Method for predicting carbon emission in power grid supply chain

Abstract

The invention provides a method for predicting carbon emission in a power grid supply chain. The method comprises the steps of 1, predicting the carbon emission of a power supply link of the power grid supply chain in a target region; 2, predicting the carbon emission of a management link in a power grid supply chain of the target region; and step 3, according to the carbon emission of the power supply link and the carbon emission of the management link, predicting the carbon emission of the target area in the power grid supply chain in the future target time period. According to the carbon emission prediction method for the power grid supply chain provided by the invention, the power grid supply chain is divided into prediction combination of a power supply link and a management link, the two pieces of data are integrated, the final carbon emission of the power grid supply chain is determined, and the prediction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of carbon emission prediction, and in particular to a method for predicting carbon emission in a power grid supply chain. Background Art

[0002] Global climate change has become the focus of international attention, and an efficient and accurate carbon emission prediction method is urgently needed.

[0003] At present, scholars have conducted research on carbon emission prediction and achieved fruitful results.

[0004] Green supply chain construction has become the focus of the power grid supply chain. Research on carbon emission accounting methods in the power grid supply chain management link will help to understand the carbon emission intensity in the warehousing link, provide data support for power grid supply chain logistics carbon reduction decisions, and assist the green and low-carbon transformation of the power grid supply chain.

[0005] At present, power grid companies have been applying some carbon emission forecasting schemes, but the current power grid carbon emission forecasts are aimed at predicting the carbon emissions of power grid companies in the medium and long term. In order to formulate reasonable low-carbon development strategies for power grid companies in the long term and achieve sustainable development, there is still a lack of solutions for carbon emission accounting and forecasting in the power grid supply chain. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for predicting carbon emissions in a power grid supply chain, which can calculate and predict carbon emissions in a power grid supply chain.

[0007] A carbon emission prediction method in a power grid supply chain, comprising:

[0008] Step 1: Predict the carbon emissions of the power supply link in the target area power grid supply chain;

[0009] Step 2: Predict the carbon emissions of the management links in the target area power grid supply chain;

[0010] Step 3: Based on the carbon emissions in the power supply link and the carbon emissions in the management link, predict the carbon emissions in the power grid supply chain in the target area during the future target time period.

[0011] Furthermore, step one specifically includes:

[0012] Step 1.1: Collect historical overall carbon emissions data of the target area power grid supply chain;

[0013] Step 1.2: Based on the historical carbon emission data of the target area power grid supply chain, predict the carbon emission of the power supply link of the target area power grid supply chain.

[0014] Furthermore, step 1.1 specifically includes:

[0015] Step 1.1.1: Identify the various links in the power grid supply chain

[0016] Power generation: including thermal power generation, hydropower generation, wind power generation, and solar power generation;

[0017] Transmission link: including high-voltage transmission lines and substations;

[0018] Power distribution link: including low-voltage distribution lines and distribution equipment;

[0019] Step 1.1.2: Collect historical data on the target area power grid supply chain

[0020] Collect energy consumption data from all links of the target area’s power grid supply chain;

[0021] Collect carbon emission factors of equipment, technologies and materials used in each link of the target area power grid supply chain;

[0022] Step 1.1.3: Calculate the historical overall carbon emissions of the target area power grid supply chain:

[0023] Calculate the historical overall carbon emissions of each link based on energy consumption data and carbon emission factors;

[0024] Carbon emissions calculation formula: Pi = Ei * Fi

[0025] Among them, Pi represents the carbon emission of the i-th link, i represents the link of the power grid supply chain, Ei is the energy consumption of the i-th link, and Fi is the carbon emission factor of the i-th link;

[0026] Step 1.1.4: Summarize the carbon emissions of each link in the target area power supply chain to obtain the historical carbon emissions data of the power supply chain:

[0027] P=PA+PB+PC

[0028] PA represents the carbon emissions in the power generation link, PB represents the carbon emissions in the transmission link, and PC represents the carbon emissions in the distribution link.

[0029] Furthermore, step 1.2 specifically includes:

[0030] Step 1.2.1: Based on the historical carbon emission data of the target area power grid supply chain, preliminarily predict the carbon emission of the power supply link of the target area power grid supply chain in the future;

[0031] Step 1.2.2, determine the current power transformation development state, and obtain the parameters of the current power transformation development state;

[0032] Step 1.2.3: Combine the parameters of the power transformation development status and the carbon emissions of the power supply link initially predicted in step 1.2.1 to obtain the final carbon emissions of the power supply link of the target area power grid supply chain.

[0033] Furthermore, in step 1.2.1, the network model is trained based on the historical carbon emission data of the power grid supply chain in the target area, and the overall carbon emissions of the power grid supply chain in the target area in a certain time period in the future are predicted by collecting the carbon emission data of the target area at a certain time at the current time.

[0034] Furthermore, in step 1.2.1, the overall carbon emissions of the power grid supply chain in the target area are predicted in a certain period of time in the future. The specific prediction method is as follows:

[0035] (1) Data Collection

[0036] Obtain daily / monthly / yearly carbon emission data from the target area power grid supply chain over the past few years. The data includes date, time, and carbon emissions at each stage.

[0037] (2) Data preprocessing

[0038] Clean the collected data to remove missing values ​​and outliers;

[0039] Create new feature data, including seasonal characteristics, holiday markers, and energy consumption type proportions;

[0040] Standardize or normalize different types of data to ensure that data of different dimensions can be compared on the same scale, thus improving model training efficiency;

[0041] (3) Training Model

[0042] Considering the needs of time series prediction, we choose the long short-term memory network LSTM, gated recurrent unit GRU or Transformer neural network model.

[0043] The specific steps of training the model are as follows:

[0044] Divide the data into training set, validation set and test set, where 70% of the data is used for training, 15% for validation, and 15% for testing;

[0045] Determine learning rate, batch size, training rounds, optimizer parameters;

[0046] Train the model using the training set data and evaluate its performance on the validation set;

[0047] Adjust model structure or parameters as needed to improve prediction accuracy;

[0048] Use the test set data to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE).

[0049] Check the model’s forecasting performance over different time periods;

[0050] Adjust the model structure or parameters based on the evaluation results, including increasing the number of hidden layers, adjusting the number of neurons, and trying different activation functions;

[0051] (4) Prediction

[0052] Use the trained model, input the current carbon emission data and other influencing factors, and predict the overall carbon emissions of the power grid supply chain in a certain period of time in the future.

[0053] Furthermore, in step 1.2.2, considering the changes in power generation structure and the continuous changes in the development level of coal-fired power technology, three different power transformation development scenarios are set, namely, the baseline scenario, the fast transformation scenario, and the slow transformation scenario; the average annual change rate of the proportion of power generation of each power generation mode in the baseline scenario is set according to the average annual change rate within its preset period, the slow transformation scenario slows down the speed of transformation from thermal power to clean energy power generation on the basis of the baseline scenario, and the fast transformation scenario accelerates the speed of transformation from thermal power to clean power. Fixed parameters λ are set for the three different power transformation development scenarios.

