A carbon emission reduction prediction system and method based on big data

Abstract

The present invention relates to the field of data processing, and in particular to a carbon emission reduction prediction system and method based on big data. It includes: collecting and integrating multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set; preprocessing, feature extraction and optimization processing of the integrated comprehensive data set to obtain an optimized feature data set; constructing and training a carbon emission prediction model, and obtaining global difference feedback through a multi-dimensional difference feedback optimization algorithm; optimizing the trained carbon emission prediction model based on global difference feedback; verifying the optimized carbon emission prediction model to obtain an optimized and verified carbon emission prediction model; applying the optimized and verified carbon emission prediction model to the optimized feature data set to obtain the final carbon emission prediction result, and generate carbon emission reduction strategy recommendations. It solves the technical problems of poor accuracy in processing and analyzing multi-source heterogeneous data and poor adaptability in carbon emission prediction analysis.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a carbon emission reduction prediction system and method based on big data. Background Art

[0002] The main sources of carbon emissions include energy consumption, industrial production, transportation and other fields. The carbon emissions in these fields directly affect the global greenhouse gas concentration. In order to achieve the carbon reduction target, scientific and accurate carbon emission forecasts and effective emission reduction strategy formulation are needed.

[0003] However, the existing carbon emission forecasting and analysis has the following technical problems: poor accuracy in processing and analyzing multi-source heterogeneous data and poor adaptability in carbon emission forecasting and analysis. Summary of the invention

[0004] The present invention provides a carbon emission reduction prediction system and method based on big data to solve the technical problems of poor accuracy in processing and analyzing multi-source heterogeneous data and poor adaptability in carbon emission prediction and analysis.

[0005] The present invention provides a carbon emission reduction prediction system and method based on big data, which specifically includes the following technical solutions:

[0006] A carbon emission reduction prediction method based on big data includes the following steps:

[0007] S1. Collect and integrate multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set; preprocess the integrated comprehensive data set, extract features from the preprocessed data set, and optimize the extracted original feature data to obtain an optimized feature data set;

[0008] S2. Construct and train a carbon emission prediction model. Based on the trained carbon emission prediction model, introduce a multidimensional difference feedback optimization algorithm. In the implementation process of the multidimensional difference feedback optimization algorithm, use the trained carbon emission prediction model to process the optimized feature data set to obtain a preliminary carbon emission prediction result; then introduce the impact value of external environmental changes , analyze the impact of external environmental changes on carbon emission forecast results, and the calculation formula is:

[0009] ,

[0010] in, Indicates External environmental factors at a point in time state variables; is the total number of external environmental factors; function Describes the The nonlinear impact of various external environmental factors on carbon emissions; is the weight coefficient;

[0011] Analyze the difference between the preliminary carbon emission forecast results and the actual carbon emission values, as well as the impact of changes in the external environment, form a comprehensive difference analysis, generate a comprehensive difference value, and introduce a comprehensive difference analysis formula, as follows:

[0012] ,

[0013] in, It's time point The comprehensive difference value of It's time point Preliminary carbon emission forecast results; It's time point Actual carbon emissions value; It is External environmental factors at a point in time The impact value of

[0014] Based on the comprehensive difference value, a global difference feedback is generated. The generation formula of the global difference feedback is:

[0015] ,

[0016] in, Represents the global difference feedback value; is the total number of time points analyzed; , , is the weight coefficient of each part;

[0017] Based on the global difference feedback on the gradient of the carbon emission prediction model parameters, the carbon emission prediction model parameters are adjusted to obtain an optimized carbon emission prediction model; the optimized carbon emission prediction model is verified to obtain an optimized and verified carbon emission prediction model;

[0018] S3. Apply the optimized and verified carbon emission prediction model to the optimized feature data set to obtain the final carbon emission prediction result; based on the final carbon emission prediction result, generate carbon emission reduction strategy recommendations.

