Carbon emission prediction system and method

Abstract

The invention provides a carbon emission prediction system and method, and relates to the technical field of carbon emission prediction, and the system comprises a carbon emission collection module which is used for collecting carbon emission data of different industries in a monitoring region; the data transmission and analysis module is used for binding the acquired carbon emission with the industry and characteristics thereof, transmitting data, comparing the carbon emission data with other main data of the industry, and selecting data influencing the carbon emission in the industry; and the feature selection module extracts features having important influence on industry carbon emission prediction from selected data, applies a feature selection algorithm, monitors carbon emission data of different industries, then imports the carbon emission data and industry data into a prediction system, generates an industry prediction system, and in the system generation process, generates a prediction result of the industry carbon emission prediction. Experiments related to influence factors are carried out, the accuracy of the prediction system is ensured through multiple experiments and verification, and then the prediction system is used for predicting the subsequent carbon emission of the industry.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon emission prediction systems, and in particular to a carbon emission prediction system and method. Background Art

[0002] As the global climate change problem becomes increasingly serious, accurate prediction of carbon emissions is crucial for formulating effective emission reduction policies and promoting sustainable development. Existing carbon emissions prediction methods have some limitations. On the one hand, some prediction models rely solely on a single data source, such as energy consumption data or economic activity data, ignoring the impact of other important factors on carbon emissions, resulting in insufficient accuracy and comprehensiveness of the prediction results. On the other hand, traditional prediction models are often static and difficult to adapt to the ever-changing economic structure, technological progress and policy environment, and lack the ability to effectively capture the dynamic trends of carbon emissions. In addition, existing prediction systems also have certain problems in data processing and analysis efficiency. They are unable to quickly process large-scale, multi-dimensional carbon emission-related data and are unable to meet the needs of real-time prediction and decision support.

[0003] The existing authorization announcement number CN 118674087 A is a carbon emission prediction system, including: a carbon emission data acquisition module: used to collect different types of carbon emission data in the monitoring area; a data transmission processing module: used to receive the carbon emission data and process it to obtain the total carbon emission; a correlation analysis module: used to analyze the correlation factors affecting carbon emissions through grey correlation analysis to obtain a carbon emission sequence data set; a carbon emission prediction module: used to predict carbon emissions through a neural network prediction model to obtain a prediction result. The present invention improves the accuracy of the prediction, but when this prediction system is used, the predicted value cannot be corrected to different degrees according to the industry, and it is difficult to meet the prediction needs of carbon emissions in different industries.

[0004] Therefore, the present invention proposes a carbon emission prediction system and method for correcting carbon emissions for different industries and different influencing factors in different industries to solve such problems. Summary of the invention

[0005] To achieve the above objectives, the present invention proposes a carbon emission prediction system and method, comprising: Carbon emission collection module: used to collect carbon emission data of different industries in the monitoring area; Data transmission and analysis module: Bind the collected carbon emissions with the industry and its characteristics, transmit the data, compare the carbon emissions data with other major data in the industry, and select the data that affects carbon emissions in the industry; Feature selection module: extract features that have a significant impact on the prediction of carbon emissions in the industry from the selected data, apply feature selection algorithms, use correlation analysis, chi-square test, and recursive feature elimination to select the most representative and predictive feature data from the many extracted features; Data preprocessing module: cleans and converts the collected carbon emission data; Forecasting system module: imports carbon emission data, industry characteristic data and influencing factors into a module that integrates a variety of advanced forecasting models; Result analysis and evaluation module: Considering the impact of future economy, policies and technologies on carbon emissions in different industries, a variety of evaluation indicators are used to evaluate the forecast results.

[0006] In one example, the carbon emissions collection module collects carbon emissions from different industries in the monitoring area. When it is necessary to predict the industry carbon emissions data, the collected data is directly used for prediction. When it is necessary to predict the total carbon emissions of different industries in the monitoring area, the collected carbon emissions data can be added together for prediction, or the carbon emissions of each industry can be predicted and added together.

