Prediction method and system combining deep learning and ARIMA model

Abstract

The invention is suitable for the technical field of enterprise carbon emission prediction, and provides a deep learning and ARI MA model combined prediction method and system, and the method comprises the steps: collecting the historical data of X sample enterprises, comprising enterprise historical carbon emission data, shipment quantity data, consumption and proportion data of different energy sources, price data of different energy sources, temperature change and other monthly data; using a KNN algorithm to classify the enterprises; and estimating a carbon emission range of a to-be-classified enterprise, classifying the enterprise by applying the KNN algorithm according to a part of historical data of the to-be-classified enterprise, and estimating the carbon emission range of the to-be-classified enterprise or predicting the carbon emission of the to-be-classified enterprise according to the category of the to-be-classified enterprise. According to the prediction method based on the KNN algorithm in combination with deep learning and the ARI MA model, the prediction accuracy and practicability can be improved, and a more reliable decision basis is provided for an enterprise to realize carbon emission management.

Description

Technical Field

[0001] The present invention belongs to the technical field of enterprise carbon emission prediction, and in particular relates to a prediction method and system combining deep learning and ARIMA model. Background Art

[0002] As the world pays more and more attention to climate change, the monitoring and management of corporate carbon emissions has become the focus of governments and the public. In order to achieve sustainable development goals, companies are seeking effective carbon emission forecasting methods to formulate scientific and reasonable emission reduction strategies. However, the carbon emission forecasting methods currently used in the market have certain limitations, which brings challenges to companies in providing accurate carbon emission forecasts.

[0003] The carbon emission prediction methods currently used are usually based on traditional statistical models such as linear regression and time series analysis. These methods have their advantages in data processing and result analysis, but they also face the following shortcomings: These methods usually assume that the data has a linear relationship, while in reality, carbon emissions are affected by many factors and may show nonlinear characteristics, resulting in the model being unable to fully capture complex behavior patterns. When dealing with different types of enterprises or industries, traditional methods require readjustment of model parameters and lack efficient versatility. Summary of the invention

[0004] In view of this, the present invention provides a prediction method and system combining deep learning and ARIMA model to solve the technical problems that the prior art cannot well predict the carbon emission data of the enterprise to be analyzed and cannot provide carbon emission reduction recommendations.

[0005] A first aspect of the present invention provides a prediction method combining deep learning and an ARIMA model, the prediction method combining deep learning and an ARIMA model comprising:

[0006] Collect historical data of X sample companies, including historical carbon emissions data, shipment data, consumption and proportion of different energy sources, price data of different energy sources, and monthly data such as temperature changes;

[0007] Collect the company's historical carbon emission data, including time series data, and record the carbon emission value at each time point. Collect the company's shipment data, energy consumption combination data (such as the consumption and proportion of different energy sources), and temperature change data for relevant time periods. This will provide a comprehensive understanding of the company's energy consumption: By collecting the consumption and proportion data of different energy sources, you can clearly understand the company's dependence on various energy sources, providing a basis for subsequent analysis and decision-making.

[0008] Enterprises are classified using the KNN algorithm, wherein the classification features of the KNN algorithm include at least energy prices and time factors; using the KNN algorithm to classify enterprises and considering energy prices and time factors as classification features can bring more accurate classification results, personalized analysis and suggestions, which can help enterprises optimize energy management, reduce risks, enhance competitiveness, and promote sustainable development.

[0009] Furthermore, the use of the KNN algorithm to classify enterprises specifically includes:

[0010] Mark the historical data of X sample companies as data points A = (a1, a2, ..., a n ), B=(b1,b2,...,b n )..., where a i and b i They are the values ​​of data points A and B on the i-th feature respectively. Each category of historical data of an enterprise can be used as a feature, such as the enterprise’s historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and temperature changes.

[0011] Calculate the distance between the data points. For data point A = (a1, a2, ..., a n ) and data points B=(b1,b2,...,b n ), the distance between them is:

[0012]

[0013] The K data points closest to each other are grouped into the same category, and the category to which the data point belongs is determined based on the majority voting principle, where K = 0.2 × X. By setting an appropriate clustering coefficient, the X companies in the sample data can be clustered into roughly five categories, so that the companies are not subdivided too finely, and the carbon emission characteristics of different companies cannot be effectively distinguished and classified.