[0054] Furthermore, step 1.2.3 uses the following formula to calculate the carbon emissions of the power supply link in the target area power grid supply chain:

[0055] P2=P1*λ

[0056] Among them, P2 is the carbon emissions of the power supply link of the final future target regional power grid supply chain, P1 is the carbon emissions of the power supply link initially predicted in step 1.2.1, and λ is a set fixed parameter;

[0057] Among them, the parameter λ in the baseline scenario is set to 1 based on the overall carbon emissions calculated in step 1.1.4, indicating that it is the same as in step 1.1.4. The parameter λ in the fast transition scenario is set to 0.8, indicating that the transformation speed from thermal power to clean power is accelerated, and its carbon emissions need to be converted based on conventional forecasts. In theory, the carbon emissions are less than the baseline scenario. The slow transition scenario is to slow down the transformation speed from thermal power to clean energy power generation based on the baseline scenario. In theory, the carbon emissions are more than the baseline scenario, and the parameter λ is set to 1.2.

[0058] Furthermore, step 2 specifically includes:

[0059] Step 2.1, collect historical carbon emission data of the storage link in the power grid supply chain;

[0060] Industrial transport vehicles are used for loading and unloading of goods in the receiving, inventory and shipping operations in the warehousing process. For the above-mentioned operations, industrial transport vehicles are energy-consuming equipment, and the carbon emissions calculation method is to measure the actual electricity or fuel consumption of industrial transport vehicles for each operation. Among them: the electricity / diesel emission factor is obtained by citing the electricity and diesel carbon emission factors of industrial transport vehicles released by the Ministry of Ecology and Environment to calculate the carbon emissions of industrial transport vehicles; or the electricity and diesel carbon emission factors are estimated according to the model, working conditions and power source of the forklift; if none of the above methods can be found, it is set by experimental measurement or expert estimation;

[0061] The carbon emission calculation of industrial handling vehicles is to calculate the product of the power consumption and the electric emission factor of the electric industrial handling vehicle for a certain period of time, and the sum of the product of the fuel consumption and the diesel emission factor of the diesel industrial handling vehicle. The specific carbon emission calculation formula is as follows:

[0062] Carbon emissions from warehousing in the power grid supply chain = carbon emissions from receiving operations + carbon emissions from tallying + carbon emissions from shipping operations + carbon emissions from storage operations + carbon emissions from the operation process

[0063] in:

[0064] Carbon emissions from receiving operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 1

[0065] Carbon emissions from shipping operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 2

[0066] Carbon emissions from tallying = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 3

[0067] Storage operation carbon emissions = industrial handling vehicle energy consumption data × electricity or oil emission factor + crane power consumption × electricity emission factor 4

[0068] Among them, the energy consumption data of industrial handling vehicles refers to the power consumption / fuel consumption of industrial handling vehicles. By collecting the power or fuel consumption and business volume data of a single operation of industrial handling vehicles, the power and fuel consumption of industrial handling vehicles in a certain period of time is calculated; by collecting the operation time and business volume data of a crane loading and unloading operation, the power consumption of the crane in a certain period of time is calculated; by collecting the operation time and business volume data of automatic shelves loading and unloading shelves, the power consumption of the crane in a certain period of time is calculated;

[0069] Step 2.2: Collect historical carbon emission data of the operation links in the power grid supply chain

[0070] Carbon emissions from operations = public electricity × electricity emission factor 5 + energy consumption of purchased products and services × electricity emission factor 6 per unit of goods or services

[0071] Step 2.3, based on the carbon emission data of the storage link and the carbon emission data of the operation link in the power grid supply chain, calculate the historical carbon emission data of the management link of the power grid supply chain;

[0072] Step 2.4, based on the historical carbon emission data of the management link of the power grid supply chain, predict the carbon emission data of the management link of the power grid supply chain in a certain time period in the future;

[0073] The specific prediction method is as follows:

[0074] (1) Data Collection

[0075] Obtain daily / monthly / yearly carbon emission data for the past few years from the historical carbon emission data of the target area power grid supply chain management link. The data includes date, time, and carbon emission amount of each link.

[0076] (2) Data preprocessing

[0077] Clean the collected data to remove missing values ​​and outliers;

[0078] Create new feature data, including seasonal characteristics, holiday markers, and energy consumption type proportions;

[0079] Standardize or normalize different types of data to ensure that data of different dimensions can be compared on the same scale, thus improving model training efficiency;

[0080] (3) Training Model

[0081] Considering the needs of time series prediction, choose the long short-term memory network LSTM, gated recurrent unit GRU or Transformer neural network model;

[0082] The specific steps of training the model are as follows:

[0083] Divide the data into training set, validation set and test set, where 70% of the data is used for training, 15% for validation, and 15% for testing;

[0084] Determine learning rate, batch size, training rounds, optimizer parameters;

[0085] Train the model using the training set data and evaluate its performance on the validation set;

[0086] Adjust model structure or parameters as needed to improve prediction accuracy;

[0087] Use the test set data to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE).

[0088] Check the model’s forecasting performance over different time periods;

[0089] Adjust the model structure or parameters based on the evaluation results, including increasing the number of hidden layers, adjusting the number of neurons, and trying different activation functions;

[0090] (4) Prediction

[0091] Use the trained model, input the current carbon emission data and other influencing factors, and predict the carbon emissions of the power grid supply chain management link in a certain period of time in the future.

[0092] Furthermore, step three specifically includes:

[0093] The carbon emissions of the power supply link are set as E_supply, the carbon emissions of the management link are set as E_management, and the weighted coefficients of the power supply link and the management link are set as W_supply and W_management respectively, which are used to reflect the relative importance of the power supply link and the management link in the carbon emissions of the power grid supply chain. The sum of the weighted coefficients is 1;

[0094] The total carbon emissions E_total in the power grid supply chain are calculated using the following formula:

[0095] E_total=W_supply×E_supply+W_management×E_management

[0096] in:

[0097] E_total represents the total carbon emissions in the predicted power grid supply chain;

[0098] W_supply represents the weighted coefficient of the power supply link;

[0099] E_supply represents the predicted carbon emissions from the electricity supply link;

[0100] W_management represents the weighted coefficient of the management link;

[0101] E_management represents the predicted carbon emissions in the management process.