[0019] Preferably, the S1 specifically includes:

[0020] The preprocessed data set is subjected to feature extraction to obtain original feature data; a multidimensional structured feature mapping and a nonlinear interactive optimization algorithm are introduced to optimize the original feature data, wherein the multidimensional structured feature mapping and the nonlinear interactive optimization algorithm perform multidimensional structured feature mapping on the original feature data to obtain mapping feature data; nonlinear interactive optimization is performed on the mapping feature data, and a nonlinear interactive feature data set is constructed based on the feature data after the nonlinear interactive optimization; global feature selection is performed on the nonlinear interactive feature data set to obtain an optimized feature data set.

[0021] Preferably, the S1 specifically includes:

[0022] The multi-dimensional structured feature mapping and nonlinear interactive optimization algorithm obtains mapping feature data by performing multi-dimensional structured feature mapping on the original feature data; the multi-dimensional structured feature mapping formula is as follows:

[0023] ,

[0024] in, Representation Dimension The mapping eigenvalues ​​of ; It is in the dimension Multi-dimensional structured feature mapping function; is the original feature data Total number of is the original feature data The Characteristic data; It is Feature data In Dimension The weight coefficient of It is the control Feature data In Dimension The standard deviation of the fluctuations; is the original feature data The Characteristic data; It is Feature data In Dimension The scaling factor of It is Feature data In Dimension Nonlinear expansion parameters of ; Is the dimension The global scaling factor of ; is the original feature data The Characteristic data; It is Feature data In Dimension The nonlinear amplitude control parameters of the image are as follows; and mapping feature data are generated based on the mapping feature values ​​of each dimension.

[0025] Preferably, the S1 specifically includes:

[0026] The mapping feature data is subjected to nonlinear interactive optimization, and a nonlinear interactive feature data set is formed based on the feature data after nonlinear interactive optimization; the nonlinear interactive optimization formula is defined as follows:

[0027] ,

[0028] in, Is the mapping feature data and map feature data The feature data after nonlinear interactive optimization, , , the feature data after all nonlinear interaction optimization constitutes the nonlinear interaction feature data set ; is the dimension of the mapping feature data; Is the mapping feature data and map feature data The nonlinear interaction weight coefficient between ; is the hyperbolic tangent function; is the interaction offset; Is the control mapping feature data and map feature data Parameters for the range of nonlinear interactions between ; is the difference weight coefficient; is the hyperbolic cosine function; is the differential impact control parameter;

[0029] Perform global feature selection on the nonlinear interactive feature data set to obtain an optimized feature data set.

[0030] Preferably, the S3 specifically includes:

[0031] Analyze the final carbon emission forecast results and identify the characteristics of carbon emission trends; generate carbon emission reduction strategy recommendations based on the final carbon emission forecast results and the characteristics of carbon emission trends.

[0032] A carbon emission reduction prediction system based on big data, including the following parts:

[0033] Data collection and integration module, data preprocessing module, feature engineering and selection module, carbon emission prediction model module, difference analysis feedback module, carbon emission reduction strategy recommendation module;

[0034] The data collection and integration module is responsible for collecting and integrating multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set, and then passing the integrated comprehensive data set to the data preprocessing module;

[0035] The data preprocessing module is responsible for preprocessing the integrated comprehensive data set to obtain the preprocessed data set, and then passing the preprocessed data set to the feature engineering and selection module;

[0036] The feature engineering and selection module is responsible for extracting features from the preprocessed data set and optimizing them to obtain an optimized feature data set, and then passing the optimized feature data set to the carbon emission prediction model module;

[0037] The carbon emission prediction model module builds and trains the carbon emission prediction model, obtains preliminary carbon emission prediction results based on the trained carbon emission prediction model, and sends the preliminary carbon emission prediction results to the difference analysis feedback module; optimizes the carbon emission prediction model according to the global difference feedback to obtain the optimized carbon emission prediction model, verifies the optimized carbon emission prediction model, and inputs the optimized feature data set into the optimized and verified carbon emission prediction model to obtain the final carbon emission prediction results, and transmits the final carbon emission prediction results to the carbon emission reduction strategy recommendation module;

[0038] The difference analysis feedback module conducts a comprehensive difference analysis on the preliminary carbon emission forecast results, the actual carbon emission values ​​and the differences in external environmental changes, obtains global difference feedback, and sends the global difference feedback to the carbon emission forecast model module;

[0039] The carbon emission reduction strategy recommendation module generates carbon emission reduction strategy recommendations based on the final carbon emission forecast results.

