. The method comprises: real-time monitoring to obtain the real-time net power generation and the number of electric energy pulses on the user side, and establishing a dynamic estimation model of the carbon emission factor according to historical data and real-time meteorological data; establishing a carbon emission prediction model, establishing a differentiated power load scheduling optimization model, and generating a power load scheduling plan; setting user area division rules according to the power load scheduling plan, dividing the area according to the user's carbon emission behavior, and setting a regional independence strategy; establishing a regional carbon emission model for each area according to the regional independence strategy; achieving the effect of accurately obtaining the real-time power and carbon emission factor of the distributed power source on the user side, improving the accuracy of carbon emission calculation, and establishing a regional carbon emission model to improve the efficiency and accuracy of carbon emission detection.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission, and in particular to a method for detecting carbon emission of an electricity user. Background Art

[0002] With the advancement of new energy and photovoltaic projects, the penetration rate of distributed power generation on the user side has increased significantly, making it more difficult to detect carbon emissions on the user side. However, carbon emissions detection on the user side is an important basis for users to save energy and reduce carbon emissions.

[0003] The current method based on power flow tracking requires the real-time power and carbon emission factors of each power generation node to be known. However, as the penetration rate of distributed power sources in low-voltage distribution networks increases, it is difficult for the control center to accurately obtain the real-time power and node carbon emission factors of each user's distributed power source without real-time power flow calculation. It can be seen that the current detection of carbon emissions on the user side is difficult.

[0004] Application number 202210402190.4 discloses a carbon emissions detection method and device for electricity users. By obtaining the real-time net power generation power and the number of electricity energy pulses on the user side, and using a preset carbon emission factor model, the average carbon emission factor of the user side within the target period is determined; finally, based on the average carbon emission factor and the number of electricity energy pulses, the first carbon emissions of the user side within the target period are generated to realize carbon emissions detection on the user side.

[0005] However, in the prior art, there is uncertainty in obtaining the real-time power and carbon emission factors of distributed power sources on the user side, resulting in inaccurate calculation of carbon emissions. The preset carbon emission factor model is established based on uncertain data analysis and is in a static state, which makes it impossible to obtain accurate carbon emissions. Summary of the invention

[0006] This application solves the problem of data information uncertainty in the prior art, which leads to reduced carbon emission detection accuracy, by providing a method for detecting carbon emissions of electricity users, and achieves the technical effect of accurate identification and improving detection accuracy by establishing a dynamic model.

[0007] The present application provides a method for detecting carbon emissions of an electricity user, the method comprising:

[0008] S100: Real-time monitoring to obtain the real-time net power generation and power consumption pulse number on the user side, and establish a dynamic estimation model of carbon emission factors based on historical data and real-time meteorological data, update the carbon emission factors in real time, and generate a carbon emission analysis report;

[0009] S200: Obtain the carbon emission analysis report, establish a carbon emission prediction model, identify the change trend of carbon emissions and high emission periods, establish a differentiated power load scheduling optimization model, and generate a power load scheduling plan;

[0010] S300: setting a user area division rule according to the power load dispatching scheme, dividing the area according to the user's carbon emission behavior, and setting a regional independence strategy;

[0011] S400: Establishing a regional carbon emission model for each region according to the regional independence strategy, preliminarily calculating the user-side carbon emissions of each partition in the next time period according to the regional carbon emission model, and evaluating and optimizing the preliminary calculation results according to the actual carbon emissions;

[0012] Among them, the calculation formula for user-side carbon emissions is:

[0013]

[0014] in, represents the carbon emissions on the user side, represents the user's electricity consumption in the i-th time period, represents the average carbon emission factor of the user side in the i-th time period, represents the net power generation of distributed generation in the jth time period, represents the carbon emission factor of distributed generation in the jth time period, represents the length of the jth time period, The number of time periods representing electricity consumption, The number of time periods representing the net generated power.

[0015] Furthermore, the real-time meteorological data is converted into characteristic data related to carbon emissions according to conversion rules, and different conversion rules are set according to different distributed power generation characteristics for separate conversions.

[0016] Furthermore, the dynamic estimation model selects a linear regression model as a basic model, takes historical data and converted real-time meteorological data as input, outputs a carbon emission factor, and uses cross-validation and calculation of mean square error to evaluate the output effect.