[0102] The present invention has the following beneficial effects:

[0103] 1. Provide a carbon emission prediction method for the power grid supply chain. By splitting the power grid supply chain into the power supply link and the management link prediction, the two data are integrated to determine the final carbon emission of the power grid supply chain and improve the prediction accuracy;

[0104] 2. When predicting carbon emissions in the power supply chain, the specific calculation method of historical data of the power grid supply chain is more detailed, making the overall prediction more accurate in the later stage;

[0105] 3. When predicting carbon emissions in the power supply sector, the current state of power transformation and development is taken into account, making the prediction more accurate;

[0106] 4. In the forecasting management stage, the division is more detailed, including the calculation of carbon emissions such as public electricity consumption and electricity consumption for purchased products or services. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] Figure 1 is a schematic diagram of carbon emissions in various management links in the power grid supply chain in an embodiment of the present invention;

[0108] Figure 2 It is a method for calculating the carbon emissions of energy-consuming equipment in each management link of the power grid supply chain. DETAILED DESCRIPTION

[0109] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0110] See also Figure 1 The embodiment of the present invention provides a method for predicting carbon emissions in a power grid supply chain, comprising the following steps:

[0111] Step 1: Predict the carbon emissions of the power supply link in the target area power grid supply chain, including the following steps:

[0112] Step 1.1: Count the historical overall carbon emissions data of the target area power grid supply chain. Step 1.1 specifically includes:

[0113] Step 1.1.1: Identify the various links in the power grid supply chain

[0114] Power generation link: including thermal power generation, hydropower generation, wind power generation, solar power generation, etc.

[0115] Transmission link: including high-voltage transmission lines, substations, etc.

[0116] Power distribution link: including low-voltage distribution lines, distribution equipment, etc.

[0117] Step 1.1.2: Collect historical data on the target area power grid supply chain

[0118] Collect energy consumption data (such as coal, oil, natural gas, electricity, etc.) at all links of the power grid supply chain in the target area.

[0119] Collect carbon emission factors of equipment, technologies and materials used in each link of the power grid supply chain in the target area.

[0120] For the power generation link, the energy consumed by power generation can be directly calculated, for example, the energy consumed E1 (coal) is 1 million tons. For the carbon emission factor, the parameters can be set directly, for example, the carbon emission factor of coal is 2.5 tons CO2 / ton of coal, hydropower consumes water resources but has no carbon emissions, wind power and solar power have no carbon emissions, and the carbon emission factor is 0.

[0121] For the transmission link, first of all, it is necessary to identify the equipment used in the transmission link, such as high-voltage transmission lines, substations, etc. Collect the energy consumption data of these devices, which can usually be obtained from the equipment's operation records, energy efficiency reports, or data provided by the manufacturer. Add up the energy consumption of all equipment to get the total energy consumption of the transmission link.

[0122] For example, X = X1 = X2 + ... + Xn, where n is the number of devices used in the transmission link. Assume that there are two high-voltage transmission lines and a substation in the transmission link, and their energy consumption is 50,000 kWh, 30,000 kWh, and 20,000 kWh respectively. Then the total energy consumption of the transmission link is:

[0123] Total energy consumption in the transmission link = 50,000 kWh + 30,000 kWh + 20,000 kWh = 100,000 kWh. For the transmission link, the carbon emission factor is usually related to the type of energy used (such as electricity) and the efficiency of energy transmission. This factor can be obtained from energy suppliers, research institutions or government reports and set as a parameter.

[0124] For the distribution link, similar to the transmission link, first determine the equipment used in the distribution link, such as low-voltage distribution lines, distribution transformers, etc., collect the energy consumption data of these equipment, add up the energy consumption of all equipment, and get the total energy consumption of the distribution link. Similar to the transmission link, the total energy consumption of the distribution link is also calculated by adding up the energy consumption of all equipment. The carbon emission factor of the distribution link is also usually related to the type of energy used and the energy transmission efficiency, and can be obtained from the same source.

[0125] Assume that there are 10 low-voltage distribution lines and one distribution transformer in the distribution link, and their energy consumption is 5,000 kWh, 3,000 kWh and 2,000 kWh respectively. Then the total energy consumption of the distribution link is:

[0126] Total energy consumption in the power distribution link = 10*5,000 kWh + 3,000 kWh + 2,000 kWh = 55,000 kWh

[0127] Step 1.1.3: Calculate the historical overall carbon emissions of the target area power grid supply chain:

[0128] Based on energy consumption data and carbon emission factors, the historical overall carbon emissions of each link are calculated.

[0129] Carbon emissions calculation formula:

[0130] Pi=Ei*Fi

[0131] Among them, A: power generation link, B: transmission link, C: distribution link. Pi represents the carbon emissions of the i-th link, i represents the link of the power grid supply chain (such as A, B, C), Ei is the energy consumption of the i-th link, and Fi is the carbon emission factor of the i-th link.

[0132] Step 1.1.4: Summarize the carbon emissions of each link in the target area power supply chain to obtain the historical carbon emissions data of the power supply chain.

[0133] The carbon emissions of each link in the target area power grid supply chain are aggregated to obtain the overall carbon emissions of the power grid supply chain.

[0134] P=PA+PB+PC

[0135] PA represents the carbon emissions in the power generation link, PB represents the carbon emissions in the transmission link, and PC represents the carbon emissions in the distribution link.

[0136] Step 1.2: Based on the historical carbon emissions data of the target area power grid supply chain, predict the carbon emissions of the power supply link of the target area power grid supply chain

[0137] Step 1.2 specifically includes:

[0138] Step 1.2.1: Based on the historical carbon emissions data of the target area power grid supply chain, a preliminary forecast of the carbon emissions of the power supply link of the target area power grid supply chain in the future is made.

[0139] According to the historical carbon emission data of the power grid supply chain in the target area, the network model is trained, and the overall carbon emission of the power grid supply chain in the target area in a certain time period in the future is predicted by collecting the carbon emission data of the target area at a certain time at the current time. The network model can be various current models, such as CNN model, long short-term memory network (LSTM) or Transformer neural network model.

[0140] The specific prediction method is as follows:

[0141] (1) Data Collection

[0142] Obtain daily / monthly / yearly carbon emission data for the past few years from the historical carbon emission data of the target area power grid supply chain. The data should include date, time, carbon emissions at each link, etc.

[0143] (2) Data preprocessing

[0144] The collected data is cleaned to remove missing values ​​and outliers (such as abnormally high or low emissions caused by extreme weather).

[0145] Create new feature data, such as seasonal characteristics (month, quarter), holiday marks, energy consumption type proportion, etc.

[0146] Standardize / normalize different types of data to ensure that data of different dimensions can be compared on the same scale and improve model training efficiency.

[0147] (3) Training Model

[0148] Considering the needs of time series prediction, neural network models such as long short-term memory network (LSTM), gated recurrent unit (GRU) or Transformer can be selected, which are good at processing time dependencies in sequence data.

[0149] Taking the long short-term memory network (LSTM) as an example, the model structure is constructed as follows:

[0150] Input layer: receives preprocessed time series data (such as carbon emission data for the past N days and other influencing factors).

[0151] Hidden layer: Contains LSTM / GRU / Transformer layers, which are used to learn temporal dependencies and feature relationships in the data.

[0152] Output layer: predict the overall carbon emissions of the power grid supply chain in the next M days.

[0153] The specific steps of training the model are as follows:

[0154] Divide the dataset:

[0155] The data is divided into training set, validation set and test set, where 70% of the data is used for training, 15% for validation and 15% for testing.

[0156] Setting training parameters

[0157] Determine parameters such as learning rate, batch size, training rounds, optimizer, etc.

[0158] Training the model

[0159] The model is trained using the training set data and its performance is evaluated on the validation set.