[0040] The beneficial effects of the technical solution of the present invention are:

[0041] 1. It uses multi-dimensional structured feature mapping and nonlinear interactive optimization algorithms. Through complex nonlinear mapping and feature interactive optimization, it can capture the complex relationship between feature data and generate a high-dimensional, low-redundancy optimized feature data set. Further, through global feature selection, the most informative and predictive features are retained, thereby significantly improving the accuracy of the carbon emission prediction model.

[0042] 2. The introduction of multi-dimensional difference feedback optimization algorithm realizes the dynamic adjustment of carbon emission prediction model, and realizes the continuous optimization of carbon emission prediction model through comprehensive difference analysis and global difference feedback. This closed-loop global difference feedback optimization process ensures the robustness and adaptability of carbon emission prediction model under various environmental change conditions, thus providing a solid data foundation and theoretical support for the scientific formulation of carbon emission reduction strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a structural diagram of a carbon emission reduction prediction system based on big data according to the present invention;

[0044] Figure 2 This is a flow chart of a carbon emission reduction prediction method based on big data described in the present invention. DETAILED DESCRIPTION

[0045] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The following is a detailed description of a carbon emission reduction prediction system and method based on big data provided by the present invention in conjunction with the accompanying drawings.

[0048] See attached Figure 1 , which shows a structural diagram of a carbon emission reduction prediction system based on big data provided by an embodiment of the present invention, the system includes the following parts:

[0049] Data collection and integration module, data preprocessing module, feature engineering and selection module, carbon emission prediction model module, difference analysis feedback module, carbon emission reduction strategy recommendation module;

[0050] The data collection and integration module is responsible for automatically collecting multi-source heterogeneous data related to carbon emissions from various data sources, including energy consumption data, industrial production data, transportation data, and meteorological data, and integrating and processing the multi-source heterogeneous data to obtain an integrated comprehensive data set, and then passing the integrated comprehensive data set to the data preprocessing module;

[0051] The data preprocessing module is responsible for preprocessing the integrated comprehensive data set, such as cleaning, standardization, denoising, missing value processing, and outlier detection, to obtain the preprocessed data set, and then pass the preprocessed data set to the feature engineering and selection module;

[0052] The feature engineering and selection module is responsible for extracting relevant features from the preprocessed data set, optimizing it, obtaining an optimized feature data set, and passing the optimized feature data set to the carbon emission prediction model module;

[0053] The carbon emission prediction model module constructs and trains the carbon emission prediction model based on the collected historical data related to carbon emissions to obtain a trained carbon emission prediction model, obtains preliminary carbon emission prediction results based on the trained carbon emission prediction model, and sends the preliminary carbon emission prediction results to the difference analysis feedback module, and then optimizes the carbon emission prediction model according to the global difference feedback of the difference analysis feedback module to obtain an optimized carbon emission prediction model; verifies the optimized carbon emission prediction model to obtain an optimized and verified carbon emission prediction model, and applies it to the optimized feature data set to predict future carbon emission trends to obtain the final carbon emission prediction results, and passes the final carbon emission prediction results to the carbon emission reduction strategy recommendation module;

[0054] The difference analysis feedback module conducts periodic comprehensive difference analysis on the preliminary carbon emission prediction results of the trained carbon emission prediction model, the actual carbon emission values, and the changes in the external environment (such as economic fluctuations and technological progress), obtains global difference feedback, and sends the global difference feedback to the carbon emission prediction model module;

[0055] The carbon emission reduction strategy recommendation module generates specific carbon emission reduction strategy recommendations based on the final carbon emission prediction results obtained by the carbon emission prediction model module.

