CN117172347A - Carbon emission prediction method based on energy big data - Google Patents

Abstract

The invention discloses a carbon emission prediction method based on energy big data, which relates to the technical field of carbon emission prediction, including: S1, data integration of carbon emission influencing factors; S2, carbon emission prediction model construction; S3, prediction model training and impact simulation ; S4, impact analysis of carbon emission prediction; S5, carbon emission prediction and result correction. This carbon emission prediction method based on energy big data studies the relationship between the preliminary prediction of carbon emissions and the impact prediction, establishes a mathematical equation relationship, calculates the preliminary prediction results of carbon emissions, and then uses the mathematical equation relationship to predict the preliminary prediction results. Calibration not only builds a carbon emission prediction model, but also simulates the impact of various influencing factors on the prediction model under different circumstances to avoid the impact of differences in carbon emission influencing factors in different fields and different equipment on the prediction model, thereby avoiding its impact on the prediction model. Cause interference to carbon emission predictions and improve the accuracy and reliability of carbon emission prediction results.

Description

A carbon emission prediction method based on energy big data

Technical field

The invention relates to the technical field of carbon emission prediction, specifically a carbon emission prediction method based on energy big data.

Background technique

Energy big data is a related technology and application that comprehensively collects, processes, analyzes and applies energy data such as electricity, gas, and oil as well as data in economic, population, geography, climate and other related fields. With the continuous deepening of 5G, AI intelligence and Internet of Things technologies and applications, on the basis of mobile Internet and enterprise informatization, relying on technologies such as big data and blockchain, the collection, analysis and analysis of structured and unstructured information can be realized. Integration enables timely, accurate, credible, usable and traceable collection of massive data. It is not only a deep integration of energy production, management, consumption and related technological revolutions with big data concepts, but also can find new directions for the development of the energy industry and business model innovation.

For example, the patent document with application number 202111516267.2 discloses a method for predicting carbon emissions in the energy industry based on power data. This patent builds a conversion relationship between power data in the energy industry - energy consumption data - carbon emissions, based on power data in the energy-using industry. Improve the carbon emission management function, and at the same time predict and warn corporate carbon emissions, fully tap the carbon emission reduction potential of enterprises, realize the prediction and early warning of carbon emissions of key energy-consuming enterprises, and ultimately achieve a data-driven multi-industry energy consumption and carbon emission panorama forecast, and at the same time serve key energy-consuming enterprises to tap their carbon emission reduction potential and carry out energy conservation and consumption reduction.

However, carbon emission prediction methods similar to the above documents still have the following shortcomings:

Existing carbon emission prediction methods generally build a carbon emission prediction model and calculate carbon emissions by substituting relevant parameters and variables. The accuracy and reliability of the carbon emission prediction results directly depend on the prediction model, but due to different fields and even There are differences in the factors affecting carbon emissions of different equipment, so the prediction model is easily affected by these differences, which interferes with the carbon emission prediction and affects the accuracy and reliability of the carbon emission prediction results.

Therefore, there is an urgent need to improve this shortcoming. The present invention studies and improves the existing structure and shortcomings, and provides a carbon emission prediction method based on energy big data.

Contents of the invention

The purpose of the present invention is to provide a carbon emission prediction method based on energy big data to solve the problems raised in the above background technology.

In order to achieve the above objectives, the present invention provides the following technical solution: a carbon emission prediction method based on energy big data, including the following specific steps:

S1. Data integration of factors affecting carbon emissions:

Analyze and determine the factors affecting carbon emissions in the energy industry, connect to the energy information database, retrieve relevant information and data on factors affecting carbon emissions in the database, and establish an energy information data set. Set multiple subsets in the energy information data set to combine different carbon emissions. The influencing factor information data is placed inside the subset, and information subsets of different categories of carbon emission influencing factors are established;

S2. Construction of carbon emission prediction model:

Clean and correct the data in the data subset, and use the random forest algorithm to determine the importance of influencing factors, sort the influencing factors according to their importance, select the main influencing factors, generate a feature matrix, and construct a carbon emission prediction Model;

S3. Predictive model training and impact simulation:

The carbon emission prediction model is trained according to the characteristic matrix to test the carbon emission prediction model, and after training, a preliminary prediction of carbon emissions is made based on the input values of each influencing factor, and then the parameters of the influencing factors are adjusted to simulate Be affected by different situations and conduct carbon emission predictions;

S4. Impact analysis of carbon emission forecast:

Study the relationship between preliminary prediction of carbon emissions and impact prediction, combine energy data to analyze the impact and change patterns of influencing factors, determine the internal correlation between the two through cluster analysis and association rule analysis, draw a causal relationship diagram, and establish mathematical equations relation;

S5. Carbon emission prediction and result correction:

The energy data in the energy information database is retrieved, substituted into the carbon emission prediction model to calculate the preliminary prediction results of carbon emissions, and then the preliminary prediction results are corrected using mathematical equations to remove the interference caused by the impact of influencing factors on the carbon emission prediction, and the calculation The final carbon emission prediction results are obtained.