[0160] Adjust the model structure or parameters as needed to improve prediction accuracy.

[0161] Model evaluation and optimization

[0162] The test set data is used to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), etc.

[0163] Check the model's prediction performance in different time periods (such as holidays, different seasons).

[0164] Adjust the model structure or parameters based on the evaluation results, such as increasing the number of hidden layers, adjusting the number of neurons, trying different activation functions, etc.

[0165] In the embodiments of the present invention, an integrated learning method (such as model averaging and stacking generalization) can be introduced to further improve the prediction accuracy.

[0166] (4) Prediction

[0167] Use the trained model, input the current carbon emission data and other influencing factors, and predict the overall carbon emissions of the power grid supply chain in a certain time period in the future (such as the next week or month).

[0168] Suppose we collected daily carbon emission data of a power grid supply chain for the past five years, as well as weather data and energy consumption data for the same period. After preprocessing, we used the LSTM model for training and adjusted parameters such as the learning rate and the number of hidden layer units to achieve a low RMSE value on the test set. Then, we used the model to predict daily carbon emissions for the next month.

[0169] Step 1.2.2: Determine the current state of power transformation and development, and obtain the parameters of the current state of power transformation and development.

[0170] Taking into account the changes in power generation structure and the level of development of coal-fired power technology as society develops, three different power transformation development scenarios are set, namely the baseline scenario, the fast transformation scenario, and the slow transformation scenario, as shown in Table 1:

[0171] Table 1 Correspondence table of power transformation development status and its corresponding parameters

[0172] Power Transformation Development Status Parameter λ Baseline scenario 1 Fast Transformation Scenario 0.8 Slow transition scenario 1.2

[0173] The average annual change rate of the proportion of power generation of each power generation mode under the baseline scenario is set according to the average annual change rate within the preset period (such as 2015-2021). The slow transformation scenario slows down the transformation of thermal power to clean energy power generation on the basis of the baseline scenario, while the fast transformation scenario accelerates the transformation of thermal power to clean electricity.

[0174] Step 1.2.3: Combine the parameters of the power transformation development status and the carbon emissions of the power supply link initially predicted in step 2.1 to obtain the final carbon emissions of the power supply link of the target area power grid supply chain.

[0175] After determining the parameters of the power transformation development state in step 1.2.2, the parameters of the power transformation development state and the preliminary predicted overall carbon emissions of the future target area power grid supply chain are combined to obtain the final carbon emissions of the power supply link of the future target area power grid supply chain. Specifically, the following formula can be used:

[0176] P2=P1*λ

[0177] Among them, P2 is the carbon emissions of the power supply link of the final future target regional power grid supply chain, P1 is the carbon emissions of the power supply link initially predicted in step 1.2.1, and λ is a set fixed parameter.

[0178] Among them, under the baseline scenario, the parameter λ is set to 1 based on the overall carbon emissions calculated in step 1.1.4, indicating that it is the same as in step 1.1.4. Under the fast transition scenario, the parameter λ is set to 0.8, indicating that the transformation speed of thermal power to clean power is accelerated, and its carbon emissions need to be converted based on conventional forecasts. In theory, the carbon emissions are less than the baseline scenario; the slow transition scenario is to slow down the transformation speed of thermal power to clean energy power generation based on the baseline scenario. In theory, the carbon emissions are more than the baseline scenario, and the parameter λ is set to 1.2. Of course, the above parameters are examples and can be fine-tuned according to actual conditions.

[0179] Step 2: Predict the carbon emissions of each management link in the target area power grid supply chain.

[0180] The present invention can also predict the carbon emissions of each management link in the target area power supply chain, and can predict the carbon emissions of each management link in the target area power supply chain based on the carbon emission factor accounting method.

[0181] Through field research on a power grid supply chain warehouse, we identified the primary, secondary, and tertiary energy emissions in all links of the power grid supply chain, systematically understood the energy-consuming links, energy-consuming equipment, and energy types in the warehousing business activities in the power grid supply chain, clarified the boundaries of warehousing emissions accounting in the power grid supply chain, and supported the research on carbon emission calculation methods. Through warehouse research, we understand the current energy consumption status and existing problems in the power grid supply chain. Based on the boundaries of power grid supply chain management, we refine and sort out the energy-consuming equipment and corresponding energy consumption types in each operation link, and then identify the carbon emission sources in each link. The carbon emissions of each management link in the power grid supply chain are as follows: Figure 1 shown.

[0182] By calculating the carbon emissions of each operation link in the power grid supply chain, accumulating the emissions of all links, the total carbon emissions in the power grid supply chain are obtained, and the data collection requirements are clarified according to the calculation rules to form a complete power grid supply chain carbon emissions calculation method.

[0183] Since the warehousing link is the most important link in the warehousing operation in the power grid supply chain, industrial transport vehicles (such as forklifts) are the most frequently used equipment. Therefore, the electricity or fuel consumption of industrial transport vehicles is used as an indicative evaluation parameter for warehousing operations in the power grid supply chain. In the embodiment of the present invention, the energy-consuming equipment in the power grid supply chain is divided into industrial transport vehicles and other operating equipment, and the carbon emissions of industrial transport vehicles in different operating links (as carbon emissions data in the warehousing link) and the carbon emissions of other operating equipment (as carbon emissions data in the operating link) are counted separately.

[0184] Among them, the operation links of industrial handling vehicles include receiving, sorting, shipping, and storage, while the carbon emissions of other operating equipment are uniformly classified as part of the operation management link.

[0185] Step 2 specifically includes:

[0186] Step 2.1: Collect historical carbon emission data of the storage link in the power grid supply chain

[0187] The specific methods are as follows: Industrial handling vehicles are used for loading and unloading of goods in the receiving, inventory, and shipping operations.

[0188] like Figure 2 As shown, for the above-mentioned operation links, industrial handling vehicles are energy-consuming equipment, and the carbon emission calculation method can currently be achieved by measuring the actual electricity or oil consumption of industrial handling vehicles for each operation. Among them: the electricity / diesel emission factor can be obtained by citing the electricity and diesel carbon emission factor data of industrial handling vehicles released by the Ministry of Ecology and Environment and other institutions to calculate the carbon emissions of industrial handling vehicles. For example, taking industrial handling vehicles as forklifts as an example, this information can be obtained through official data released by the government or research reports of relevant research institutions, or by referring to data provided by forklift manufacturers. In addition, some professional carbon emission calculators can estimate the electricity / diesel emission factor based on the forklift model, working conditions, and power source. If none of the above methods can be found, it can also be set through experimental measurement or expert estimation.