[0014] Furthermore, based on the above KNN algorithm, the carbon emission range of the enterprise to be classified can be estimated, which specifically includes: classifying the enterprise according to some historical data of the enterprise to be classified by applying the KNN algorithm, and estimating the carbon emission range of the enterprise to be classified or predicting the carbon emission of the enterprise to be classified according to the category of the enterprise to be classified.

[0015] Estimation based on historical data and KNN algorithm provides a data-driven decision-making method. This can avoid subjective guesswork and blind decision-making, and improve the accuracy and reliability of decision-making. By estimating the carbon emission range of the enterprise to be classified or predicting the specific carbon emissions, it provides a reference for the enterprise on its carbon emissions. This helps enterprises understand their own carbon emission levels, compare with other enterprises in the same industry, and formulate corresponding emission reduction targets and strategies.

[0016] Furthermore, it also includes analyzing the sensitivity of different types of enterprises in the X sample enterprises to temperature and time. The specific calculation method is:

[0017]

[0018] Among them, the temperature data is the weekly or monthly average temperature, x i is the data corresponding to temperature or time in the historical data of each sample enterprise, is the average value of the data related to temperature or time in the historical data of all sample enterprises, y i is the carbon emission data in the historical data of each sample enterprise, The average value of carbon emissions in the historical data of each sample enterprise.

[0019] Identifying enterprises that are sensitive to weekly or monthly temperatures and time can prompt them to take appropriate measures to improve energy efficiency and deal with risks such as energy supply interruptions or price fluctuations. For example, for temperature-sensitive enterprises, energy consumption can be reduced by improving insulation and heat preservation measures; for time-sensitive enterprises, the operating time and load of equipment can be optimized to improve energy efficiency. If energy supply may be in short supply or price increases in a specific time period, measures such as energy reserve and production plan adjustment can be taken in advance to reduce risks.

[0020] Furthermore, it also includes determining the sensitivity of different categories of enterprises to temperature and time based on the calculation results of the correlation coefficient, wherein when the correlation coefficient r>0.7 or r<-0.7, the enterprises in this category are considered to be temperature and time sensitive enterprises.

[0021] The value range of the Pearson correlation coefficient is from -1 to 1, and its absolute value indicates the strength of the correlation. Generally, the correlation is judged according to the following common threshold range: When the absolute value of the Pearson correlation coefficient is greater than or equal to 0.7, it is generally considered that there is a strong correlation between the two variables. That is, when the correlation coefficient is greater than or equal to 0.7 or less than or equal to -0.7, it can be determined that there is a significant linear relationship between the variables.

[0022] Furthermore, enterprises that are sensitive to temperature and time are taken as enterprises to be classified, and an ARIMA model is established to predict the carbon emission values ​​of the enterprises to be classified on a quarterly basis, including:

[0023]

[0024] Among them, μ is a constant term, is the predicted carbon emission at time t, y t-i is the observed value of carbon emissions at a past time point, and θ j is the model parameter, ∈ t is the error term. Try different combinations of ARIMA model parameters p (autoregressive order) and q (moving average order), where p is an integer from 1 to 6 and q is an integer from 1 to 4. Calculate the likelihood function value L and Akaike information criterion AIC = -2 ln(L) + 6 of the ARIMA model under different value combinations. The combination of p and q with the minimum AIC is used as the value of the parameters p and q of the ARIMA model.

[0025] AIC comprehensively considers the goodness of fit and complexity of the model. Generally, different combinations of p and q are tried, and ARIMA models are established and the corresponding AIC values ​​are calculated. The p and q combination with the smallest AIC value is selected as the optimal parameter choice.