Further, in step S1, the energy information data set {Edb_Co2}={ΔCo2_1}+{ΔCo2_2}+{ΔCo2_3}+...+{ΔCo2_n}, where Co2_1, Co2_2, Co2_3,..., Co2_n respectively represent different carbons. Emission influencing factors, {ΔCo2_1}, {ΔCo2_2}, {ΔCo2_3}, ..., {ΔCo2_n} respectively represent the information subsets corresponding to different influencing factors.

Further, in step S1, the information data of carbon emission influencing factors in the same information subset are independent of each other and do not interfere with each other, and there is no overlap or interaction between the information data of carbon emission influencing factors in different information subsets.

Furthermore, the specific operation of the random forest algorithm in step S2 is to combine multiple decision trees. Each time the data subset is randomly selected with replacement, and at the same time, some features are randomly selected as input.

Furthermore, the random forest algorithm in step S2 is a bagging algorithm using decision trees as estimators. The final results are statistically calculated by the combiner. In the classification problem, the combiner selects the majority of the classification results as the final result. In the regression problem , average multiple regression results as the final result.

Further, in step S2, the ranking of influencing factors is based on importance and follows the ordering rule of increasing importance. The higher the ranking result of the influencing factors, the higher the importance of the influencing factors. On the contrary, The lower the ranking result is, the lower the importance of the influencing factor.

Further, the selection of the main influencing factors in step S2 is based on the ranking results of the influencing factors, and the main influencing factors refer to the influencing factors with the highest ranking results and high importance.

Further, the carbon emission prediction formula in step S3 is as follows:

Among them, C is the carbon emission prediction result, i=1, 2, 3,..., n, k is the importance parameter, P _i is the carbon emission coefficient of the i-th main influencing factor, and D _i is the i-th main influencing factor. consumption.

Further, the cluster analysis operation in step S4 is to classify the preliminary prediction and impact prediction data into different classes or clusters, and then analyze the classification results in combination, and the classes required for clustering are unknown.

Further, the operation of association rule analysis in step S4 is to find frequent patterns, associations, correlations or causal structures that exist between the preliminary prediction and the impact prediction in the data of the preliminary prediction and the impact prediction.

The present invention provides a carbon emission prediction method based on energy big data, which has the following beneficial effects: The present invention determines the factors affecting carbon emissions in the energy industry through analysis, and connects the energy information database to retrieve data to establish energy information data sets and different Information subsets of the categories of factors affecting carbon emissions, and use the random forest algorithm to identify the importance of the influencing factors, then sort and select the main influencing factors, generate a feature matrix, build a carbon emission prediction model, conduct model training, and train Then make a preliminary prediction of carbon emissions, then adjust the parameters of the influencing factors, simulate the impact and predict the carbon emissions, study the relationship between the preliminary prediction of carbon emissions and the impact prediction, analyze the impact and change patterns of the influencing factors, and perform cluster analysis and After analyzing the association rules, establish a mathematical equation relationship, substitute the energy data into the carbon emission prediction model to calculate the preliminary prediction results of carbon emissions, and then use the mathematical equation relationship to correct the preliminary prediction results to obtain the final carbon emission prediction results. After improvement The carbon emission prediction method not only builds a carbon emission prediction model, but also simulates the impact of various influencing factors on the prediction model under different circumstances to avoid the impact of differences in carbon emission influencing factors in different fields and different equipment on the prediction model, thereby Avoid interference with carbon emission predictions and improve the accuracy and reliability of carbon emission prediction results.

Description of drawings

Figure 1 is a schematic diagram of the overall operation flow of a carbon emission prediction method based on energy big data according to the present invention;

Figure 2 is a schematic diagram of the operation flow of step S1 of a carbon emission prediction method based on energy big data according to the present invention;

Figure 3 is a schematic diagram of the operation flow of step S2 of a carbon emission prediction method based on energy big data according to the present invention;

Figure 4 is a schematic diagram of the operation flow of step S3 of a carbon emission prediction method based on energy big data according to the present invention;

Figure 5 is a schematic diagram of the operation flow of step S4 of a carbon emission prediction method based on energy big data according to the present invention;

Figure 6 is a schematic diagram of the operation flow of step S5 of a carbon emission prediction method based on energy big data of the present invention.