[0007] In one example, the data transmission and analysis module transmits the carbon emission data of each industry, and analyzes and compares it with other data of the industry to obtain a correlation function, or combines other data of the industry and then associates it with the carbon emission to obtain a correlation function.

[0008] In one example, the feature selection module selects features that have an important impact on carbon emissions prediction from the integrated data, including direct features such as total energy consumption, carbon emission intensity, and industrial structure ratio; indirect features such as economic growth rate, energy price index, and technological innovation index; and meteorological related features such as the impact of temperature changes on energy demand. In addition, new features such as energy structure change rate and carbon emission elasticity coefficient are constructed through data transformation and combination. The feature selection module uses feature selection algorithm, correlation analysis, chi-square test, and recursive feature elimination to select the most representative and predictive feature subsets from the many extracted features. Feature selection can not only reduce the input dimension of the model and improve the model training efficiency, but also avoid overfitting problems and enhance the generalization ability of the model.

[0009] In one example, the data preprocessing module cleans, converts and performs feature engineering on the collected data. Data cleaning is to remove noise, duplicate data and missing values ​​in the data, and to process missing values ​​using methods such as data interpolation and filling. Data conversion is to unify data in different formats and units, and to unify the units of energy consumption data into standard coal equivalent. Feature engineering is to extract and construct features that have an important impact on carbon emissions prediction, such as energy intensity, energy consumption per unit of GDP, carbon emission intensity, and carbon emissions per unit of GDP. The preprocessed data can improve the training efficiency and prediction accuracy of subsequent prediction models.

[0010] In one example, the prediction system module integrates a variety of advanced prediction models, including machine learning models, support vector machines, random forests, long short-term memory networks LSTM and deep learning models, and a combination model of convolutional neural networks CNN and LSTM. For different types of data and prediction needs, the system can automatically select appropriate models for training and prediction. During the model training process, historical data is used to adjust and optimize the model parameters, and the model performance is evaluated through cross-validation and other methods to select the optimal model parameters. The prediction model module also has a model update function, which can continuously update the model according to new data to adapt to the dynamic changes in carbon emissions.

[0011] In one example, the result analysis and evaluation module takes into account the uncertainty of future economic, policy and technological factors. The module predicts carbon emissions by setting different scenarios, including baseline scenarios, policy-driven scenarios, and technological innovation scenarios. In each scenario, the input parameters of the prediction model are adjusted in combination with the changing trends of relevant factors to obtain carbon emission prediction results under different scenarios. The scenario analysis module provides decision makers with a variety of possible carbon emission development paths, which helps to formulate more forward-looking and adaptive emission reduction strategies. The result evaluation and visualization module evaluates the prediction results and uses a variety of evaluation indicators, including root mean square error RMSE, mean absolute error MAE, and determination coefficient R², to compare with actual carbon emission data to verify the accuracy and reliability of the prediction model. At the same time, the prediction results are presented in an intuitive and visual way, drawing a curve of carbon emissions changing over time, a bar chart comparing carbon emissions under different scenarios, and a radar chart of the impact of various factors on carbon emissions. The visual interface facilitates decision makers and relevant personnel to understand and analyze the prediction results, providing strong support for decision-making.

[0012] A carbon emissions prediction method includes that different industries' carbon emissions need to use different prediction systems. The prediction systems vary according to the carbon emissions data and influencing factors of different industries, and have different sensitivities to different factors.

[0013] The carbon emission prediction system and method proposed by the present invention can bring the following beneficial effects: 1. The present invention generates an industry forecasting system by monitoring the carbon emission data of different industries, and then importing the carbon emission data and industry data into the forecasting system. During the system generation process, experiments on relevant influencing factors are conducted, and multiple experiments and verifications are performed to ensure the accuracy of the forecasting system, which is then used to predict subsequent carbon emissions in the industry.

[0014] 2. The present invention assists in generating carbon emission prediction systems for different industries, making the prediction system more accurate. At the same time, the monitored carbon emission data can be accumulated to facilitate the prediction of carbon emission data in the monitored area when a certain industry relocates or disappears, and can better predict the reduction in carbon emissions caused by sudden reasons. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 A schematic diagram of a carbon emission prediction system and method according to the present invention. DETAILED DESCRIPTION

[0016] In order to more clearly and completely illustrate the technical solution of the present invention, the present invention is further described below in conjunction with the accompanying drawings.