[0026] The Pearson correlation coefficient can give a specific value to indicate the degree of linear correlation between an enterprise and temperature and time. This allows the correlation to be accurately measured and compared, which helps to more objectively evaluate the relationship between them. By calculating the correlation coefficient, it can be determined whether there is a trend of positive or negative correlation between the enterprise's carbon emissions and temperature and time. A positive correlation means that as the temperature rises or time passes, the enterprise's carbon emissions increase; a negative correlation indicates the opposite trend. For example, if it is found that the enterprise is highly correlated with the temperature, measures can be considered to reduce the dependence on the temperature, such as improving insulation and adopting energy-saving technologies.

[0027] Furthermore, it also includes energy consumption analysis for classified enterprises, including:

[0028] Use the KNN algorithm to retrieve similar enterprises to the enterprise to be classified;

[0029] Based on the ARIMA, the predicted carbon emission values ​​of the enterprises to be classified are obtained, and the energy consumption data of the 20% enterprises with the lowest carbon emission values ​​among the same type of enterprises are analyzed to obtain a low-carbon emission energy combination, wherein the energy consumption data analysis includes energy consumption and proportion data analysis;

[0030] Compare the energy consumption data of the enterprise to be classified with the low-carbon emission energy consumption combination, and output energy consumption improvement suggestions on a quarterly basis.

[0031] Based on the ARIMA model and energy data analysis, improvement suggestions are data-driven. This can provide more objective and accurate information, helping enterprise management make more informed decisions and develop more effective energy management strategies. By analyzing the 20% of companies with the lowest carbon emissions among similar companies, best practices and benchmark companies in the industry can be identified. Companies to be classified can learn from the experience of these companies and their energy strategies and technologies to improve their own energy management level.

[0032] Furthermore, the suggestions for improving energy consumption by quarterly output include:

[0033] The energy consumption data analysis includes analyzing the energy consumption and proportion data of the similar enterprises in the historical quarters to obtain the average values ​​of different types of energy consumption of the similar enterprises in the corresponding quarters;

[0034] Calculate the difference between the different types of energy consumption of the enterprise to be classified in the forecast quarter and the different types of energy consumption of the same type of enterprise in the corresponding quarter;

[0035] Energy improvement suggestions are output based on the difference, and the most similar enterprise after each energy consumption adjustment is retrieved based on the KNN algorithm, and the carbon emission reduction of the most similar enterprise in the corresponding quarter is used as the estimated value of the carbon emission reduction of the enterprise to be classified.

[0036] By analyzing the energy consumption and proportion data of similar enterprises in historical quarters, the average values ​​of different types of energy consumption are obtained. Then the difference between the energy consumption of the predicted quarter of the enterprise to be classified and the corresponding quarter of similar enterprises is calculated, and energy improvement suggestions are output based on the difference. This method can provide accurate energy-saving suggestions for specific energy consumption types and help enterprises better control energy costs. The improvement suggestions are based on the analysis and calculation of historical data of similar enterprises and are a data-driven decision-making process. This can avoid subjective judgment bias and improve the accuracy and reliability of decision-making.

[0037] A second aspect of an embodiment of the present invention provides a prediction system combining deep learning and an ARIMA model, including:

[0038] The collection unit is used to collect historical data of X sample companies, including historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and monthly data such as temperature changes;

[0039] A classification unit, used for classifying enterprises using a KNN algorithm, wherein the classification features of the KNN algorithm include at least energy prices and time factors;

[0040] The estimation unit is used to classify the enterprises by applying the KNN algorithm according to part of the historical data of the enterprises to be classified, and to estimate the carbon emission range of the enterprises to be classified or predict the carbon emission of the enterprises to be classified according to the categories of the enterprises to be classified.

[0041] The third aspect of the present invention provides a terminal device, comprising: a data input module, a calculation and analysis module, a memory, a processor, and an enterprise carbon emission prediction program stored in the memory and executable on the processor, wherein the enterprise carbon emission prediction program is configured to implement the steps in the prediction method combining deep learning and ARIMA model as described in any one of the first aspects.

[0042] The fourth aspect of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the prediction method combining deep learning and ARIMA model described in the first aspect.