Detailed ways

The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the invention but are not intended to limit the scope of the invention.

As shown in Figures 1-6, a carbon emission prediction method based on energy big data includes the following specific steps:

S1. Data integration of factors affecting carbon emissions:

Analyze and determine the factors affecting carbon emissions in the energy industry, connect to the energy information database, retrieve relevant information and data on factors affecting carbon emissions in the database, and establish an energy information data set. Set multiple subsets in the energy information data set to combine different carbon emissions. The information data of influencing factors is placed inside the subset, and information subsets of different categories of carbon emission influencing factors are established. The information data of carbon emission influencing factors in the same information subset are independent of each other and do not interfere with each other. The impact of carbon emissions in different information subsets There is no overlap or interaction in factor information data;

Energy information data set {Edb_Co2}={ΔCo2_1}+{ΔCo2_2}+{ΔCo2_3}+…+{ΔCo2_n}, where Co2_1, Co2_2, Co2_3,…, Co2_n respectively represent different carbon emission influencing factors, {ΔCo2_1}, {ΔCo2_2}, {ΔCo2_3}, ..., {ΔCo2_n} respectively represent the information subsets corresponding to different influencing factors;

S2. Construction of carbon emission prediction model:

Clean and correct the data in the data subset, and use the random forest algorithm to determine the importance of influencing factors, sort the influencing factors according to their importance, select the main influencing factors, generate a feature matrix, and construct a carbon emission prediction Model;

The specific operation of the random forest algorithm is to combine multiple decision trees. Each time the data subset is randomly selected with replacement, and at the same time, some features are randomly selected as input; the random forest algorithm uses decision trees as estimators. In the bagging algorithm, the final results are counted through the combiner. In the classification problem, the combiner selects the majority of the classification results as the final result. In the regression problem, the average of multiple regression results is taken as the final result;

The ranking of influencing factors is based on importance, and follows the ordering rules from largest to smallest. The higher the ranking result of the influencing factors, the higher the importance of the influencing factors. On the contrary, the lower the ranking result, the higher the importance of the influencing factors. The lower the importance of the influencing factors; the selection of the main influencing factors is based on the ranking results of the influencing factors, and the main influencing factors refer to the influencing factors with high ranking results and high importance;

S3. Predictive model training and impact simulation:

The carbon emission prediction model is trained according to the characteristic matrix to test the carbon emission prediction model, and after training, a preliminary prediction of carbon emissions is made based on the input values of each influencing factor, and then the parameters of the influencing factors are adjusted to simulate Be affected by different situations and conduct carbon emission predictions;

The carbon emission prediction formula is as follows:

Among them, C is the carbon emission prediction result, i=1, 2, 3,..., n, k is the importance parameter, P _i is the carbon emission coefficient of the i-th main influencing factor, and D _i is the i-th main influencing factor. consumption;

S4. Impact analysis of carbon emission forecast:

Study the relationship between preliminary prediction of carbon emissions and impact prediction, combine energy data to analyze the impact and change patterns of influencing factors, determine the internal correlation between the two through cluster analysis and association rule analysis, draw a causal relationship diagram, and establish mathematical equations relation;

The operation of cluster analysis is to classify the data of preliminary prediction and impact prediction into different classes or clusters, and then combine the classification results for analysis, and the classes required for clustering are unknown; the operation of association rule analysis is to perform preliminary prediction and shock prediction data, looking for frequent patterns, associations, correlations, or causal structures that exist between preliminary forecasts and shock predictions;

S5. Carbon emission prediction and result correction:

The energy data in the energy information database is retrieved, substituted into the carbon emission prediction model to calculate the preliminary prediction results of carbon emissions, and then the preliminary prediction results are corrected using mathematical equations to remove the interference caused by the impact of influencing factors on the carbon emission prediction, and the calculation The final carbon emission prediction results are obtained.