[0017] Please refer to Figure 1 The present invention proposes a carbon emission prediction system and method, comprising: Carbon emission collection module: used to collect carbon emission data of different industries in the monitoring area; Data transmission and analysis module: Bind the collected carbon emissions with the industry and its characteristics, transmit the data, compare the carbon emissions data with other major data in the industry, and select the data that affects carbon emissions in the industry; Feature selection module: extract features that have a significant impact on the prediction of carbon emissions in the industry from the selected data, apply feature selection algorithms, use correlation analysis, chi-square test, and recursive feature elimination to select the most representative and predictive feature data from the many extracted features; Data preprocessing module: cleans and converts the collected carbon emission data; Forecasting system module: imports carbon emission data, industry characteristic data and influencing factors into a module that integrates a variety of advanced forecasting models; Result analysis and evaluation module: Based on the impact of future economy, policies and technologies on carbon emissions in different industries, a variety of evaluation indicators are used to evaluate the forecast results.

[0018] The data collection module establishes connections with data sources such as the energy department, statistics department, and meteorological department through a network interface to obtain energy consumption data, economic activity data, policy and regulatory data, and meteorological data in real time. For example, it obtains monthly coal, oil, natural gas and other energy consumption data from the management system of energy companies; obtains quarterly GDP data, industrial structure data, etc. from the website of the Statistics Bureau; obtains carbon emission policy documents and related regulatory information from the website of the government environmental protection department; obtains daily temperature, precipitation, wind speed and other meteorological data from the meteorological station database, and the collected data is stored in the system database in the prescribed format.

[0019] The data preprocessing module cleans the collected data and fills the missing values ​​in the energy consumption data with linear interpolation. It identifies and corrects the outliers in the economic activity data through statistical methods, converts the units of energy consumption data into standard coal equivalent, normalizes the economic activity data with different statistical calibers, extracts features such as energy intensity and carbon emission intensity, constructs new feature variables, energy structure ratio, and industry carbon emission contribution ratio, and stores the preprocessed data in the data warehouse to provide data support for subsequent prediction model training.

[0020] According to the characteristics of the data and the prediction requirements, the prediction model module automatically selects the appropriate prediction model. For carbon emission data with time series characteristics, the long short-term memory network LSTM model is selected for training. If there are a large number of nonlinear features in the data, the support vector machine SVM model may be more suitable. For multi-dimensional and complex data structures, a combination of convolutional neural network CNN and LSTM can be used. For example, when predicting short-term carbon emissions, the LSTM model is selected because the carbon emission data has obvious time series characteristics; when predicting long-term carbon emissions and considering the combined impact of multiple factors, a combination of CNN and LSTM is used.

[0021] The selected model is trained using historical data, and the preprocessed data is divided into training set, validation set and test set in a ratio of 7:2:1. During the training process, appropriate model parameters are set, such as the number of hidden layer nodes, learning rate, number of iterations, etc. of the LSTM model. The weight of the model is adjusted through the back propagation algorithm to minimize the loss function of the model on the training set. The performance of the model is evaluated on the validation set, and indicators such as root mean square error (RMSE) and mean absolute error (MAE) are used to measure it. According to the evaluation results of the validation set, the model parameters are adjusted to prevent the model from overfitting or underfitting. After multiple iterative training, the model parameters with the best performance on the validation set are selected.

[0022] The trained model is applied to the test set data for prediction to obtain the predicted value of carbon emissions, which is compared with the actual value and the prediction error is calculated. The accuracy and reliability of the model are further evaluated based on the prediction error. If the prediction error is within an acceptable range, the model training is considered successful and can be used for actual carbon emissions prediction.