[0017] Furthermore, the specific calculation formula of the average carbon emission factor is as follows:

[0018]

[0019] in, represents the average carbon emission factor on the user side, represents the power generation of the kth distributed generation in a specific time period, represents the carbon emission factor corresponding to the kth distributed generation, Indicates the number of types of distributed power sources.

[0020] Furthermore, characteristic parameters are extracted according to the carbon emission analysis report and historical data, and a recursive feature elimination method is used to select characteristic parameters. A long short-term memory network model is selected according to the characteristic parameters to establish a carbon emission prediction model.

[0021] Furthermore, the user area division rule is to partition the users by extracting features reflecting the users' electricity consumption behavior and distributed power generation behavior, using a clustering algorithm, and dividing the users into different behavior areas according to the clustering results; the regional independence strategy is to set a difference independence strategy according to different regions, and formulate different identification standards for each behavior area.

[0022] Furthermore, the user area division rule also includes adjusting the area division result according to the clustering result and a preset area floating strategy, and determining whether the constraint condition is met to determine the final area division result.

[0023] Furthermore, the regional floating strategy is to adjust the region according to pre-set regional floating standards and amplitude adjustment principles. The regional floating standards include the differences in carbon emission intensity between regions, the matching degree of peak and valley periods of electricity load, and the complementarity of distributed power generation; the amplitude adjustment principle is to pre-set the amplitude of dynamic partitioning according to actual conditions.

[0024] Furthermore, the regional carbon emission model is established according to the result of each regional division combined with regional characteristics and data features, and the user-side carbon emissions for the next time period are obtained according to the set evaluation criteria based on each regional carbon emission model.

[0025] Furthermore, the pre-set evaluation criteria of the regional carbon emission model are based on statistical analysis of historical carbon emission data, comparing current carbon emissions with historical data, analyzing year-on-year and month-on-month changes, evaluating carbon emission trends, and then generating new user-side carbon emissions.

[0026] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0027] Through real-time monitoring and dynamic estimation models, and comprehensive consideration of multi-source data information, it is possible to accurately estimate the carbon emissions on the user side and dynamically evaluate the data accuracy, thereby accurately obtaining the real-time power and carbon emission factors of distributed power sources on the user side and improving the accuracy of carbon emissions calculations. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1The figure is a flow chart of a method for detecting carbon emissions of an electricity user in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which the present invention belongs; the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more related listed items.

[0031] Embodiment 1: Figure 1 As shown, a method for detecting carbon emissions of an electricity user comprises:

[0032] S100: Real-time monitoring to obtain the real-time net power generation and power consumption pulse number on the user side, and establish a dynamic estimation model of carbon emission factors based on historical data and real-time meteorological data, update the carbon emission factors in real time, and generate a carbon emission analysis report;

[0033] In some embodiments, relevant information is extracted from the historical operating data of distributed power sources (such as solar panels, wind turbines, etc.), including power generation, operating efficiency, maintenance records, etc., and real-time meteorological data such as wind speed, light intensity, and temperature are obtained through meteorological monitoring stations or meteorological data providers. According to the type of distributed power sources (such as photovoltaics, wind power), its characteristic parameters are extracted. For photovoltaic power sources, its characteristic parameters specifically include conversion efficiency, temperature coefficient, maximum power point voltage and current, open circuit voltage and short circuit current. The conversion efficiency refers to the efficiency of photovoltaic panels in converting light energy into electrical energy, expressed as a percentage, which is used to measure the performance of photovoltaic panels and directly affects their power generation; the temperature coefficient refers to the ratio of the power generation efficiency of photovoltaic panels to temperature change, and the efficiency of photovoltaic panels will decrease with increasing temperature; the maximum power point voltage and current refer to the voltage and current values ​​when the photovoltaic panel reaches maximum power output under standard test conditions (STC); the open circuit voltage and short circuit current refer to the voltage of the photovoltaic panel in the open circuit state and the current in the short circuit state, which are used to evaluate the basic performance of the photovoltaic panel. For wind power source types, their characteristic parameters specifically include power curve, cut-in wind speed and cut-out wind speed, rated wind speed, impeller diameter and swept area, etc. The power curve refers to the curve that describes the output power of the wind turbine at different wind speeds; the cut-in wind speed and cut-out wind speed refer to the wind speed thresholds at which the wind turbine starts and stops generating electricity; the rated wind speed refers to the wind speed at which the wind turbine reaches the rated power output; the impeller diameter and swept area refer to the diameter of the wind turbine impeller and the area covered when it rotates, which affect the ability of the wind turbine to capture wind energy.