[0189] The carbon emission calculation of industrial handling vehicles is to calculate the product of the power consumption and the electric emission factor of electric industrial handling vehicles for a certain period of time, and the sum of the product of the fuel consumption and the diesel emission factor of diesel industrial handling vehicles. The specific carbon emission calculation formula is as follows:

[0190] Carbon emissions from warehousing in the power grid supply chain = carbon emissions from receiving operations + carbon emissions from tallying + carbon emissions from shipping operations + carbon emissions from storage operations + carbon emissions from the operation process

[0191] in:

[0192] Carbon emissions from receiving operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 1

[0193] Carbon emissions from shipping operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 2

[0194] Carbon emissions from tallying = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 3

[0195] Storage operation carbon emissions = industrial handling vehicle energy consumption data × electricity or oil emission factor + crane power consumption × electricity emission factor 4

[0196] Among them, the energy consumption data of industrial transport vehicles refers to the electricity and fuel consumption of industrial transport vehicles (depending on whether they are electric equipment or oil-consuming equipment). The electricity and fuel consumption of industrial transport vehicles in a certain period of time can be calculated by collecting the electricity or fuel consumption and business volume data of a single operation of the industrial transport vehicles; the operation time and business volume data of a crane loading and unloading operation can be collected to calculate the electricity consumption of the crane in a certain period of time; the operation time and business volume data of automatic shelves loading and unloading operations can be collected to calculate the electricity consumption of the crane in a certain period of time.

[0197] If there are multiple industrial transport vehicles, the storage carbon emission data of multiple industrial transport vehicles will be calculated separately and then summarized. If multiple industrial transport vehicles involve different power types, such as electric forklifts and diesel forklifts, the corresponding electricity (diesel) emission factors will be the electric carbon emission factor and the diesel carbon emission factor respectively.

[0198] Generally, electricity emission factor 1, electricity emission factor 2, electricity emission factor 3, and electricity emission factor 4 can be set the same. However, in order to make the prediction more accurate, differentiated electricity emission factors can be set according to the overall workload in different operation links.

[0199] For example, during the receiving stage, forklifts are mainly used for unloading and handling operations, which are relatively labor-intensive, but the operating time is relatively concentrated. Since forklifts operate frequently and carry heavy loads during the receiving stage, it is assumed that the average carbon emission factor of the power grid during this stage is 0.6tCO2 / MWh (this value is a hypothetical value, and can actually be determined based on local power grid data and the proportion of renewable energy). Assuming that the forklifts consume a total of 100MWh of electricity during the receiving stage of a certain period, according to the carbon emission factor, the carbon emissions generated during this stage are 100MWh*0.6tCO2 / MWh=60tCO2.

[0200] During the tallying stage, forklifts are mainly used to sort and stack goods in the warehouse. The work intensity is moderate, but the operation path may be long and start and stop frequently. Considering the operation characteristics of forklifts during the tallying stage, it is assumed that the average carbon emission factor of the power grid during this stage is slightly lower than that during the receiving stage, which is 0.55tCO2 / MWh (also an assumed value). Assuming that the forklifts consume a total of 80MWh of electricity during the tallying stage, the carbon emissions are 80MWh*0.55tCO2 / MWh=44tCO2.

[0201] During the delivery phase, forklifts are mainly used for loading and unloading of goods. The workload is similar to that of the receiving phase, but may involve more types of goods and different loading requirements. Since the workload of the delivery phase is similar to that of the receiving phase, it is assumed that the average carbon emission factor of the power grid during this phase is also set to 0.6tCO2 / MWh. Assuming that the forklifts consume a total of 90MWh of electricity during the delivery phase, the carbon emissions are 90MWh*0.6tCO2 / MWh=54tCO2.

[0202] During the inventory phase, forklifts are mainly used for daily management and inventory of goods, with relatively low work intensity and scattered operation time. Considering the operation characteristics and low energy consumption requirements of forklifts during the inventory phase, it is assumed that the average carbon emission factor of the power grid during this phase is further reduced to 0.5tCO2 / MWh (this reflects the assumption that the proportion of renewable energy in the power grid may be higher). Assuming that the forklifts consume a total of 60MWh of electricity during the inventory phase, the carbon emissions are 60MWh*0.5tCO2 / MWh=30tCO2.

[0203] From the above examples, we can see that it is reasonable to set different carbon emission factors for forklift energy consumption at different operation stages, which helps to more accurately estimate and manage the carbon emissions of warehouse management in the power supply chain. It should be noted that the above carbon emission factors are all hypothetical values. In actual application, they can also be appropriately adjusted according to local power grid data, the proportion of renewable energy and the specific operation of the forklift. In addition, with the advancement of renewable energy technology and the optimization of the power grid structure, the carbon emission factor will also change, so it is necessary to regularly update the relevant carbon emission factor data to ensure the accuracy of the estimation results.

[0204] Step 2.2: Collect historical carbon emission data of the operation links in the power grid supply chain

[0205] Carbon emissions from operations = public electricity × electricity emission factor 5 + energy consumption of purchased products and services × electricity emission factor 6 per unit of goods or services

[0206] There are many public equipment involved in the operation process, which can be divided into public energy consumption and purchased energy consumption. For example, the electricity consumption of public equipment, such as printers, lighting equipment, office equipment, etc., and some energy consumption involving purchased products or services, can be set based on historical data. Electricity emission factor 5 and electricity emission factor 6 can be set. The setting of electricity emission factor can refer to the forklift method, for example, this information can be obtained through official data released by the government or research reports of relevant research institutions, or refer to the data provided by the manufacturer of the corresponding equipment (such as office equipment or purchased equipment). In addition, some professional carbon emission calculators can estimate its electricity / diesel emission factor based on the equipment model, working conditions and power source information. If the above methods cannot be found, it can also be set by experimental measurement or expert estimation. Specifically, for example, diesel emission factor, electricity emission factor: diesel emission factor refers to the diesel emission factor data under the National VI emission standard for diesel vehicles, and electricity emission factor refers to the electricity emission factor data of the "Electricity System Emission Factor Calculation Tool" under the United Nations Clean Development Mechanism.

[0207] Among them, energy-consuming equipment for public electricity can be collected in the following ways:

[0208] (1) Collect electricity consumption for warehouse lighting, office, etc., design a facility electricity collection plan, and collect warehouse electricity consumption to calculate carbon emissions data that cannot be allocated to specific operating links.

[0209] (2) Collect carbon emission data of office products such as paper, computers, and office furniture published by authoritative organizations;

[0210] Among them, the energy consumption calculation method for the quantity of purchased products and services is as follows;

[0211] Collect carbon emission data of outsourced services or products for calculating indirect carbon emission data.

[0212] Carbon emissions data for outsourced services or products can include the following, and the specific collection methods are as follows:

[0213] Energy consumption data: Collect information on the type, amount and source of energy consumed by suppliers of outsourced products or services during the production process.

[0214] Raw material data: Collect information on the sources of raw materials required for the production of purchased products, carbon emissions during the processing process, and waste treatment.

[0215] Transportation data: Collect information on the mode, distance, and carbon emissions of transportation of purchased products from suppliers to the company.

[0216] Collect other indirect emissions: such as the amount of purchased electricity and heat and their embodied carbon emissions.

[0217] Step 2.3: Calculate the historical carbon emission data of the management link of the power grid supply chain based on the statistical carbon emission data of the storage link and the carbon emission data of the operation link in the power grid supply chain.