[0043] Compared with the prior art, the present invention has the following beneficial effects: the prediction method based on the KNN algorithm combined with deep learning and the ARIMA model can improve the accuracy and practicality of the prediction, and provide a more reliable decision-making basis for enterprises to achieve carbon emission management. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 A schematic flow chart of a prediction method combining deep learning and ARIMA model provided by the present invention is shown;

[0046] Figure 2 A schematic diagram of a prediction system combining deep learning and an ARIMA model provided by an embodiment of the present invention is shown;

[0047] Figure 3 A schematic diagram of a terminal device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0048] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0049] The first aspect of the embodiment of the present invention provides a prediction method and system combining deep learning and ARIMA model, see Figure 1 , Figure 1 A schematic flow chart of a prediction method combining deep learning and ARIMA model provided by the present invention is shown. The prediction method combining deep learning and ARIMA model includes:

[0050] Step 101: Collect historical data of X sample companies, including historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and monthly data such as temperature changes;

[0051] Collect the company's historical carbon emission data, including time series data, and record the carbon emission value at each time point. Collect the company's shipment data, energy consumption combination data (such as the consumption and proportion of different energy sources), and temperature change data for relevant time periods. This will provide a comprehensive understanding of the company's energy consumption: By collecting the consumption and proportion data of different energy sources, you can clearly understand the company's dependence on various energy sources, providing a basis for subsequent analysis and decision-making.

[0052] Step 102: Use the KNN algorithm to classify the enterprises, wherein the classification features of the KNN algorithm include at least energy price and time factor; using the KNN algorithm to classify the enterprises and considering energy price and time factor as classification features can bring more accurate classification results, personalized analysis and suggestions, which will help enterprises optimize energy management, reduce risks, enhance competitiveness, and promote sustainable development.

[0053] K-Nearest Neighbors (KNN) is a basic machine learning algorithm that is suitable for classification and regression problems, especially when the data distribution has no obvious regularity and the training sample is small. It determines the category of the sample to be predicted by calculating the distance between the sample to be predicted and the samples in the training set in the feature space. The KNN algorithm is a non-parametric learning algorithm that does not require assumptions about the data. The KNN algorithm can quickly determine the range of a company's carbon emissions or predict specific carbon emissions based on the above factors.

[0054] Furthermore, the use of the KNN algorithm to classify enterprises specifically includes:

[0055] Mark the historical data of X sample companies as data points A = (a1, a2, ..., a n ), B=(b1,b2,...,b n )..., where ai and bi are the values ​​of data points A and B on the i-th feature respectively. Each category of enterprise historical data can be used as a feature, such as enterprise historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and temperature changes.

[0056] Calculate the distance between the data points. For data point A = (a1, a2, ..., a n ) and data points B=(b1,b2,...,b n ), the distance between them is:

[0057]

[0058] The K data points closest to each other are grouped into the same category, and the category to which the data point belongs is determined based on the majority voting principle, where K = 0.2 × X. By setting an appropriate clustering coefficient, the X companies in the sample data can be clustered into roughly five categories, so that the companies are not subdivided too finely, and the carbon emission characteristics of different companies cannot be effectively distinguished and classified.

[0059] Step 103: Further, based on the above KNN algorithm, the carbon emission range of the enterprise to be classified can be estimated. Specifically, it includes: according to part of the historical data of the enterprise to be classified, applying the KNN algorithm to classify the enterprise, and estimating the carbon emission range of the enterprise to be classified or predicting the carbon emission of the enterprise to be classified according to the category of the enterprise to be classified.

[0060] Estimation based on historical data and KNN algorithm provides a data-driven decision-making method. This can avoid subjective guesswork and blind decision-making, and improve the accuracy and reliability of decision-making. By estimating the carbon emission range of the enterprise to be classified or predicting the specific carbon emissions, it provides a reference for the enterprise on its carbon emissions. This helps enterprises understand their own carbon emission levels, compare with other enterprises in the same industry, and formulate corresponding emission reduction targets and strategies.