[0034] In some embodiments, the real-time meteorological data is converted into features related to carbon emissions according to conversion rules, such as the effect of wind speed on the output power of wind turbines and the effect of light intensity on the power generation of photovoltaic panels. The specific conversion process of the conversion rules includes mapping the real-time wind speed to the corresponding output power according to the power curve of the wind turbine. The specific calculation formula for the output power of the wind turbine is:

[0035]

[0036] in, Represents the output power of the wind turbine generator, represents the air density, Indicates the swept area, Indicates wind speed, Represents the power coefficient, which is related to the tip speed ratio and blade angle Related.

[0037] The actual power generation of photovoltaic panels is calculated as follows:

[0038]

[0039] in, Indicates the actual power generation. represents the theoretical power generation, Indicates the loss rate.

[0040] The output power at a specific wind speed can be determined by looking up the power curve table or using an interpolation method; for example, if the real-time wind speed is 10m / s, the output power of the wind turbine at this wind speed can be obtained as 1.5MW by looking up the power curve or interpolating. According to the conversion efficiency and temperature coefficient of the photovoltaic panel, the actual conversion efficiency under the current light intensity and temperature conditions is calculated. Then, the actual conversion efficiency and the physical area of ​​the photovoltaic panel are used to calculate the theoretical power generation; for example, assuming that the conversion efficiency of the photovoltaic panel is , the physical area is , the real-time light intensity is , the temperature is 25 degrees. Considering the effect of temperature coefficient on conversion efficiency (assuming the temperature coefficient is -0.4% / degree), the actual conversion efficiency is 15%. Therefore, the theoretical power generation is .

[0041] The linear regression model is selected as the basic model to establish a dynamic estimation model, and it has a good fitting effect on linear relationship data. Collect historical power generation data of distributed power sources, determine the type of distributed power sources and the power generation and carbon emission factors in a specific time period. The specific time period refers to the time period pre-selected in the specific experimental measurement process. Convert historical meteorological data (such as wind speed, light intensity, temperature, etc.) into features related to carbon emissions, collect or calculate the carbon emission factors of distributed power sources in the corresponding time period as label data, process missing values, fill or delete data records containing missing values; normalize / standardize data, normalize or standardize feature data to improve the convergence speed and performance of the model; use preprocessed historical data and meteorological features as input, and carbon emission factors as output, use cross-validation (such as K-fold cross-validation) to evaluate the generalization ability of the model to avoid overfitting, and evaluate the model by calculating the mean square error (MSE). To improve the accuracy of the model, meteorological data and distributed power supply operation data are collected in real time through meteorological stations, distributed power supply monitoring systems, etc., the real-time data is cleaned and preprocessed, and the preprocessed real-time data is input into the trained linear regression model to estimate the carbon emission factor. New historical data and corresponding meteorological characteristics are collected at regular intervals (such as monthly and quarterly); the newly collected data is merged with the historical data, and data preprocessing, model training and verification are re-performed to update the estimation accuracy of the model; the updated model is deployed to the production environment to replace the old model for real-time carbon emission factor estimation.

[0042] Specifically, the specific calculation formula for the average carbon emission factor is as follows:

[0043]

[0044] in, represents the average carbon emission factor on the user side, represents the power generation of the kth distributed generation in a specific time period, represents the carbon emission factor corresponding to the kth distributed generation, Indicates the number of types of distributed power sources.

[0045] In some embodiments, real-time meteorological data directly affects the output power of distributed power sources. For example, an increase in wind speed will increase the output power of wind turbines, and changes in light intensity will affect the power generation of photovoltaic panels. These changes will indirectly affect the carbon emission factor because the operating efficiency will also be different under different output powers.

[0046] In some embodiments, step S100 also includes: fusing multi-source data information, implementing cross-checking, and automatically identifying and correcting abnormal data to generate a carbon emissions analysis report on the user side.