[0218] The data in step 2.1 and step 2.2 can be directly added to obtain the carbon emission data of the management link of the power grid supply chain, or it can be calculated according to a certain weighting method.

[0219] The formula can be used:

[0220] Ptotal = αPm + βPn, where Pm is the carbon emission data of the warehousing link of the power grid supply chain, Pn is the historical carbon emission data of the management link of the power grid supply chain, α and β are pre-set weight factors, and the sum of α and β is 2. α and β can be set by experts in the field based on experience. With the passage of time and the continuous improvement of the prediction method, the accuracy of the prediction should be re-evaluated regularly, and the weight factors α and β should be updated accordingly. Statistical methods such as regression analysis and variance analysis can also be used to quantitatively analyze the relationship between the final predicted value Ptotal and the actual value to adjust the weight factors.

[0221] Step 2.4: Based on the historical carbon emission data of the management link of the power grid supply chain, the carbon emission data of the management link of the power grid supply chain in a certain time period in the future is predicted.

[0222] According to the historical carbon emission data of the power supply chain management link in the target area, the network model is trained, and the overall carbon emission of the power supply chain management link in the target area in a certain time period in the future is predicted by collecting the carbon emission data of the target area at a certain time at the current time. The network model can be various current models, such as CNN model, long short-term memory network (LSTM) or Transformer neural network model.

[0223] The specific prediction method is as follows:

[0224] (1) Data Collection

[0225] Obtain daily / monthly / yearly carbon emission data for the past few years from the historical carbon emission data of the target area power grid supply chain management link. The data should include date, time, carbon emissions of each link, etc.

[0226] (2) Data preprocessing

[0227] The collected data is cleaned to remove missing values ​​and outliers (such as abnormally high or low emissions caused by extreme weather).

[0228] Create new feature data, such as seasonal characteristics (month, quarter), holiday marks, energy consumption type proportion, etc.

[0229] Standardize / normalize different types of data to ensure that data of different dimensions can be compared on the same scale and improve model training efficiency.

[0230] (3) Training Model

[0231] Considering the needs of time series prediction, neural network models such as long short-term memory network (LSTM), gated recurrent unit (GRU) or Transformer can be selected, which are good at processing time dependencies in sequence data.

[0232] Taking the long short-term memory network (LSTM) as an example, the model structure is constructed as follows:

[0233] Input layer: receives preprocessed time series data (such as carbon emission data for the past N days and other influencing factors).

[0234] Hidden layer: Contains LSTM / GRU / Transformer layers, which are used to learn temporal dependencies and feature relationships in the data.

[0235] Output layer: predict the overall carbon emissions of the power grid supply chain in the next M days.

[0236] The specific steps of training the model are as follows:

[0237] Divide the dataset

[0238] The data is divided into training set, validation set and test set, where 70% of the data is used for training, 15% for validation and 15% for testing.

[0239] Setting training parameters

[0240] Determine parameters such as learning rate, batch size, training rounds, optimizer, etc.

[0241] Training the model

[0242] The model is trained using the training set data and its performance is evaluated on the validation set.

[0243] Adjust the model structure or parameters as needed to improve prediction accuracy.

[0244] Model evaluation and optimization

[0245] The test set data is used to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), etc.

[0246] Check the model's prediction performance in different time periods (such as holidays, different seasons).

[0247] Adjust the model structure or parameters based on the evaluation results, such as increasing the number of hidden layers, adjusting the number of neurons, trying different activation functions, etc.

[0248] In the embodiments of the present invention, an integrated learning method (such as model averaging and stacking generalization) can be introduced to further improve the prediction accuracy.

[0249] (4) Prediction

[0250] Use the trained model, input the current carbon emission data and other influencing factors, and predict the carbon emissions of the power grid supply chain management link in a certain time period in the future (such as the next week or month).

[0251] Step 3: Based on the carbon emissions in the power supply link and the carbon emissions in the management link, predict the carbon emissions in the power grid supply chain in the target area during the future target time period.

[0252] Assumptions:

[0253] Carbon emissions from electricity supply (E_supply)

[0254] Carbon emissions from management (E_management)

[0255] Weighting coefficients (W_supply and W_management): used to reflect the relative importance of power supply and management in the carbon emissions of the power supply chain. The weighting coefficients can be set according to actual conditions, such as based on historical data, expert opinions or policy orientations, and the sum of the weighting coefficients is 1.

[0256] The total carbon emissions (E_total) in the power grid supply chain can be calculated using the following formula:

[0257] E_total=W_supply×E_supply+W_management×E_management

[0258] in:

[0259] E_total represents the predicted total carbon emissions in the power grid supply chain.

[0260] W_supply represents the weighted coefficient of the power supply link.

[0261] E_supply represents the predicted carbon emissions in the electricity supply sector.

[0262] W_management represents the weighted coefficient of the management link.

[0263] E_management represents the predicted carbon emissions in the management process.

[0264] Assume that the carbon emissions in the power supply link of a target area in the next year are 10 million tons, and the carbon emissions in the management link are 2 million tons. According to expert opinions and policy guidance, the weighted coefficient of the power supply link is set to 0.8, and the weighted coefficient of the management link is set to 0.2. The total carbon emissions in the power grid supply chain can be calculated by the following formula:

[0265] E_total=0.8×1000+0.2×200=840(10,000 tons)

[0266] Step 4: Post-forecast management of carbon emissions

[0267] In this solution, the carbon emission forecast results can be further analyzed to identify possible trends and anomalies, and the forecast results can be used to formulate emission reduction strategies, optimize energy structure, adjust production plans, etc.

[0268] For example, based on the carbon emission forecast results, we can identify carbon emission anomalies: according to the fluctuation range of historical carbon emission data, we can set a reasonable carbon emission anomaly threshold. When the forecast results exceed this threshold, it is considered an abnormal phenomenon.

[0269] For example, using carbon emission monitoring systems and data analysis tools, we can monitor carbon emission data in real time and compare it with the predicted results. Once an abnormality is found, we can immediately analyze the cause and take corresponding countermeasures.

[0270] Formulate emission reduction strategies based on carbon emission forecast results, such as gradually reducing the use of fossil energy and increasing the proportion of clean energy such as wind energy and solar energy. For example, in electricity production, increase the installed capacity and power generation of renewable energy.

[0271] According to the carbon emission forecast results, production plans and production rhythms are adjusted to avoid large-scale production during peak carbon emission periods. For example, high-energy-consuming products can be produced during periods of low electricity demand.

[0272] According to the carbon emission forecast results, we can adjust the product structure, reduce the production of high-carbon emission products, and increase the proportion of low-carbon emission products. For example, in the automobile manufacturing industry, we can increase the research and development and production of new energy vehicles.

[0273] Implement carbon trading and carbon tax policies based on carbon emission forecast results: Encourage enterprises to reduce carbon emissions and purchase carbon emission rights through the carbon trading market. Enterprises with lower carbon emissions can obtain economic benefits by selling carbon emission rights, thereby encouraging more enterprises to participate in emission reduction actions. Carbon taxes are imposed on high-carbon emission enterprises to increase their carbon emission costs and force them to take emission reduction measures. Carbon tax revenue can be used to support the development of clean energy, energy-saving and emission reduction technology research and development, etc.