In summary, as shown in Figures 1 to 6, the carbon emission prediction method based on energy big data includes the following specific steps when used:

S1. Data integration of factors affecting carbon emissions:

Analyze and determine the factors affecting carbon emissions in the energy industry, connect to the energy information database, retrieve relevant information and data on factors affecting carbon emissions in the database, and establish an energy information data set. Set multiple subsets in the energy information data set to combine different carbon emissions. The information data of influencing factors are placed inside the subsets to establish information subsets of different categories of carbon emission influencing factors; the information data of carbon emission influencing factors in the same information subset are independent of each other and do not interfere with each other, and the carbon emission impacts in different information subsets There is no overlap or interaction in factor information data;

Energy information data set {Edb_Co2}={ΔCo2_1}+{ΔCo2_2}+{ΔCo2_3}+…+{ΔCo2_n}, where Co2_1, Co2_2, Co2_3,…, Co2_n respectively represent different carbon emission influencing factors, {ΔCo2_1}, {ΔCo2_2}, {ΔCo2_3}, ..., {ΔCo2_n} respectively represent the information subsets corresponding to different influencing factors;

S2. Construction of carbon emission prediction model:

Clean and correct the data in the data subset, and use the random forest algorithm to determine the importance of influencing factors. The specific operation of the random forest algorithm is to combine multiple decision trees together. Each data subset is randomly placed At the same time, some features are randomly selected as input; and the random forest algorithm is a Bagging algorithm that uses decision trees as estimators. The final results are statistically calculated through the combiner. In the classification problem, the combiner selects the majority of the classification results as the final result. In the regression problem, the average of multiple regression results is taken as the final result, and the influencing factors are sorted according to their importance. The ranking of the influencing factors is based on their importance, and follows the order from the largest to the smallest. According to the rules, the higher the ranking result of influencing factors is, the higher the importance of the influencing factor is. On the contrary, the lower the ranking result is, the lower the importance of the influencing factor is. Select the main influencing factors and select the main influencing factors. Based on the ranking results of influencing factors, and the main influencing factors refer to the influencing factors with high ranking results and high importance, a feature matrix is generated and a carbon emission prediction model is constructed;

S3. Predictive model training and impact simulation:

The carbon emission prediction model is trained according to the characteristic matrix to test the carbon emission prediction model, and after training, a preliminary prediction of carbon emissions is made based on the input values of each influencing factor, and then the parameters of the influencing factors are adjusted to simulate Be affected by different situations and conduct carbon emission predictions;

The carbon emission prediction formula is as follows:

Among them, C is the carbon emission prediction result, i=1, 2, 3,..., n, k is the importance parameter, P _i is the carbon emission coefficient of the i-th main influencing factor, and D _i is the i-th main influencing factor. consumption;

S4. Impact analysis of carbon emission forecast:

Study the relationship between the preliminary prediction and impact prediction of carbon emissions, and analyze the impact and change patterns of influencing factors based on energy data. Through cluster analysis and association rule analysis, the operation of cluster analysis is to classify the data of preliminary prediction and impact prediction into Different classes or clusters are analyzed together with the classification results, and the classes required for clustering are unknown. The operation of association rule analysis is to find the data that exists between the preliminary prediction and impact prediction. Frequent patterns, associations, correlations or causal structures between the two, determine the internal correlation between the two, draw a causal relationship diagram, and establish a mathematical equation relationship;

S5. Carbon emission prediction and result correction:

The energy data in the energy information database is retrieved, substituted into the carbon emission prediction model to calculate the preliminary prediction results of carbon emissions, and then the preliminary prediction results are corrected using mathematical equations to remove the interference caused by the impact of influencing factors on the carbon emission prediction, and the calculation The final carbon emission prediction results are obtained.

This invention determines the factors affecting carbon emissions in the energy industry through analysis, and connects the energy information database to retrieve data to establish an energy information data set and information subsets of different categories of carbon emission influencing factors, and determines the importance of the influencing factors through a random forest algorithm. Discriminate, then sort and select the main influencing factors, generate a feature matrix, build a carbon emission prediction model, conduct model training, and make a preliminary prediction of carbon emissions after training, then adjust the parameters of the influencing factors, simulate the impact and conduct carbon emissions Forecasting, study the relationship between preliminary predictions of carbon emissions and impact predictions, analyze the impact and change patterns of influencing factors, establish mathematical equation relationships after cluster analysis and association rule analysis, and substitute energy data into the carbon emission prediction model to calculate The preliminary prediction results of carbon emissions are then corrected using mathematical equations to obtain the final carbon emission prediction results. The improved carbon emission prediction method not only builds a carbon emission prediction model, but also simulates the impact of various influencing factors on different The impact of different circumstances on the prediction model can be avoided, and the impact of differences in carbon emission influencing factors in different fields and different equipment on the prediction model can be avoided, so as to avoid interference with carbon emission prediction and improve the accuracy and reliability of carbon emission prediction results.

