Urban green planning system based on integrated learning of urban carbon emissions

Abstract

The present invention discloses an urban green planning system based on integrated learning of urban carbon emissions, relates to the technical field of urban carbon emissions planning, and is intended to solve the problem of poor learning effect of urban carbon emissions in the integrated learning process. The present invention divides the standardized carbon emissions data into a reasonable data set to obtain a training set, a validation set, and a test set, which helps the model to conduct targeted learning and evaluation at different stages. By optimizing the learning rate, batch size, number of hidden layers, number of hidden layer nodes, regularization coefficient, and other hyperparameters of the validation set, the performance of the model can be significantly improved. By calculating the area of ​​the non-overlapping area, the gap between the urban planning carbon emissions and the target carbon emissions can be objectively reflected. The threshold assessment results are divided into low-risk emissions, medium-risk emissions, and high-risk emissions, and the results are labeled, which helps decision makers take corresponding measures according to different risk levels.

Description

Technical Field

[0001] The present invention relates to the technical field of urban carbon emission planning, and in particular to an urban green planning system based on urban carbon emission integrated learning. Background Art

[0002] Urban carbon emission planning refers to a series of goals and measures formulated for greenhouse gas emissions within the city, especially carbon dioxide emissions.

[0003] The patent application with publication number CN115438908A discloses a carbon emission prediction method, system, storage medium and electronic device, which mainly realizes the prediction of carbon emission reduction by formulating corresponding carbon reduction plans and analyzing the changes in energy consumption and carbon emissions. It proposes a planning algorithm with different carbon reduction plans as input and carbon emission prediction as output, which provides support for relevant carbon emission planners to formulate carbon reduction strategies. Although the above patent solves the problem of carbon emission prediction, there are still the following problems in actual operation:

[0004] 1. After obtaining the carbon emission data in the city, it is not classified and processed according to the carbon emission type, resulting in poor quality of carbon emission data.

[0005] 2. When constructing a model using urban carbon emission data, no better integrated learning model is selected for the construction process, resulting in reduced efficiency in integrated learning of carbon emission data.

[0006] 3. The city’s carbon emission data is not used for further risk assessment and early warning, resulting in reduced subsequent processing capabilities. Summary of the invention

[0007] The purpose of the present invention is to provide an urban green planning system based on integrated learning of urban carbon emissions. By reasonably dividing the standardized carbon emission data into a data set, a training set, a validation set and a test set are obtained, which helps the model to perform targeted learning and evaluation at different stages. By optimizing the learning rate, batch size, number of hidden layers, number of hidden layer nodes, regularization coefficient and other hyperparameters of the validation set, the performance of the model can be significantly improved. By calculating the area of ​​the non-overlapping area, the gap between the urban planning carbon emissions and the target carbon emissions can be objectively reflected. The threshold assessment results are divided into low-risk emissions, medium-risk emissions and high-risk emissions, and the results are labeled, which helps decision makers take corresponding measures according to different risk levels and can solve the problems in the prior art.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] The urban green planning system based on integrated learning of urban carbon emissions includes:

[0010] Carbon emission data acquisition unit, used for:

[0011] Retrieving different types of carbon emission data from the database, uniquely coding each type of carbon emission data, and marking the carbon emission data after the unique coding is completed as carbon emission data to be processed;

[0012] Carbon emission data processing unit, used for:

[0013] The carbon emission data to be processed are preprocessed, and after the data preprocessing is completed, standardized carbon emission data are obtained;

[0014] Processing data model building blocks for:

[0015] The standardized carbon emission data is used to construct an integrated learning model, and after the integrated learning model is constructed, the target carbon emission data is obtained;

[0016] Model data evaluation unit for:

[0017] Compare the target carbon emission data with the urban planning carbon emission data, conduct a threshold assessment on the target carbon emission data based on the data comparison results, and obtain the carbon emission decision data after the threshold assessment is completed;

[0018] Carbon emission assessment data decision unit, used for:

[0019] Targeted decision-making plans are formulated according to the evaluation scope of the carbon emission decision data, and the formulated plans and carbon emission decision data are transmitted to the display terminal for display of the data and plans.