[0274] Take a power company as an example. The company established a carbon emission prediction model based on historical carbon emission data and multiple factors. The prediction results show that the company's carbon emissions will continue to grow in the future and may exceed the set abnormal threshold.

[0275] In response to this forecast, the company has adopted the following emission reduction strategies:

[0276] Optimize energy structure: Increase investment in renewable energy power generation and increase the proportion of clean energy such as wind power and photovoltaic power generation. At the same time, carry out technical transformation and upgrading of existing coal-fired power generation units to improve power generation efficiency and reduce carbon emissions.

[0277] Adjust production plans: Carry out inspection and maintenance of high-energy-consuming equipment during periods of low electricity demand to reduce unnecessary energy consumption and carbon emissions. At the same time, optimize power dispatch plans and reasonably arrange the power generation sequence and load distribution of different units.

[0278] Implement carbon trading and carbon tax policies: actively participate in the carbon trading market and meet their own production needs by purchasing carbon emission rights. At the same time, pay attention to the dynamic changes in the government's carbon tax policy, make preparations in advance and adjust their own business strategies.

[0279] Through the implementation of the above measures, the company successfully controlled the growth trend of carbon emissions and avoided the occurrence of abnormal phenomena. At the same time, these emission reduction strategies also brought the company both economic and social benefits.

[0280] Assume that the carbon emissions of a power grid in the power supply link in the past year were 10 million tons, and the carbon emissions in the management link were 2 million tons. By predicting the carbon emissions in the power supply link and the management link respectively, and predicting the carbon emissions in the next year, the following results are obtained:

[0281] Predicted carbon emissions from power supply: 10.5 million tons (actual value: 10.2 million tons, error: 3%)

[0282] Management-level predicted carbon emissions: 2.1 million tons (actual value: 2.05 million tons, error: 2.5%)

[0283] By integrating the prediction results of the two links, the predicted value of carbon emissions in the power supply chain is 12.6 million tons (actual value: 12.25 million tons, error: 2.8%). It can be seen that by predicting the carbon emissions of the power supply link and the management link separately and integrating the two, a more accurate prediction value of carbon emissions in the power supply chain can be obtained.

[0284] Therefore, the present invention can improve the prediction accuracy of carbon emissions in the power supply chain by splitting the power supply chain into the power supply link and the management link for prediction, and integrating the data of the two links. This method not only takes into account the impact of the power supply link on carbon emissions, but also takes into account the contribution of the management link to carbon emissions, thereby more comprehensively reflecting the carbon emissions of the power supply chain. At the same time, by comparing the predicted results with the actual values, the accuracy of the model can be verified, and a scientific basis can be provided for future carbon emission management and emission reduction strategies.

[0285] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting carbon emissions in a power grid supply chain, characterized in that: The steps include: Step 1: Predict the carbon emissions of the power supply link in the target area power grid supply chain; Step 2: Predict the carbon emissions of the management links in the target area power grid supply chain; Step 3: Based on the carbon emissions in the power supply link and the carbon emissions in the management link, predict the carbon emissions in the power grid supply chain in the target area during the future target time period.

2. The method for predicting carbon emissions in a power grid supply chain according to claim 1, characterized in that: Step 1 specifically includes: Step 1.1: Collect historical overall carbon emissions data of the target area power grid supply chain; Step 1.2: Based on the historical carbon emission data of the target area power grid supply chain, predict the carbon emission of the power supply link of the target area power grid supply chain.

3. The carbon emission prediction method in the power grid supply chain according to claim 2, characterized in that: Step 1.1 specifically includes: Step 1.1.1: Identify the various links in the power grid supply chain Power generation: including thermal power generation, hydropower generation, wind power generation, and solar power generation; Transmission link: including high-voltage transmission lines and substations; Power distribution link: including low-voltage distribution lines and distribution equipment; Step 1.1.2: Collect historical data on the target area power grid supply chain Collect energy consumption data from all links of the target area’s power grid supply chain; Collect carbon emission factors of equipment, technologies and materials used in each link of the target area power grid supply chain; Step 1.1.3: Calculate the historical overall carbon emissions of the target area power grid supply chain: Calculate the historical overall carbon emissions of each link based on energy consumption data and carbon emission factors; Carbon emissions calculation formula: Pi = Ei * Fi Among them, Pi represents the carbon emission of the i-th link, i represents the link of the power grid supply chain, Ei is the energy consumption of the i-th link, and Fi is the carbon emission factor of the i-th link; Step 1.1.4: Summarize the carbon emissions of each link in the target area power supply chain to obtain the historical carbon emissions data of the power supply chain: P=PA+PB+PC PA represents the carbon emissions in the power generation link, PB represents the carbon emissions in the transmission link, and PC represents the carbon emissions in the distribution link.

4. The method for predicting carbon emissions in a power grid supply chain according to claim 3, characterized in that: Step 1.2 specifically includes: Step 1.2.1: Based on the historical carbon emission data of the target area power grid supply chain, make a preliminary forecast of the carbon emission of the power supply link of the target area power grid supply chain in the future; Step 1.2.2, determine the current power transformation development state, and obtain the parameters of the current power transformation development state; Step 1.2.3: Combine the parameters of the power transformation development status and the carbon emissions of the power supply link initially predicted in step 1.2.1 to obtain the final carbon emissions of the power supply link of the target area power grid supply chain.

5. The method for predicting carbon emissions in a power grid supply chain according to claim 4, characterized in that: In step 1.2.1, the network model is trained based on the historical carbon emission data of the power grid supply chain in the target area. By collecting the carbon emission data of the target area at a certain time at the current time, the overall carbon emissions of the power grid supply chain in the target area in a certain time period in the future are predicted.

6. The method for predicting carbon emissions in a power grid supply chain according to claim 5, characterized in that: In step 1.2.1, the overall carbon emissions of the target area power grid supply chain are predicted in a certain period of time in the future. The specific prediction method is as follows: (1) Data Collection Obtain daily / monthly / yearly carbon emission data from the target area power grid supply chain over the past few years. The data includes date, time, and carbon emissions at each stage. (2) Data preprocessing Clean the collected data to remove missing values ​​and outliers; Create new feature data, including seasonal characteristics, holiday markers, and energy consumption type proportions; Standardize or normalize different types of data to ensure that data of different dimensions can be compared on the same scale, thus improving model training efficiency; (3) Training Model Considering the needs of time series prediction, we choose the long short-term memory network LSTM, gated recurrent unit GRU or Transformer neural network model. The specific steps of training the model are as follows: Divide the data into training set, validation set and test set, where 70% of the data is used for training, 15% for validation, and 15% for testing; Determine learning rate, batch size, training rounds, optimizer parameters; Train the model using the training set data and evaluate its performance on the validation set; Adjust model structure or parameters as needed to improve prediction accuracy; Use the test set data to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE). Check the model’s forecasting performance over different time periods; Adjust the model structure or parameters based on the evaluation results, including increasing the number of hidden layers, adjusting the number of neurons, and trying different activation functions; (4) Prediction Use the trained model, input the current carbon emission data and other influencing factors, and predict the overall carbon emissions of the power grid supply chain in a certain period of time in the future.