[0047] In some embodiments, the power generation of the power plant, the power flow data of the power grid, the transmission loss, the node voltage, the power consumption of the user, the output power of the distributed power source, and the real-time meteorological data are collected; the collected multi-source data are integrated to form a unified data format and storage structure. Compare the data from different sources to check the consistency and rationality of the data. For example, compare whether the power generation of the power plant matches the input power on the power grid side; use the dynamic estimation model, combined with the real-time meteorological data and the historical operation data of the distributed power source, to predict the carbon emission factor of the distributed power source in real time. According to the power grid power flow balance relationship and the carbon emission balance relationship model, calculate the average carbon emission factor of the user side after considering the influence of the distributed power source. According to the number of electric energy pulses (i.e., power consumption) and the average carbon emission factor on the user side, calculate the first carbon emission of the user side in each target period. Consider the actual power consumption of the user and the net power generation of the distributed power source (i.e., the power generation after deducting its own losses). Accumulate the first carbon emissions in multiple target periods to obtain the carbon emissions on the user side. The formula for calculating the carbon emissions on the user side is as follows:

[0048]

[0049] in, represents the carbon emissions on the user side, represents the user's electricity consumption in the i-th time period, represents the average carbon emission factor of the user side in the i-th time period, represents the net power generation of distributed generation in the jth time period, represents the carbon emission factor of distributed generation in the jth time period, represents the length of the jth time period, The number of time periods representing electricity consumption, The number of time periods representing the net generated power.

[0050] The calculated carbon emission data is collated, including the carbon emission of each target period and related statistical data and analysis results. A carbon emission analysis report is generated based on the collated data.

[0051] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0052] This application uses real-time monitoring and dynamic estimation models, and comprehensively considers multi-source data information to accurately estimate carbon emissions on the user side and dynamically evaluate data accuracy, thereby accurately obtaining the real-time power and carbon emission factors of distributed power sources on the user side and improving the accuracy of carbon emissions calculations.

[0053] Embodiment 2: In embodiment 1, a dynamic estimation model of carbon emission factors is established, the carbon emission factors are updated in real time, and a carbon emission analysis report on the user side is generated. This embodiment makes further improvements on the basis of the above embodiment.

[0054] S200: Obtain the carbon emission analysis report, establish a carbon emission prediction model, identify the changing trend of carbon emissions and high emission periods, establish a differentiated power load scheduling optimization model, and generate a power load scheduling plan.

[0055] In some embodiments, a carbon emission prediction model is established based on the carbon emission analysis report and historical data information on the user side. The carbon emission data in the past period of time is obtained through the data collection of the above embodiments, including the carbon emission of each time period on the user side, the carbon emission data of each distributed power source, etc. The real-time net power generation, the number of power consumption pulses, the power grid flow data, etc. on the user side are collected through a high-precision real-time data collection system. The data collection frequency is set according to actual needs, such as every minute, every hour, etc. The real-time meteorological data includes meteorological factors such as temperature, humidity, wind speed, and light intensity that may affect the power generation of distributed power sources and the power grid flow.

[0056] Feature parameters are extracted according to the carbon emission analysis report and historical data, feature parameters are selected using a recursive feature elimination method, and a long short-term memory network model is selected according to the feature parameters to establish a carbon emission prediction model.

[0057] Specifically, feature extraction is the process of obtaining useful information about carbon emissions from raw data. For carbon emission prediction, it is necessary to extract features related to distributed power sources, grid operation, meteorological conditions, etc. Specifically, they include: Distributed power source type: Different types of distributed power sources (such as solar power, wind power) have different effects on carbon emissions; Distributed power source capacity: The capacity determines the power generation potential, which affects carbon emissions; Distributed power generation power: The real-time power generation power directly affects the current carbon emissions; Grid flow distribution: Changes in grid flow reflect the transmission and distribution of electric energy, which has an indirect impact on carbon emissions; Temperature: Temperature affects electricity demand, which in turn affects carbon emissions; Humidity: Humidity is also related to electricity demand, especially under certain climate conditions. These features provide information on multiple dimensions that affect carbon emissions, which helps the model predict carbon emissions more accurately.

[0058] In some embodiments, Recursive Feature Elimination is a feature selection method that eliminates the least important features by recursively considering the importance of each feature in the model. Select a base model to provide feature importance scores, use all features to train the base model, output the importance scores of each feature based on the model, remove the features with the lowest scores, and repeat the above steps until a predetermined number of features is reached or a stopping condition is met. It helps to reduce the number of features and reduce model complexity while retaining the features that have the greatest impact on the prediction target and improving model performance. Feature conversion is the process of mathematically transforming the original features to improve model performance. Normalization is performed to scale the data to the same scale, which helps in the training of some models. Improving the distribution of data and improving the model's ability to fit the data improves prediction performance.