[0056] See attached Figure 2 , which shows a flow chart of a carbon emission reduction prediction method based on big data provided by an embodiment of the present invention, the method comprising the following steps:

[0057] S1. Collect and integrate multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set; preprocess the integrated comprehensive data set, extract features from the preprocessed data set, and optimize the extracted original feature data to obtain an optimized feature data set;

[0058] Automatically collect multi-source heterogeneous data related to carbon emissions from multiple data sources, where the multi-source heterogeneous data may include energy consumption data, industrial production data, transportation data, and meteorological data; integrate and process the multi-source heterogeneous data, using data integration technology, including ETL (extraction, transformation, loading) process, data fusion algorithm (such as weighted fusion), to obtain an integrated comprehensive data set, where the data fusion algorithm is a technical means well known to those skilled in the art, and is selected according to specific circumstances, and will not be elaborated here;

[0059] The integrated comprehensive data set is preprocessed by cleaning, standardization, denoising, missing value processing and outlier detection to obtain a preprocessed data set. The preprocessing means are well known to those skilled in the art and will not be described in detail here.

[0060] The feature engineering technology is used to extract features from the preprocessed data set to obtain the original feature data; the multidimensional structured feature mapping and nonlinear interactive optimization algorithm are introduced to optimize the original feature data to obtain the optimized feature data set; the multidimensional structured feature mapping and nonlinear interactive optimization algorithm gradually generates the optimized feature data set through multidimensional structured feature mapping, nonlinear interactive optimization and global feature selection; the specific implementation process is as follows:

[0061] First, the original feature data Multidimensional structured feature mapping is performed to capture complex nonlinear relationships; the formula for multidimensional structured feature mapping is as follows:

[0062] ,

[0063] in, Representation Dimension The mapping feature value is obtained by The new eigenvalues ​​obtained after complex nonlinear mapping are used for the next step of global feature selection; It is in the dimension The multi-dimensional structured feature mapping function combines multiple nonlinear transformations to improve the information content and representation ability of feature data; is the original feature data The number of bits, that is, the original feature data Total number of is the original feature data The Characteristic data; It is Feature data In Dimension The weight coefficient of Feature data Mapping to dimensions The contribution of is set by statistics or prior knowledge; It is the control Feature data In Dimension The standard deviation of the fluctuation determines the size of the fluctuation range of the feature data in multi-dimensional structured feature mapping; is the original feature data The Characteristic data; It is Feature data In Dimension The scaling factor affects the smoothness of the feature data in the logarithmic transformation; It is Feature data In Dimension The nonlinear expansion parameter controls the Feature data The degree of expansion in the exponential function mapping; Is the dimension The global scaling factor controls the scaling ratio of the overall feature mapping; is the original feature data The Characteristic data; It is Feature data In Dimension The nonlinear amplitude control parameter affects the amplitude of the characteristic data in cubic and fractional form; Use the sine function to express the fluctuation characteristics of feature data in different dimensions; Combining logarithmic function and exponential function, it expresses the complex nonlinear interaction between feature data; By introducing cubic terms and fractional forms, the complexity of feature data after multi-dimensional structured feature mapping is enhanced;

[0064] After using multi-dimensional structured feature mapping, generate mapping feature data , is the dimension of the mapping feature data; in order to analyze the complex nonlinear interaction between the mapping feature data and extract the feature combination with significant influence, the mapping feature data Perform nonlinear interaction optimization. The nonlinear interaction optimization formula is used to realize the nonlinear interaction optimization calculation between the mapping feature data; the nonlinear interaction optimization formula is defined as follows:

[0065] ,

[0066] in, Is the mapping feature data and map feature data The feature data after nonlinear interactive optimization, , At the same time, the feature data after all nonlinear interaction optimization constitutes the nonlinear interaction feature data set ; Is the mapping feature data and map feature data The nonlinear interaction weight coefficient between is used to indicate the strength of the nonlinear interaction between the two mapping feature data; It is a hyperbolic tangent function, which is used to compress the nonlinear interaction results between the mapped feature data into a specific range (between -1 and 1), thereby reflecting the smoothness of the nonlinear interaction relationship; is the interaction offset, is the mapping feature data and map feature data An offset parameter introduced during nonlinear interaction can adjust the center position of the nonlinear interaction between two mapped feature data; Is the control mapping feature data and map feature data The parameters of the nonlinear interaction range between them determine the decay rate of the nonlinear interaction influence; is the difference weight coefficient, used to represent the mapping feature data and map feature data The contribution of the differences between them to the overall nonlinear interaction; It is a hyperbolic cosine function, which is used to describe the influence of the difference between mapping features on the nonlinear interaction results. It mainly amplifies or reduces the effect of the difference and controls its influence range. It is the difference that affects the control parameters and adjusts the mapping feature data and map feature data The degree of influence of the differences between them on the nonlinear interaction results.

[0067] Furthermore, the most informative and helpful features for carbon emission prediction are selected from the nonlinear interactive feature data set through global feature selection to obtain an optimized feature data set; global feature selection is achieved by minimizing feature redundancy and maximizing feature contribution. The objective function of global feature selection is The definition is as follows:

[0068] ,

[0069] in, is the dimension of the nonlinear interaction feature dataset; It is the coefficient of redundancy between features, which is used to measure the nonlinear interaction feature data and nonlinear interaction feature data If the similarity between two nonlinear interaction feature data is high, then The value of will be large, which means that there is a large redundancy between the nonlinear interaction feature data; is the nonlinear interaction feature data The contribution coefficient of For indicators such as expected carbon emissions The size of the contribution; is the regularization coefficient, which controls the complexity of nonlinear interactive feature data in the global feature selection process to prevent overfitting; is the nonlinear interaction feature data The inverse value of the scaling factor is used to adjust the contribution of nonlinear interaction feature data to the objective function;

[0070] The above objective function is to optimize the global feature selection of nonlinear interactive feature data to ensure that feature redundancy is reduced while retaining the feature data with the largest amount of information; global feature selection minimizes the objective function of global feature selection through existing optimization algorithms (such as global optimization methods based on simulated annealing algorithms) To select the optimized feature data set ; In the process of global feature selection, the redundancy and contribution of feature data are comprehensively considered through complex nonlinear interactive optimization formulas to ensure that the final selected feature data has both high information content and avoids excessive redundancy.

[0071] S2. Construct and train a carbon emission prediction model. Based on the trained carbon emission prediction model, obtain global difference feedback through a multi-dimensional difference feedback optimization algorithm; optimize the trained carbon emission prediction model based on the global difference feedback to obtain an optimized carbon emission prediction model; verify the optimized carbon emission prediction model to obtain an optimized and verified carbon emission prediction model;

[0072] Collect historical data related to carbon emissions, including energy consumption, industrial production index, transportation volume and meteorological data; integrate, preprocess, extract features and optimize the historical data through the same integration, preprocessing, feature extraction and optimization methods as step S1 to obtain an optimized historical feature data set; professional and technical personnel select an existing machine learning model as a carbon emission prediction model according to specific needs, such as a gradient boosting decision tree model, to perform carbon emission prediction; and train and verify the selected carbon emission prediction model based on the optimized historical feature data set to obtain a trained carbon emission prediction model.