The embodiments of the present invention are presented for purposes of illustration and description, and are not intended to be exhaustive or to limit the invention to the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiment was chosen and described in order to best explain the principles of the invention and the practical application, and to enable others of ordinary skill in the art to understand the invention and design various embodiments with various modifications as are suited to the particular use contemplated.

Claims (10)

1. A carbon emission prediction method based on energy big data, which is characterized by including the following specific steps: S1. Data integration of factors affecting carbon emissions: Analyze and determine the factors affecting carbon emissions in the energy industry, connect to the energy information database, retrieve relevant information and data on factors affecting carbon emissions in the database, and establish an energy information data set. Set multiple subsets in the energy information data set to combine different carbon emissions. The influencing factor information data is placed inside the subset, and information subsets of different categories of carbon emission influencing factors are established; S2. Construction of carbon emission prediction model: Clean and correct the data in the data subset, and use the random forest algorithm to determine the importance of influencing factors, sort the influencing factors according to their importance, select the main influencing factors, generate a feature matrix, and construct a carbon emission prediction Model; S3. Predictive model training and impact simulation: The carbon emission prediction model is trained according to the characteristic matrix to test the carbon emission prediction model, and after training, a preliminary prediction of carbon emissions is made based on the input values of each influencing factor, and then the parameters of the influencing factors are adjusted to simulate Be affected by different situations and conduct carbon emission predictions; S4. Impact analysis of carbon emission forecast: Study the relationship between preliminary prediction of carbon emissions and impact prediction, combine energy data to analyze the impact and change patterns of influencing factors, determine the internal correlation between the two through cluster analysis and association rule analysis, draw a causal relationship diagram, and establish mathematical equations relation; S5. Carbon emission prediction and result correction: The energy data in the energy information database is retrieved, substituted into the carbon emission prediction model to calculate the preliminary prediction results of carbon emissions, and then the preliminary prediction results are corrected using mathematical equations to remove the interference caused by the impact of influencing factors on the carbon emission prediction, and the calculation The final carbon emission prediction results are obtained. 2. A carbon emission prediction method based on energy big data according to claim 1, characterized in that in step S1, the energy information data set {Edb_Co2}={ΔCo2_1}+{ΔCo2_2}+{ΔCo2_3}+ …+{ΔCo2_n}, where Co2_1, Co2_2, Co2_3,…, Co2_n respectively represent different influencing factors of carbon emissions, {ΔCo2_1}, {ΔCo2_2}, {ΔCo2_3},…, {ΔCo2_n} respectively represent the corresponding information subset. 3. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the carbon emission influencing factor information data in the same information subset in step S1 are independent of each other and do not interfere with each other. There is no overlap or interaction in the information data of carbon emission influencing factors within the information subset. 4. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the specific operation of the random forest algorithm in step S2 is to combine multiple decision trees together. The set is randomly selected with replacement, and some features are randomly selected as input. 5. A carbon emission prediction method based on energy big data according to claim 4, characterized in that the random forest algorithm in step S2 is a bagging algorithm using a decision tree as an estimator, and the final result is processed by a combiner Statistics, the combiner selects the majority classification result as the final result in classification problems, and averages multiple regression results as the final result in regression problems. 6. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the ranking of influencing factors in step S2 is based on importance and follows the ranking from largest to smallest importance. According to the rule, the higher the ranking result of an influencing factor is, the higher the importance of the influencing factor is. On the contrary, the lower the ranking result is, the lower the importance of the influencing factor is. 7. A carbon emission prediction method based on energy big data according to claim 6, characterized in that the selection of the main influencing factors in step S2 is based on the ranking results of the influencing factors, and the main influencing factors refer to the ranking results. The first and most important influencing factors. 8. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the carbon emission prediction formula in step S3 is as follows: Among them, C is the carbon emission prediction result, i=1, 2, 3,..., n, k is the importance parameter, P _i is the carbon emission coefficient of the i-th main influencing factor, and D _i is the i-th main influencing factor. consumption. 9. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the cluster analysis operation in step S4 is to classify the preliminary prediction and impact prediction data into different classes or Clusters are combined with the classification results for analysis, and the classes required for clustering are unknown. 10. A carbon emission prediction method based on energy big data according to claim 1, characterized in that the operation of association rule analysis in step S4 is to search for the data that exists in the preliminary prediction and impact prediction. Frequent patterns, associations, correlations, or causal structures between forecasts and shock predictions.