[0059] In some embodiments, the long short-term memory network model normalizes / standardizes the extracted and selected features, and divides them into training sets and test sets, defines the number of LSTM layers, the number of units and the activation function, adds a fully connected layer for output, selects the loss function and optimizer, uses the training set to train the LSTM model, applies cross-validation to evaluate the generalization ability of the model, and uses grid search to adjust hyperparameters (such as learning rate, batch size, number of LSTM layers, etc.) to optimize model performance. Use the test set to evaluate the trained model, calculate indicators such as prediction error (such as MSE) and accuracy, draw a comparison chart of the predicted value and the actual value, and intuitively evaluate the model performance. Repeat the training and evaluation steps until satisfactory prediction performance is achieved. Using a neural network as a basic model for training is a technical method well known to people in the field.

[0060] The predicted carbon emissions in the future time period are plotted into a time series graph in chronological order. The X-axis represents time and the Y-axis represents carbon emissions. Identify the overall trend of carbon emissions, carefully observe the time series graph, identify the upward or downward trend of carbon emissions, accurately describe the trend of carbon emissions and analyze the reasons for the trend change, and find out the possible high emission periods in the future time period. Set the carbon emission threshold based on historical carbon emission data. When the predicted value exceeds this threshold, the period is considered to be a high emission period. Mark all periods where the predicted value exceeds the threshold in the time series graph. These periods are possible high emission periods. For each high emission period, deeply analyze the possible causes of high emissions, such as the peak power generation of a specific type of distributed power source, extreme weather conditions, specific economic activities, etc., and organize the trend analysis, high emission period identification and cause analysis into a report.

[0061] In some embodiments, the differentiated power load scheduling optimization model is to minimize the carbon emissions of different users in the future time period. The specific calculation formula for minimizing carbon emissions is as follows:

[0062]

[0063] in, is the scheduling period, is the number of distributed generation sources, is the carbon emission factor of the i-th distributed generation, is the power of the ith distributed generation in period t, It is the carbon emission factor of the i-th distributed generation in period t, which integrates the carbon emission factor after the influence of factors such as time and weather.

[0064] And set constraints to constrain, build the objective function, use the optimization tool to establish, and determine the decision variables, including: the power generation plan of distributed power sources (such as the power generation in each time period); the distribution of power loads (such as the power consumption of each load in each time period); the power grid flow control parameters (such as line transmission power, reactive power compensation, etc., depending on specific needs). The constraints include: power balance constraints: ensure that the power generation in each time period matches the load demand, considering the power grid loss; distributed power operation constraints: such as maximum / minimum power generation, ramp rate limit, etc.; power grid safety constraints: such as line transmission power limit, voltage stability, etc.; user demand constraints: such as power supply guarantee for key loads, user power consumption plan, etc.; meteorological factor constraints: such as the predicted wind speed and light intensity on the power generation of distributed power sources. According to the carbon emission calculation formula, the decision variables such as the power generation of distributed power sources and the load power consumption are linked to carbon emissions. Consider the differences in carbon emission factors in different time periods and different meteorological conditions. Through the differentiated power load scheduling optimization model, the power load scheduling plan for the future time period is generated.

[0065] S300: Setting user area division rules according to the power load dispatching plan, dividing areas according to users' carbon emission behaviors, and setting regional independence strategies; being able to realize real-time detection of partitions and form multi-area and multi-standard carbon emission detection.

[0066] In some embodiments, the user area division rule is to use a clustering algorithm to partition users by extracting features that reflect the user's electricity consumption behavior and distributed power generation behavior, and to divide users with similar electricity consumption behavior and distributed power generation behavior into the same area. According to the clustering results, the users are divided into different behavior areas; the regional independence strategy is to set a difference independence strategy according to different areas, and formulate different identification standards for each behavior area, such as users in high-carbon emission areas have large electricity consumption and are concentrated in peak hours, and users in low-carbon emission areas have stable electricity consumption and large distributed power generation. By real-time monitoring of user electricity consumption behavior and distributed power generation behavior, the user's current behavior area is identified. Differentiated carbon emission monitoring strategies are formulated according to the characteristics of different behavior areas. For example, users in high-carbon emission areas are monitored with emphasis, and the monitoring frequency and accuracy are increased; users in low-carbon emission areas are monitored routinely to ensure that carbon emissions remain at a low level.