[0073] Furthermore, in order to effectively identify and optimize the deficiencies of the carbon emission prediction model in a dynamically changing environment, thereby improving the accuracy of carbon emission prediction, a multidimensional difference feedback optimization algorithm is introduced. The multidimensional difference feedback optimization algorithm generates global difference feedback by comprehensively considering the difference between the preliminary carbon emission prediction results and the changes in the external environment, and optimizes the carbon emission prediction model based on the global difference feedback. The specific implementation process is as follows:

[0074] The optimized feature data set is input into the trained carbon emission model to obtain the preliminary carbon emission prediction results; then the impact of external environmental changes on the carbon emission prediction results is analyzed. External environmental changes can include technological progress, economic fluctuations, etc.; the impact value of external environmental changes is introduced , and its calculation formula is:

[0075] ,

[0076] in, Indicates External environmental factors at a point in time state variables, such as the magnitude of economic fluctuations; is the total number of external environmental factors; function Describes the The nonlinear or complex impacts of various external environmental factors on carbon emissions are obtained through modeling of the impacts of various external factors and fitting of historical data; is the weight coefficient, reflecting the The contribution of various external factors to overall environmental change;

[0077] Next, we analyze the difference between the preliminary carbon emission forecast results and the actual carbon emission values, and take into account the impact of external environmental changes to form a comprehensive difference analysis. The comprehensive difference analysis formula is introduced as follows:

[0078] ,

[0079] in, It's time point The comprehensive difference value represents the comprehensive difference between the predicted value and the actual value and the external environment impact; It's time point Preliminary carbon emission forecast results; It's time point The actual carbon emission value is used to compare with the preliminary carbon emission forecast results; It is External environmental factors at a point in time The impact value of

[0080] At each time point The comprehensive difference value Afterwards, multiple time points The difference information on the above is combined to generate global difference feedback, so as to facilitate the overall optimization of the carbon emission prediction model. The generation formula of global difference feedback is:

[0081] ,

[0082] in, It represents the global difference feedback value, which indicates the overall evaluation of the forecast deviation within the cycle and is used as a reference for optimizing the carbon emission forecast model; is the total number of time points analyzed and used to calculate the mean of global differential feedback; , , It is the weight coefficient of each part, which is used to control the impact of comprehensive difference value, external environment change and time periodicity on global difference feedback; It directly reflects the impact of differences at a single time point on global feedback; The cumulative effect of changes in the external environment is taken into account, including Smooths out large environmental changes; finally, The influence of time periodicity is introduced. When carbon emissions have significant seasonal or periodic fluctuations, the time function can capture this pattern and ensure the accuracy of global difference feedback.

[0083] In order to feedback the global difference To optimize the carbon emission prediction model, it is necessary to calculate the gradient of the global difference feedback on the carbon emission prediction model parameters to guide the further adjustment of the carbon emission prediction model. The formula is:

[0084] ,

[0085] in, are the optimized carbon emission prediction model parameters; are the trained carbon emission prediction model parameters; is the learning rate, which is used to control the amplitude of the carbon emission prediction model parameter update and ensure the stability of the carbon emission prediction model adjustment; It is the gradient of the global difference feedback on the carbon emission prediction model parameters, indicating the direction and size of the adjustment of each control carbon emission prediction model parameter.

[0086] After the above processing, an optimized carbon emission prediction model is obtained, and then verified by a method such as cross-validation to obtain an optimized and verified carbon emission prediction model.

[0087] S3. Apply the optimized and verified carbon emission prediction model to the optimized feature data set to obtain the final carbon emission prediction result; based on the final carbon emission prediction result, generate carbon emission reduction strategy recommendations.

[0088] The optimized feature data set is used as input to the optimized and verified carbon emission prediction model. According to the optimized and verified carbon emission prediction model, the future carbon emission trend is predicted to obtain the final carbon emission prediction result. The final carbon emission prediction result includes the carbon emission trend in the future time period, and the final carbon emission prediction result is analyzed using the existing time series analysis algorithm to identify the main characteristics of the carbon emission trend, including trend changes, seasonal fluctuations and fluctuation analysis. Trend changes include trend analysis to identify long-term trends, such as whether carbon emissions continue to grow or decline; seasonal analysis to analyze whether there are periodic fluctuations in the final carbon emission prediction results, such as seasonal fluctuations or annual cycles; fluctuation analysis to identify the causes of short-term fluctuations, such as sudden changes in economic activities;