[0061] Furthermore, it also includes analyzing the sensitivity of different types of enterprises in the X sample enterprises to temperature and time. The specific calculation method is:

[0062]

[0063] Among them, the temperature data is the weekly or monthly average temperature, x iis the data corresponding to temperature or time in the historical data of each sample enterprise, is the average value of the data related to temperature or time in the historical data of all sample enterprises, y i is the carbon emission data in the historical data of each sample enterprise, The average value of carbon emissions in the historical data of each sample enterprise.

[0064] Identifying enterprises that are sensitive to weekly or monthly temperatures and time can prompt them to take appropriate measures to improve energy efficiency and deal with risks such as energy supply interruptions or price fluctuations. For example, for temperature-sensitive enterprises, energy consumption can be reduced by improving insulation and heat preservation measures; for time-sensitive enterprises, the operating time and load of equipment can be optimized to improve energy efficiency. If energy supply may be in short supply or price increases in a specific time period, measures such as energy reserve and production plan adjustment can be taken in advance to reduce risks.

[0065] Furthermore, it also includes determining the sensitivity of different categories of enterprises to temperature and time based on the calculation results of the correlation coefficient, wherein when the correlation coefficient r>0.7 or r<-0.7, the enterprises in this category are considered to be temperature and time sensitive enterprises.

[0066] The value range of the Pearson correlation coefficient is from -1 to 1, and its absolute value indicates the strength of the correlation. Generally, the correlation is judged according to the following common threshold range: When the absolute value of the Pearson correlation coefficient is greater than or equal to 0.7, it is generally considered that there is a strong correlation between the two variables. That is, when the correlation coefficient is greater than or equal to 0.7 or less than or equal to -0.7, it can be determined that there is a significant linear relationship between the variables.

[0067] Furthermore, enterprises that are sensitive to temperature and time are taken as enterprises to be classified, and an ARIMA model is established to predict the carbon emission values ​​of the enterprises to be classified on a quarterly basis, including:

[0068]

[0069] Among them, μ is a constant term, is the predicted carbon emission at time t, y t-i is the observed value of carbon emissions at a past time point, and θ j is the model parameter, ∈ tis the error term. Try different combinations of ARIMA model parameters p (autoregressive order) and q (moving average order), where p is an integer from 1 to 6 and q is an integer from 1 to 4. Calculate the likelihood function value L and Akaike information criterion AIC = -2ln(L) + 6 of the ARIMA model under different value combinations. The combination of p and q with the minimum AIC is used as the value of the parameters p and q of the ARIMA model.

[0070] The ARIMA model is suitable for data with obvious time correlation and certain trends, and can be used to predict values ​​at future time points. The ARIMA model makes predictions by considering the autocorrelation and trend of time series data.

[0071] AIC comprehensively considers the goodness of fit and complexity of the model. Generally, different combinations of p and q are tried, and ARIMA models are established and the corresponding AIC values ​​are calculated. The p and q combination with the smallest AIC value is selected as the optimal parameter choice.

[0072] The Pearson correlation coefficient can give a specific value to indicate the degree of linear correlation between an enterprise and temperature and time. This allows the correlation to be accurately measured and compared, which helps to more objectively evaluate the relationship between them. By calculating the correlation coefficient, it can be determined whether there is a trend of positive or negative correlation between the enterprise's carbon emissions and temperature and time. A positive correlation means that as the temperature rises or time passes, the enterprise's carbon emissions increase; a negative correlation indicates the opposite trend. For example, if it is found that the enterprise is highly correlated with the temperature, measures can be considered to reduce the dependence on the temperature, such as improving insulation and adopting energy-saving technologies.

[0073] Step 104: further includes performing energy consumption analysis on the classified enterprises, specifically including:

[0074] Use the KNN algorithm to retrieve similar enterprises to the enterprise to be classified;

[0075] Based on the ARIMA, the predicted carbon emission values ​​of the enterprises to be classified are obtained, and the energy consumption data of the 20% enterprises with the lowest carbon emission values ​​among the same type of enterprises are analyzed to obtain a low-carbon emission energy combination, wherein the energy consumption data analysis includes energy consumption and proportion data analysis;

[0076] Compare the energy consumption data of the enterprise to be classified with the low-carbon emission energy consumption combination, and output energy consumption improvement suggestions on a quarterly basis.