[0067] In some embodiments, the objective function of the clustering algorithm is specifically:

[0068]

[0069] in, is the clustering objective function, is the user feature vector, is the center of the kth cluster, is the kth cluster; Refers to the specific number of regions generated by clustering. For example, if the user obtains three regions through clustering, then is 3.

[0070] In some embodiments, differentiated intelligent scheduling plans are formulated for users in different behavior areas. For example, for users in high-carbon emission areas, carbon emissions are reduced by optimizing distributed power generation plans and power load distribution; for users in low-carbon emission areas, the existing scheduling plan is maintained to ensure stable carbon emissions. Intelligent optimization algorithms such as genetic algorithms and particle swarm optimization algorithms are used to iteratively optimize the scheduling plan with the goal of minimizing carbon emissions. The scheduling plan is adjusted in real time based on real-time monitoring results and the execution effect of the intelligent scheduling plan. If it is found that the carbon emissions in a certain behavior area continue to rise or exceed the preset threshold, the reasons are analyzed in a timely manner and the scheduling plan is adjusted. Regularly evaluate the effectiveness of carbon emission monitoring and intelligent scheduling plans, including the degree of reduction in carbon emissions, user satisfaction, and distributed power generation efficiency.

[0071] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0072] This application can generate power load scheduling plans for future time periods by establishing a differentiated power load scheduling optimization model. Through user area division and real-time detection of partitions, it can formulate differentiated carbon emission monitoring strategies and intelligent scheduling plans for users in different behavior areas. By setting user area division rules, regional divisions are carried out according to users' carbon emission behaviors, and regional independence strategies are set; real-time detection of partitions is achieved, and the effect of accurately identifying carbon emissions is achieved.

[0073] Embodiment 3: After dividing the regions according to the clustering results, it is often unable to satisfy the real-time changes of the regions. This embodiment makes further improvements on the basis of the above embodiments.

[0074] The user area division rule also includes adjusting the area division result according to the clustering result and a preset area floating strategy, and determining whether the constraint condition is met to determine the final area division result.

[0075] In some embodiments, clustering results are obtained. The clustering results are obtained by processing user data through a clustering algorithm, dividing users into several areas with similar characteristics (such as carbon emission intensity, peak and valley periods of electricity load, distributed power generation characteristics, etc.), and can be represented in the form of a data table or graph.

[0076] The regional floating strategy is to adjust the region according to the pre-set regional floating standards and amplitude adjustment principles. The regional floating standards include the differences in carbon emission intensity between regions, the matching degree of peak and valley periods of electricity load, and the complementarity of distributed power generation; the amplitude adjustment principle is to pre-set the amplitude of dynamic partitioning according to actual conditions.

[0077] In some embodiments, the regional floating standard includes the difference in carbon emission intensity between regions, the matching degree of peak and valley periods of electricity load, and the complementarity of distributed power generation. Specifically, a carbon emission intensity difference threshold (such as 10%) is set based on historical data. When the carbon emission intensity difference between regions exceeds this threshold, dynamic adjustment is triggered to clarify the division standard, avoid excessive or insufficient division, limit the difference in carbon emission intensity, and promote carbon emission balance between regions; based on the grid operation data and user needs, a mismatch range of peak and valley periods of electricity load is set (such as 20%-30%). When the mismatch degree of peak and valley periods of electricity load between regions exceeds this range, dynamic adjustment is triggered. Setting the mismatch range helps to balance the needs of stable grid operation and power resource utilization, and avoid power waste and grid instability caused by peak and valley period misalignment; based on the distributed power generation characteristics and grid demand, a complementarity scoring threshold is set, and a comprehensive evaluation model is pre-set to calculate. When the score of distributed power generation complementarity between regions is lower than this threshold, dynamic adjustment is triggered. The scoring threshold is set to quantify the complementarity of distributed power generation, ensuring that the divided areas have better power generation complementarity and improve reliability and flexibility.