7. The method for predicting carbon emissions in a power grid supply chain according to claim 4, characterized in that: In step 1.2.2, considering the changes in power generation structure and the continuous changes in the development level of coal-fired power technology, three different power transformation development scenarios are set, namely the baseline scenario, the fast transformation scenario, and the slow transformation scenario; the average annual change rate of the proportion of power generation of each power generation mode in the baseline scenario is set according to the average annual change rate within its preset period. The slow transformation scenario slows down the transformation speed of thermal power to clean energy power generation on the basis of the baseline scenario, and the fast transformation scenario accelerates the transformation speed of thermal power to clean power. Fixed parameters λ are set for the three different power transformation development scenarios.

8. The method for predicting carbon emissions in a power grid supply chain according to claim 7, characterized in that: Step 1.2.3 uses the following formula to calculate the carbon emissions of the power supply link in the target area power grid supply chain: P2=P1*λ Among them, P2 is the carbon emissions of the power supply link of the final future target regional power grid supply chain, P1 is the carbon emissions of the power supply link initially predicted in step 1.2.1, and λ is a set fixed parameter; Among them, the parameter λ in the baseline scenario is set to 1 based on the overall carbon emissions calculated in step 1.1.4, indicating that it is the same as in step 1.1.

4. The parameter λ in the fast transition scenario is set to 0.8, indicating that the transformation speed from thermal power to clean power is accelerated, and its carbon emissions need to be converted based on conventional forecasts. In theory, the carbon emissions are less than the baseline scenario. The slow transition scenario is to slow down the transformation speed from thermal power to clean energy power generation based on the baseline scenario. In theory, the carbon emissions are more than the baseline scenario, and the parameter λ is set to 1.

2.

9. The method for predicting carbon emissions in a power grid supply chain according to claim 1, characterized in that: Step 2 specifically includes: Step 2.1, collect historical carbon emission data of the storage link in the power grid supply chain; Industrial transport vehicles are used for loading and unloading of goods in the receiving, inventory and shipping operations in the warehousing process. For the above-mentioned operations, industrial transport vehicles are energy-consuming equipment, and the carbon emissions calculation method is to measure the actual electricity or fuel consumption of industrial transport vehicles for each operation. Among them: the electricity / diesel emission factor is obtained by citing the electricity and diesel carbon emission factors of industrial transport vehicles released by the Ministry of Ecology and Environment to calculate the carbon emissions of industrial transport vehicles; or the electricity and diesel carbon emission factors are estimated according to the model, working conditions and power source of the forklift; if none of the above methods can be found, it is set by experimental measurement or expert estimation; The carbon emission calculation of industrial handling vehicles is to calculate the product of the power consumption and the electric emission factor of the electric industrial handling vehicle for a certain period of time, and the sum of the product of the fuel consumption and the diesel emission factor of the diesel industrial handling vehicle. The specific carbon emission calculation formula is as follows: Carbon emissions from warehousing in the power grid supply chain = carbon emissions from receiving operations + carbon emissions from tallying + carbon emissions from shipping operations + carbon emissions from storage operations + carbon emissions from the operation process in: Carbon emissions from receiving operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 1 Carbon emissions from shipping operations = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 2 Carbon emissions from tallying = energy consumption data of industrial handling vehicles × electricity or oil emission factor + power consumption of overhead cranes × electricity emission factor 3 Storage operation carbon emissions = industrial handling vehicle energy consumption data × electricity or oil emission factor + crane power consumption × electricity emission factor 4 Among them, the energy consumption data of industrial handling vehicles refers to the power consumption / fuel consumption of industrial handling vehicles. By collecting the power or fuel consumption and business volume data of a single operation of industrial handling vehicles, the power and fuel consumption of industrial handling vehicles in a certain period of time is calculated; by collecting the operation time and business volume data of a crane loading and unloading operation, the power consumption of the crane in a certain period of time is calculated; by collecting the operation time and business volume data of automatic shelves loading and unloading shelves, the power consumption of the crane in a certain period of time is calculated; Step 2.2: Collect historical carbon emission data of the operation links in the power grid supply chain Carbon emissions from operations = public electricity × electricity emission factor 5 + energy consumption of purchased products and services × electricity emission factor 6 per unit of goods or services Step 2.3, based on the carbon emission data of the storage link and the carbon emission data of the operation link in the power grid supply chain, calculate the historical carbon emission data of the management link of the power grid supply chain; Step 2.4, based on the historical carbon emission data of the management link of the power grid supply chain, predict the carbon emission data of the management link of the power grid supply chain in a certain time period in the future; The specific prediction method is as follows: (1) Data Collection Obtain daily / monthly / yearly carbon emission data for the past few years from the historical carbon emission data of the target area power grid supply chain management link. The data includes date, time, and carbon emission amount of each link. (2) Data preprocessing Clean the collected data to remove missing values ​​and outliers; Create new feature data, including seasonal characteristics, holiday markers, and energy consumption type proportions; Standardize or normalize different types of data to ensure that data of different dimensions can be compared on the same scale, thus improving model training efficiency; (3) Training Model Considering the needs of time series prediction, choose the long short-term memory network LSTM, gated recurrent unit GRU or Transformer neural network model; The specific steps of training the model are as follows: Divide the data into training set, validation set and test set, where 70% of the data is used for training, 15% for validation, and 15% for testing; Determine learning rate, batch size, training rounds, optimizer parameters; Train the model using the training set data and evaluate its performance on the validation set; Adjust model structure or parameters as needed to improve prediction accuracy; Use the test set data to evaluate the prediction accuracy of the model. Common indicators include mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE). Check the model’s forecasting performance over different time periods; Adjust the model structure or parameters based on the evaluation results, including increasing the number of hidden layers, adjusting the number of neurons, and trying different activation functions; (4) Prediction Use the trained model, input the current carbon emission data and other influencing factors, and predict the carbon emissions of the power grid supply chain management link in a certain period of time in the future.

10. The method for predicting carbon emissions in a power grid supply chain according to claim 1, characterized in that: Step three specifically includes: The carbon emissions of the power supply link are set as E_supply, the carbon emissions of the management link are set as E_management, and the weighted coefficients of the power supply link and the management link are set as W_supply and W_management respectively, which are used to reflect the relative importance of the power supply link and the management link in the carbon emissions of the power grid supply chain. The sum of the weighted coefficients is 1; The total carbon emissions E_total in the power grid supply chain are calculated using the following formula: E_total=W_supply×E_supply+W_management×E_management in: E_total represents the total carbon emissions in the predicted power grid supply chain; W_supply represents the weighted coefficient of the power supply link; E_supply represents the predicted carbon emissions from the electricity supply link; W_management represents the weighted coefficient of the management link; E_management represents the predicted carbon emissions in the management process.

Images