[0089] According to the final carbon emission forecast results and the main characteristics of the carbon emission trends obtained through analysis, specific carbon emission reduction strategy recommendations are generated. For example, carbon emission reduction strategies for different fields: in the energy field, it is recommended to gradually phase out high-carbon emission energy, increase the proportion of renewable energy, and promote high-efficiency energy-saving technologies; in the industrial field, it is recommended to introduce clean production technologies, promote carbon capture and storage (CCS) technologies, and strengthen energy efficiency management; in the transportation field, it is recommended to optimize the transportation structure, promote new energy vehicles, improve the public transportation system, and encourage low-carbon travel methods. For example, short-term and long-term carbon emission reduction strategies: short-term strategies propose emergency measures for short-term fluctuations in the final carbon emission forecast results, such as taking temporary emission reduction measures during the peak period of carbon emissions; long-term strategies formulate medium- and long-term structural emission reduction measures for long-term trends, such as adjusting the energy structure, optimizing industrial layout, and strengthening technological innovation and research and development.

[0090] In summary, a carbon emission reduction prediction system and method based on big data has been completed.

[0091] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0093] The above embodiments 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 above embodiments, those skilled in the art should understand that the technical solutions described in the above 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 be included in the protection scope of the present invention.

Claims

1. A carbon emission reduction prediction method based on big data, characterized in that: The following steps are involved: S1. Collect and integrate multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set; preprocess the integrated comprehensive data set, extract features from the preprocessed data set, and optimize the extracted original feature data to obtain an optimized feature data set; S2. Construct and train a carbon emission prediction model. Based on the trained carbon emission prediction model, introduce a multidimensional difference feedback optimization algorithm. In the implementation process of the multidimensional difference feedback optimization algorithm, use the trained carbon emission prediction model to process the optimized feature data set to obtain a preliminary carbon emission prediction result; then introduce the impact value of external environmental changes , analyze the impact of external environmental changes on carbon emission forecast results, and the calculation formula is: ， in, Indicates External environmental factors at a point in time state variables; is the total number of external environmental factors; function Describes the The nonlinear impact of various external environmental factors on carbon emissions; is the weight coefficient; Analyze the difference between the preliminary carbon emission forecast results and the actual carbon emission values, as well as the impact of changes in the external environment, form a comprehensive difference analysis, generate a comprehensive difference value, and introduce a comprehensive difference analysis formula, as follows: ， in, It's time point The comprehensive difference value of It's time point Preliminary carbon emission forecast results; It's time point Actual carbon emissions value; It is External environmental factors at a point in time The impact value of Based on the comprehensive difference value, a global difference feedback is generated. The generation formula of the global difference feedback is: ， in, Represents the global difference feedback value; is the total number of time points analyzed; , , is the weight coefficient of each part; Based on the global difference feedback on the gradient of the carbon emission prediction model parameters, the carbon emission prediction model parameters are adjusted to obtain an optimized carbon emission prediction model; the optimized carbon emission prediction model is verified to obtain an optimized and verified carbon emission prediction model; S3. Apply the optimized and verified carbon emission prediction model to the optimized feature data set to obtain the final carbon emission prediction result; based on the final carbon emission prediction result, generate carbon emission reduction strategy recommendations.

2. The carbon emission reduction prediction method based on big data according to claim 1 is characterized in that: The S1 specifically includes: The preprocessed data set is subjected to feature extraction to obtain original feature data; a multidimensional structured feature mapping and a nonlinear interactive optimization algorithm are introduced to optimize the original feature data, wherein the multidimensional structured feature mapping and the nonlinear interactive optimization algorithm perform multidimensional structured feature mapping on the original feature data to obtain mapping feature data; nonlinear interactive optimization is performed on the mapping feature data, and a nonlinear interactive feature data set is constructed based on the feature data after the nonlinear interactive optimization; global feature selection is performed on the nonlinear interactive feature data set to obtain an optimized feature data set.