[0077] Based on the ARIMA model and energy data analysis, improvement suggestions are data-driven. This can provide more objective and accurate information, helping enterprise management make more informed decisions and develop more effective energy management strategies. By analyzing the 20% of companies with the lowest carbon emissions among similar companies, best practices and benchmark companies in the industry can be identified. Companies to be classified can learn from the experience of these companies and their energy strategies and technologies to improve their own energy management level.

[0078] Furthermore, the suggestions for improving energy consumption by quarterly output include:

[0079] The energy consumption data analysis includes analyzing the energy consumption and proportion data of the similar enterprises in the historical quarters to obtain the average values ​​of different types of energy consumption of the similar enterprises in the corresponding quarters;

[0080] Calculate the difference between the different types of energy consumption of the enterprise to be classified in the forecast quarter and the different types of energy consumption of the same type of enterprise in the corresponding quarter;

[0081] Energy improvement suggestions are output based on the difference, and the most similar enterprise after each energy consumption adjustment is retrieved based on the KNN algorithm, and the carbon emission reduction of the most similar enterprise in the corresponding quarter is used as the estimated value of the carbon emission reduction of the enterprise to be classified.

[0082] By analyzing the energy consumption and proportion data of similar enterprises in historical quarters, the average values ​​of different types of energy consumption are obtained. Then the difference between the energy consumption of the predicted quarter of the enterprise to be classified and the corresponding quarter of similar enterprises is calculated, and energy improvement suggestions are output based on the difference. This method can provide accurate energy-saving suggestions for specific energy consumption types and help enterprises better control energy costs. The improvement suggestions are based on the analysis and calculation of historical data of similar enterprises and are a data-driven decision-making process. This can avoid subjective judgment bias and improve the accuracy and reliability of decision-making.

[0083] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: the prediction method based on the KNN algorithm combined with deep learning and the ARIMA model has significant advantages in solving the shortcomings of traditional methods, but it is still necessary to pay attention to the problems of sample selection and calculation efficiency. By further optimizing the algorithm and data processing flow, the accuracy and practicality of the prediction can be improved, providing a more reliable decision-making basis for enterprises to achieve carbon emission management.

[0084] like Figure 2 A second aspect of the embodiment of the present invention provides a prediction system combining deep learning and ARIMA model, see Figure 2 , Figure 2 A schematic diagram of a prediction system combining deep learning and ARIMA model provided by the present invention is shown, Figure 2 A forecasting system combining deep learning and ARIMA model is shown, including:

[0085] The collection unit 21 is used to collect historical data of X sample companies, including historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and monthly data such as temperature changes;

[0086] A classification unit 22, used to classify enterprises using a KNN algorithm, wherein the classification features of the KNN algorithm at least include energy prices and time factors;

[0087] The estimation unit 23 is used to classify the enterprises by applying the KNN algorithm according to part of the historical data of the enterprises to be classified, and estimate the carbon emission range of the enterprises to be classified or predict the carbon emission of the enterprises to be classified according to the categories of the enterprises to be classified.

[0088] The collection unit 21 is used to collect historical data of X sample companies, including historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and monthly data such as temperature changes;

[0089] The classification unit 22 is used to classify enterprises using the KNN algorithm, wherein the classification features of the KNN algorithm include at least energy prices and time factors; using the KNN algorithm to classify enterprises and considering energy prices and time factors as classification features can bring more accurate classification results, personalized analysis and suggestions, which will help enterprises optimize energy management, reduce risks, enhance competitiveness, and promote sustainable development.

[0090] K-Nearest Neighbors (KNN) is a basic machine learning algorithm that is suitable for classification and regression problems, especially when the data distribution has no obvious regularity and the training sample is small. It determines the category of the sample to be predicted by calculating the distance between the sample to be predicted and the samples in the training set in the feature space. The KNN algorithm is a non-parametric learning algorithm that does not require assumptions about the data. The KNN algorithm can quickly determine the range of a company's carbon emissions or predict specific carbon emissions based on the above factors.