[0023] The scenario analysis module sets multiple scenarios based on different assumptions about future economic, policy and technological development. For example, the baseline scenario assumes that the economic growth rate will remain at the current level in the next five years, and the energy structure and carbon emission policies will remain unchanged; the policy-driven scenario assumes that the government will introduce stricter carbon emission policies in the next three years, increase carbon emission taxes, implement carbon emission quota trading, etc.; the technological innovation scenario assumes that major low-carbon technological breakthroughs will occur in the next five years, such as significant progress in high-efficiency solar cell technology and carbon capture and storage technology, which will significantly reduce the cost of renewable energy and significantly improve carbon capture efficiency.

[0024] Under different scenarios, the input parameters of the prediction model are adjusted. Under the policy-driven scenario, the energy structure parameters and carbon emission intensity parameters are adjusted according to the new carbon emission policy; under the technological innovation scenario, the energy consumption efficiency parameters and renewable energy proportion parameters are adjusted according to the development of low-carbon technology. Then, the adjusted prediction model is used to predict carbon emissions and obtain the carbon emission prediction results under different scenarios.

[0025] The result evaluation and visualization module uses a variety of evaluation indicators to evaluate the prediction results, and calculates the root mean square error RMSE, mean absolute error MAE and determination coefficient R² between the predicted value and the actual value. For example, for the carbon emissions forecast for a certain period of time, RMSE reflects the average error degree between the predicted value and the actual value, MAE reflects the average absolute value of the error between the predicted value and the actual value, and R² measures the degree of fit of the prediction model to the actual data. These indicators are used to comprehensively evaluate the accuracy and reliability of the prediction model.

[0026] The forecast results are presented in an intuitive visual way, and bar charts are used to compare the predicted carbon emissions under different scenarios, so that decision makers can clearly understand the changing trends of carbon emissions under different scenarios; a curve of carbon emissions over time is drawn to display historical data and forecast data of carbon emissions, which is convenient for analyzing the long-term trend of carbon emissions; radar charts are used to show the impact of various factors, such as energy consumption, economic activities, policies and regulations, on carbon emissions, to help decision makers understand the role of different factors in carbon emissions. The visualization interface is implemented using Web technology, and decision makers can access the system through a browser to view forecast results and analysis charts.

[0027] The decision support module provides specific emission reduction suggestions to decision makers based on the carbon emission forecast results under different scenarios. If carbon emissions show an upward trend under the baseline scenario and are far from the emission reduction target, it is recommended to optimize the energy structure, increase the proportion of renewable energy, and reduce dependence on fossil energy; strengthen industrial energy conservation and emission reduction measures and improve energy utilization efficiency. If the decline in carbon emissions still does not reach the expected target under the policy-driven scenario, it is recommended to further strengthen the implementation of carbon emission policies and strictly supervise the carbon emission behavior of enterprises.

[0028] Evaluate the costs and benefits of different emission reduction measures, calculate the investment cost required to implement a certain emission reduction measure, such as building a solar power plant, promoting energy-saving technology, etc., as well as the carbon emission reduction benefits that can be brought about after the implementation of the measure, such as reduced carbon emissions, reduced environmental governance costs, etc. Through cost-benefit analysis, help decision makers choose the optimal emission reduction strategy to maximize economic and environmental benefits while achieving emission reduction goals.

[0029] Of course, the present invention may have many other implementations. Based on this implementation, other implementations obtained by ordinary technicians in this field without any creative work are all within the scope of protection of the present invention.

Claims

1. A carbon emission prediction system, characterized in that: include: Carbon emission collection module: used to collect carbon emission data of different industries in the monitoring area; Data transmission and analysis module: Bind the collected carbon emissions with the industry and its characteristics, transmit the data, compare the carbon emissions data with other major data in the industry, and select the data that affects carbon emissions in the industry; Feature selection module: extract features that have a significant impact on the prediction of carbon emissions in the industry from the selected data, apply feature selection algorithms, use correlation analysis, chi-square test, and recursive feature elimination to select the most representative and predictive feature data from the many extracted features; Data preprocessing module: cleans and converts the collected carbon emission data; Forecasting system module: imports carbon emission data, industry characteristic data and influencing factors into a module that integrates a variety of advanced forecasting models; Result analysis and evaluation module: Considering the impact of future economy, policies and technologies on carbon emissions in different industries, a variety of evaluation indicators are used to evaluate the forecast results.