3. The carbon emission reduction prediction method based on big data according to claim 2 is characterized in that: The S1 specifically includes: The multi-dimensional structured feature mapping and nonlinear interactive optimization algorithm obtains mapping feature data by performing multi-dimensional structured feature mapping on the original feature data; the multi-dimensional structured feature mapping formula is as follows: ， in, Representation Dimension The mapping eigenvalues ​​of ; It is in the dimension Multi-dimensional structured feature mapping function; is the original feature data Total number of is the original feature data The Characteristic data; It is Feature data In Dimension The weight coefficient of It is the control Feature data In Dimension The standard deviation of the fluctuations; is the original feature data The Characteristic data; It is Feature data In Dimension The scaling factor of It is Feature data In Dimension Nonlinear expansion parameters of ; Is the dimension The global scaling factor of ; is the original feature data The Characteristic data; It is Feature data In Dimension The nonlinear amplitude control parameters of the image are as follows; and mapping feature data are generated based on the mapping feature values ​​of each dimension.

4. The carbon emission reduction prediction method based on big data according to claim 3 is characterized in that: The S1 specifically includes: The mapping feature data is subjected to nonlinear interactive optimization, and a nonlinear interactive feature data set is formed based on the feature data after nonlinear interactive optimization; the nonlinear interactive optimization formula is defined as follows: ， in, Is the mapping feature data and map feature data The feature data after nonlinear interactive optimization, , , the feature data after all nonlinear interaction optimization constitutes the nonlinear interaction feature data set ; is the dimension of the mapping feature data; Is the mapping feature data and map feature data The nonlinear interaction weight coefficient between ; is the hyperbolic tangent function; is the interaction offset; Is the control mapping feature data and map feature data Parameters for the range of nonlinear interactions between ; is the difference weight coefficient; is the hyperbolic cosine function; is the differential impact control parameter; Perform global feature selection on the nonlinear interactive feature data set to obtain an optimized feature data set.

5. The carbon emission reduction prediction method based on big data according to claim 1 is characterized in that: The S3 specifically includes: Analyze the final carbon emission forecast results and identify the characteristics of carbon emission trends; generate carbon emission reduction strategy recommendations based on the final carbon emission forecast results and the characteristics of carbon emission trends.

6. A carbon emission reduction prediction system based on big data, applied to the carbon emission reduction prediction method based on big data as claimed in claim 1, characterized in that: Includes the following sections: Data collection and integration module, data preprocessing module, feature engineering and selection module, carbon emission prediction model module, difference analysis feedback module, carbon emission reduction strategy recommendation module; The data collection and integration module is responsible for collecting and integrating multi-source heterogeneous data of carbon emissions to obtain an integrated comprehensive data set, and then passing the integrated comprehensive data set to the data preprocessing module; The data preprocessing module is responsible for preprocessing the integrated comprehensive data set to obtain the preprocessed data set, and then passing the preprocessed data set to the feature engineering and selection module; The feature engineering and selection module is responsible for extracting features from the preprocessed data set and optimizing them to obtain an optimized feature data set, and then passing the optimized feature data set to the carbon emission prediction model module; The carbon emission prediction model module builds and trains the carbon emission prediction model, obtains preliminary carbon emission prediction results based on the trained carbon emission prediction model, and sends the preliminary carbon emission prediction results to the difference analysis feedback module; optimizes the carbon emission prediction model according to the global difference feedback to obtain the optimized carbon emission prediction model, verifies the optimized carbon emission prediction model, and inputs the optimized feature data set into the optimized and verified carbon emission prediction model to obtain the final carbon emission prediction results, and transmits the final carbon emission prediction results to the carbon emission reduction strategy recommendation module; The difference analysis feedback module conducts a comprehensive difference analysis on the preliminary carbon emission forecast results, the actual carbon emission values ​​and the differences in external environmental changes, obtains global difference feedback, and sends the global difference feedback to the carbon emission forecast model module; The carbon emission reduction strategy recommendation module generates carbon emission reduction strategy recommendations based on the final carbon emission forecast results.

Images