[0091] Furthermore, the use of the KNN algorithm to classify enterprises specifically includes:

[0092] Mark the historical data of X sample companies as data points A = (a1, a2, ..., a n ), B=(b1,b2,...,b n)..., where ai and bi are the values ​​of data points A and B on the i-th feature respectively. Each category of enterprise historical data can be used as a feature, such as enterprise historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and temperature changes.

[0093] Calculate the distance between the data points. For data point A = (a1, a2, ..., a n ) and data points B=(b1,b2,...,b n ), the distance between them is:

[0094]

[0095] The K data points closest to each other are grouped into the same category, and the category to which the data point belongs is determined based on the majority voting principle, where K = 0.2 × X. By setting an appropriate clustering coefficient, the X companies in the sample data can be clustered into roughly five categories, so that the companies are not subdivided too finely, and the carbon emission characteristics of different companies cannot be effectively distinguished and classified.

[0096] The estimation unit 23 is used to estimate the carbon emission range of the enterprise to be classified based on the KNN algorithm, which specifically includes: classifying the enterprise according to some historical data of the enterprise to be classified by applying the KNN algorithm, and estimating the carbon emission range of the enterprise to be classified or predicting the carbon emission of the enterprise to be classified according to the category of the enterprise to be classified.

[0097] Estimation based on historical data and KNN algorithm provides a data-driven decision-making method. This can avoid subjective guesswork and blind decision-making, and improve the accuracy and reliability of decision-making. By estimating the carbon emission range of the enterprise to be classified or predicting the specific carbon emissions, it provides a reference for the enterprise on its carbon emissions. This helps enterprises understand their own carbon emission levels, compare with other enterprises in the same industry, and formulate corresponding emission reduction targets and strategies.

[0098] Figure 3 Schematic diagram of a terminal device provided by an embodiment of the present invention. Figure 3 As shown, the third aspect of an embodiment of the present invention provides a terminal device, comprising: a data input module, a calculation and analysis module, a memory, a processor, and an enterprise carbon emission prediction program stored in the memory and executable on the processor, wherein the enterprise carbon emission prediction program is configured to implement the steps in the prediction method combining deep learning and the ARIMA model as described in any one of the first aspects.

[0099] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the prediction method combining deep learning and ARIMA model described in the first aspect.

[0100] An embodiment of the present invention provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0101] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera device / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0102] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0103] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0104] In the embodiments provided by the present invention, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0105] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed over multiple network units.

[0106] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0107] It should also be understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0108] As used in the present specification and the appended claims, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to monitoring, depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is monitored" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is monitored" or "in response to monitoring [described condition or event]", depending on the context.

[0109] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0110] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0111] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A forecasting method combining deep learning and ARIMA model, characterized in that: The prediction method combining deep learning and ARIMA model includes: Collect historical data of X sample companies, including historical carbon emissions data, shipment data, consumption and proportion of different energy sources, price data of different energy sources, and monthly data such as temperature changes; Using a KNN algorithm to classify enterprises, wherein the classification features of the KNN algorithm at least include energy prices and time factors; Estimate the range of carbon emissions of the enterprises to be classified, classify the enterprises according to some historical data of the enterprises to be classified, and estimate the range of carbon emissions of the enterprises to be classified or predict the carbon emissions of the enterprises to be classified according to the categories of the enterprises to be classified.

2. The prediction method combining deep learning and ARIMA model according to claim 1, characterized in that: The use of KNN algorithm to classify enterprises specifically includes: Mark the historical data of X sample companies as data points A = (a1, a2, ..., a n ), B=(b1,b2,...,b n )...; Calculate the distance between the data points. For data point A = (a1, a2, ..., a n ) and data points B=(b1,b2,...,b n ), the distance between them is: The K data points with the closest distance are classified into the same category, and the category to which the data point belongs is determined according to the majority voting principle, where K = 0.2 × X.