[0078] In some embodiments, the amplitude adjustment principle is a pre-set amplitude range of dynamic partitions. According to the detection constraints of the regional floating standards, the corresponding amplitude adjustment is selected. After any of the above standards triggers dynamic adjustment, dynamic regional adjustment will be implemented. The corresponding adjustment amplitude is selected according to the number of triggering conditions and the size of the excess standard. If there is one triggering condition and the size of the excess standard is low (such as 10%-30%), a slight adjustment amplitude (such as 20%) is selected. The specific correspondence between the standard size, the number of conditions and the amplitude is pre-set according to actual conditions.

[0079] In some embodiments, the clustering results and the preset floating standards are used to evaluate the current regional divisions, and the regions that need to be adjusted are identified, that is, regions that have significant differences from other regions in certain features or do not meet the preset standards. According to the amplitude adjustment principle, the adjustment amplitude for each region that needs to be adjusted is determined, and according to the determined adjustment amplitude, the regional division is actually adjusted, and the adjusted regional division is verified using the preset constraints. The constraints are the multiple standard thresholds set above to ensure balance in each feature. Analyze the distribution of feature values ​​of each region in the historical data, determine a reasonable adjustment amplitude range, and set an adjustment amplitude that conforms to the actual situation.

[0080] In this embodiment, by introducing a regional floating strategy, dynamic adjustment of regional division is achieved according to real-time changes in the region, making the regional division more flexible and adaptable; by setting standards such as the carbon emission intensity difference threshold, the peak and valley period mismatch range of electricity load, and the distributed power generation complementarity score threshold, and dynamically adjusting the region based on these standards, the balance of the region in various characteristics is optimized; by dynamically adjusting the regional division, the stable operation of the power grid and the utilization demand of power resources are balanced.

[0081] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0082] This application achieves the effect of dynamic adjustment of regional divisions according to real-time changes in regions by introducing a regional floating strategy; by setting standards such as carbon emission intensity difference thresholds, electricity load peak and valley period mismatch ranges, and distributed power generation complementarity score thresholds, and dynamically adjusting regions based on these standards, the effect of dynamic adjustment and optimization of regional divisions is achieved.

[0083] Embodiment 4: This embodiment makes further improvements on the basis of the above embodiment.

[0084] S400: Establishing a regional carbon emission model for each region according to the regional independence strategy, preliminarily calculating the user-side carbon emissions of each partition in the next time period according to the regional carbon emission model, and evaluating and optimizing the preliminary calculation results according to the actual carbon emissions.

[0085] In some embodiments, the regional carbon emission model is established separately according to the results of the division of each region, and the specific usage principles and methods of the model are adaptively adjusted according to the dynamic adjustment of the region. According to the regional characteristics and data characteristics, a suitable carbon emission calculation model is selected, such as a carbon emission model based on energy consumption, a carbon emission model based on user behavior, etc.; the model parameters are calibrated and optimized according to historical data and real-time monitoring data to ensure the accuracy and reliability of the model; the carbon emission models of each sub-region are integrated into a unified carbon emission monitoring and management system to realize real-time monitoring, analysis and display of data.

[0086] In some embodiments, the regional carbon emission model is established based on the results of each regional division combined with regional characteristics and data characteristics, and the user-side carbon emissions for the next time period are obtained according to the set evaluation criteria of each regional carbon emission model. The evaluation criteria pre-set in the regional carbon emission model are to conduct statistical analysis based on historical carbon emission data, compare the current carbon emissions with historical data, analyze year-on-year and month-on-month changes, evaluate carbon emission trends, and then generate new user-side carbon emissions.

[0087] According to the regional carbon emission model, the user-side carbon emissions can be directly obtained, which can effectively reduce the efficiency of real-time monitoring on the user side. The error value can be generated by comparing the preliminary calculated user-side carbon emissions with the actual user-side carbon emissions, and the regional carbon emission model can be further optimized. The regional carbon emission model still uses the long short-term memory network model as the basic model, and is trained, verified, and tested based on activity data and selected carbon emission factors. The activity data refers to the dynamic change data related to regional carbon emissions, including energy consumption data, user behavior data, meteorological data, distributed power generation data, etc. Energy consumption data refers to the energy consumption of different energy sources in different time periods in the region and the corresponding energy consumption types. User behavior data refers to data such as users' daily activities and electricity usage habits.