2. A carbon emission prediction system according to claim 1, characterized in that: The carbon emission collection module collects carbon emissions from different industries in the monitoring area. When it is necessary to predict the carbon emission data of an industry, the collected data is directly used for prediction. When it is necessary to predict the total carbon emissions of different industries in the monitoring area, the collected carbon emission data can be added together for prediction, or the carbon emissions of each industry can be predicted and added together.

3. A carbon emission prediction system according to claim 1, characterized in that: The data transmission and analysis module transmits the carbon emission data of each industry, and analyzes and compares it with other data of the industry to obtain a correlation function, or combines other data of the industry and then associates it with the carbon emission to obtain a correlation function.

4. A carbon emission prediction system according to claim 1, characterized in that: The feature selection module selects features that have an important impact on carbon emission prediction from the integrated data, such as the total energy consumption, carbon emission intensity, and industrial structure ratio of direct features; Indirect features include economic growth rate, energy price index, technological innovation index; as well as meteorological related features, the impact of temperature changes on energy demand. In addition, new features are constructed through data transformation and combination, such as energy structure change rate and carbon emission elasticity coefficient. The feature selection module uses feature selection algorithm, correlation analysis, chi-square test, and recursive feature elimination to select the most representative and predictive feature subsets from the many extracted features. Through feature selection, not only can the input dimension of the model be reduced and the model training efficiency be improved, but also overfitting problems can be avoided and the generalization ability of the model can be improved.

5. A carbon emission prediction system according to claim 1, characterized in that: The data preprocessing module cleans, converts and performs feature engineering on the collected data. Data cleaning is to remove noise, duplicate data and missing values ​​in the data, and to process missing values ​​using methods such as data interpolation and filling. Data conversion is to unify data of different formats and units, and to unify the units of energy consumption data into standard coal equivalent. Feature engineering is to extract and construct features that have an important impact on carbon emissions prediction, such as energy intensity, energy consumption per unit of GDP, carbon emission intensity, and carbon emissions per unit of GDP. The preprocessed data can improve the training efficiency and prediction accuracy of subsequent prediction models.

6. A carbon emission prediction system according to claim 1, characterized in that: The prediction system module integrates a variety of advanced prediction models, including machine learning models, support vector machines, random forests, long short-term memory networks LSTM and deep learning models, and a combination model of convolutional neural networks CNN and LSTM. For different types of data and prediction needs, the system can automatically select appropriate models for training and prediction. During the model training process, historical data is used to adjust and optimize the model parameters, and the model performance is evaluated through cross-validation and other methods to select the optimal model parameters. The prediction model module also has a model update function, which can continuously update the model according to new data to adapt to the dynamic changes in carbon emissions.

7. A carbon emission prediction system according to claim 1, characterized in that: The result analysis and evaluation module takes into account the uncertainty of future economic, policy and technological factors. The module predicts carbon emissions by setting different scenarios, including baseline scenarios, policy-driven scenarios and technological innovation scenarios. In each scenario, the input parameters of the prediction model are adjusted in combination with the changing trends of relevant factors to obtain carbon emission prediction results under different scenarios. The scenario analysis module provides decision makers with a variety of possible carbon emission development paths, which helps to formulate more forward-looking and adaptive emission reduction strategies. The result evaluation and visualization module evaluates the prediction results and uses a variety of evaluation indicators, including root mean square error RMSE, mean absolute error MAE and determination coefficient R², to compare with actual carbon emission data to verify the accuracy and reliability of the prediction model. At the same time, the prediction results are presented in an intuitive and visual way, drawing a curve of carbon emissions changing over time, a bar chart comparing carbon emissions under different scenarios, and a radar chart of the impact of various factors on carbon emissions. The visual interface facilitates decision makers and relevant personnel to understand and analyze the prediction results, providing strong support for decision-making.

8. A method for predicting carbon emissions, characterized in that: The carbon emission data of different industries are imported into different forecasting systems. The forecasting systems vary according to the carbon emission data and influencing factors of different industries, and have different sensitivities to different factors.

Images