3. The prediction method combining deep learning and ARIMA model as claimed in claim 2, characterized in that: It also includes analyzing the sensitivity of different types of enterprises in the X sample enterprises to temperature and time. The specific calculation is: Among them, the temperature data is the weekly or monthly average temperature, x i is the data corresponding to temperature or time in the historical data of each sample enterprise, is the average value of the data related to temperature or time in the historical data of all sample enterprises, y i is the carbon emission data in the historical data of each sample enterprise, The average value of carbon emissions in the historical data of each sample enterprise.

4. The forecasting method combining deep learning and ARIMA model as claimed in claim 3, characterized in that: It also includes determining the sensitivity of different categories of enterprises to temperature and time based on the calculation results of the correlation coefficient. When the correlation coefficient r>0.7 or r<-0.7, enterprises in this category are considered to be temperature and time sensitive enterprises.

5. The forecasting method combining deep learning and ARIMA model as claimed in claim 4, characterized in that: Enterprises that are sensitive to temperature and time are taken as enterprises to be classified, and an ARIMA model is established to predict the carbon emission values ​​of the enterprises to be classified on a quarterly basis, including: Among them, μ is a constant term, is the predicted carbon emission at time t, y t-i is the observed value of carbon emissions at a past time point, and θ j is the model parameter, ∈ t is the error term. Try different combinations of ARIMA model parameters p (autoregressive order) and q (moving average order), where p is an integer from 1 to 6 and q is an integer from 1 to 4. Calculate the likelihood function value L and Akaike information criterion AIC = -2ln(L) + 6 of the ARIMA model under different value combinations. The combination of p and q with the minimum AIC is used as the value of the parameters p and q of the ARIMA model.

6. The forecasting method combining deep learning and ARIMA model as claimed in claim 5, characterized in that: It also includes energy consumption analysis for classified enterprises, including: Use the KNN algorithm to retrieve similar enterprises to the enterprise to be classified; Based on the ARIMA, the predicted carbon emission values ​​of the enterprises to be classified are obtained, and the energy consumption data of the 20% enterprises with the lowest carbon emission values ​​among the same type of enterprises are analyzed to obtain a low-carbon emission energy combination, wherein the energy consumption data analysis includes energy consumption and proportion data analysis; Compare the energy consumption data of the enterprise to be classified with the low-carbon emission energy consumption combination, and output energy consumption improvement suggestions on a quarterly basis.

7. The forecasting method combining deep learning and ARIMA model as claimed in claim 6, characterized in that: The specific suggestions for improving energy consumption by quarterly output include: The energy consumption data analysis includes analyzing the energy consumption and proportion data of the similar enterprises in the historical quarters to obtain the average values ​​of different types of energy consumption of the similar enterprises in the corresponding quarters; Calculate the difference between the different types of energy consumption of the enterprise to be classified in the forecast quarter and the different types of energy consumption of the same type of enterprise in the corresponding quarter; Energy improvement suggestions are output based on the difference, and the most similar enterprise after each energy consumption adjustment is retrieved based on the KNN algorithm, and the carbon emission reduction of the most similar enterprise in the corresponding quarter is used as the estimated value of the carbon emission reduction of the enterprise to be classified.

8. A forecasting system combining deep learning and ARIMA model, characterized in that: The enterprise carbon emission prediction device comprises: The collection unit is used to collect historical data of X sample companies, including historical carbon emission data, shipment data, consumption and proportion data of different energy sources, price data of different energy sources, and monthly data such as temperature changes; A classification unit, used for classifying enterprises using a KNN algorithm, wherein the classification features of the KNN algorithm include at least energy prices and time factors; The estimation unit is used to classify the enterprises by applying the KNN algorithm according to part of the historical data of the enterprises to be classified, and to estimate the carbon emission range of the enterprises to be classified or predict the carbon emission of the enterprises to be classified according to the categories of the enterprises to be classified.

9. A terminal device, characterized in that: The terminal device includes: a data input module, a calculation and analysis module, a memory, a processor, and an enterprise carbon emission prediction program stored in the memory and executable on the processor, wherein the enterprise carbon emission prediction program is configured to implement the steps in the prediction method combining deep learning and ARIMA model as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the forecasting method combining deep learning and ARIMA model as described in any one of claims 1 to 7 are implemented.

Images