[0088] In this embodiment, by establishing a regional carbon emission model, the user-side carbon emissions for each partition in the next time period are preliminarily calculated, and the initial carbon emissions are directly obtained by combining historical data, which saves tedious steps such as real-time monitoring and analysis and improves detection efficiency. The regional carbon emission model still uses the long short-term memory network model as the basic model, and is trained, verified and tested according to activity data and selected carbon emission factors to further improve the prediction accuracy of carbon emissions.

[0089] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0090] This application achieves the effects of improving the prediction accuracy of carbon emissions, saving real-time monitoring and analysis steps, and optimizing regional carbon emissions management by adopting a method of establishing a regional carbon emissions model.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting carbon emissions of an electricity user, characterized in that: The method comprises: S100: Real-time monitoring to obtain the real-time net power generation and power consumption pulse number on the user side, and establish a dynamic estimation model of carbon emission factors based on historical data and real-time meteorological data, update the carbon emission factors in real time, and generate a carbon emission analysis report; S200: Obtain the carbon emission analysis report, establish a carbon emission prediction model, identify the change trend of carbon emissions and high emission periods, establish a differentiated power load scheduling optimization model, and generate a power load scheduling plan; S300: setting a user area division rule according to the power load dispatching scheme, dividing the area according to the user's carbon emission behavior, and setting a regional independence strategy; S400: Establishing a regional carbon emission model for each region according to the regional independence strategy, preliminarily calculating the user-side carbon emissions of each partition in the next time period according to the regional carbon emission model, and evaluating and optimizing the preliminary calculation results according to the actual carbon emissions; Among them, the calculation formula for user-side carbon emissions is: in, represents the carbon emissions on the user side, represents the user's electricity consumption in the i-th time period, represents the average carbon emission factor of the user side in the i-th time period, represents the net power generation of distributed generation in the jth time period, represents the carbon emission factor of distributed generation in the jth time period, represents the length of the jth time period, The number of time periods representing electricity consumption, The number of time periods representing the net generated power; The dynamic estimation model selects a linear regression model as a basic model, takes historical data and converted real-time meteorological data as input, outputs a carbon emission factor, and uses cross-validation and calculation of mean square error to evaluate the output effect; The specific calculation formula of the average carbon emission factor is as follows: in, represents the average carbon emission factor on the user side, represents the power generation of the kth distributed generation in a specific time period, represents the carbon emission factor corresponding to the kth distributed generation, Indicates the number of types of distributed power sources; The user area division rule is to use the clustering algorithm to partition the users by extracting the features reflecting the user's electricity consumption behavior and the distributed power generation behavior, and divide the users into different behavior areas according to the clustering results; the regional independence strategy is to set a difference independence strategy according to different regions, and formulate different identification standards for each behavior area; The user area division rule also includes adjusting the area division result according to the clustering result according to the pre-set area floating strategy, and determining whether the constraint condition is met to determine the final area division result; The regional floating strategy is to adjust the region according to the pre-set regional floating standards and amplitude adjustment principles. The regional floating standards include the differences in carbon emission intensity between regions, the matching degree of peak and valley periods of electricity load, and the complementarity of distributed power generation; the amplitude adjustment principle is to pre-set the amplitude of dynamic partitioning according to actual conditions.

2. A method for detecting carbon emissions of an electricity user as claimed in claim 1, characterized in that: The real-time meteorological data is converted into characteristic data related to carbon emissions according to conversion rules, and different conversion rules are set according to different distributed power generation characteristics for separate conversions.

3. A method for detecting carbon emissions of an electricity user as claimed in claim 1, characterized in that: Feature parameters are extracted according to the carbon emission analysis report and historical data, feature parameters are selected using a recursive feature elimination method, and a long short-term memory network model is selected according to the feature parameters to establish a carbon emission prediction model.

4. A method for detecting carbon emissions of an electricity user as claimed in claim 1, characterized in that: The regional carbon emission model is established according to the result of each regional division combined with regional characteristics and data characteristics, and the user-side carbon emissions in the next time period are obtained according to the set evaluation criteria based on each regional carbon emission model.

5. A method for detecting carbon emissions of an electricity user as claimed in claim 4, characterized in that: The pre-set evaluation standard of the regional carbon emission model is to conduct statistical analysis based on historical carbon emission data, compare current carbon emissions with historical data, analyze year-on-year and month-on-month changes, evaluate carbon emission trends, and then generate new user-side carbon emissions.

Images