[0020] Preferably, the carbon emission data acquisition unit includes:

[0021] Carbon emission data classification module, used for:

[0022] Confirm different types of carbon emission data from the database;

[0023] The types of carbon emissions include transportation carbon emissions data, industrial carbon emissions data, building carbon emissions data, energy supply carbon emissions data, agricultural carbon emissions data, waste treatment carbon emissions data and indirect carbon emissions data;

[0024] Among them, transportation carbon emission data refers to the energy consumption data of private cars, public transportation, rail transportation, aviation, water transportation and various means of transportation; industrial carbon emission data refers to the greenhouse gas emission data in the industrial production process; building carbon emission data refers to the energy consumption data of heating, air conditioning, hot water supply and lighting systems in residential and public buildings; energy supply carbon emission data refers to the carbon emission data in the process of fossil energy mining, processing, transportation and use; agricultural carbon emission data refers to the greenhouse gas emission data generated by fertilizer use, animal husbandry, deforestation and land use change activities in agricultural activities; waste treatment carbon emission data refers to the greenhouse gas emission data generated by garbage incineration and landfill treatment methods; indirect carbon emission data refers to purchased electricity, heating and cooling, as well as other indirect emission data caused by internal urban activities but generated outside the jurisdiction.

[0025] Preferably, the carbon emission data acquisition unit further includes:

[0026] Carbon emission data labeling module, used for:

[0027] Confirm each type of carbon emission data and uniquely code each type of carbon emission data;

[0028] Timestamp the acquisition time of each type of carbon emission data;

[0029] Using a hash algorithm to convert one or more field values ​​of each type of carbon emission data into a unique hash code, and using the hash code as a serial number;

[0030] The serial number and timestamp are uniquely coded to generate the carbon emission data, and the generated uniquely coded carbon emission data is marked as the carbon emission data to be processed.

[0031] Preferably, the carbon emission data processing unit is further used for:

[0032] Preprocess the carbon emission data to be processed;

[0033] The data preprocessing process is to sequentially clean, integrate, convert, reduce dimension and verify the carbon emission data to be processed;

[0034] First, the carbon emission data to be processed is cleaned. Data cleaning is to correct or delete missing values, erroneous values ​​and duplicate values ​​in the data;

[0035] After data cleaning is completed, data integration is performed to merge data from different data sources in the carbon emission data to be processed into one data set;

[0036] After the data integration is completed, data conversion is performed to convert data of different dimensions in the carbon emission data to be processed into the same dimension;

[0037] After the data conversion is completed, data dimension reduction is performed. Data dimension reduction is to convert the carbon emission data to be processed into a new coordinate system through linear transformation, so that the first largest variance of any projection is on the first coordinate, the second largest variance is on the second coordinate, and so on;

[0038] After the data dimension reduction is completed, data verification is performed to detect the data range, logic and missing values ​​of the data after data cleaning, data integration, data conversion and data dimension reduction;

[0039] After data verification is completed, standardized carbon emission data is obtained.

[0040] Preferably, data cleaning corrects or deletes erroneous data, including:

[0041] Extract erroneous data that occurs during data cleaning;

[0042] Extracting two pieces of data information adjacent to the collection time of the erroneous data as the data information at the previous time and the data information at the next time;

[0043] Extract data cleaning history records with corrected error data;

[0044] The error degree coefficient corresponding to the error data is obtained by using the previous data information and the subsequent data information in combination with the data cleaning history record and the corrected error data; wherein the error degree coefficient is obtained by the following formula:

[0045]

[0046] Among them, R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X x and X h They represent the data values ​​corresponding to the previous data information and the subsequent data information respectively; n represents the number of corrected error data in the data cleaning history record; X i represents the corrected data value corresponding to the corrected erroneous data in the i-th data cleaning history record; X yi Indicates the data value corresponding to the corrected erroneous data in the i-th data cleaning history record;

[0047] Comparing the error degree coefficient corresponding to the error data with a preset degree coefficient threshold;

[0048] When the error degree coefficient corresponding to the erroneous data exceeds a preset degree coefficient threshold, the erroneous data is deleted;

[0049] When the error degree coefficient corresponding to the erroneous data does not exceed a preset degree coefficient threshold, the erroneous data is corrected.

[0050] Preferably, when the error degree coefficient corresponding to the erroneous data does not exceed a preset degree coefficient threshold, the erroneous data is corrected, including:

[0051] Extracting a plurality of correct data before the collection time of the erroneous data, wherein the value range of the correct data is 50-100;

[0052] Extracting the subsequent data information corresponding to the erroneous data;

[0053] The modification amount corresponding to the erroneous data is obtained by using the multiple correct data collected before the time and the data information corresponding to the erroneous data at the time after the time; wherein the modification amount corresponding to the erroneous data is obtained by the following formula:

[0054]

[0055] Among them, ΔX represents the modification amount corresponding to the erroneous data; R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X h represents the data value corresponding to the data information at the later time; m represents the number of correct data corresponding to the previous collection time; X j represents the data value corresponding to the correct data before the jth acquisition time; w represents the bit coefficient, when (X e -X h )>0, then let w=1; when (X e -X h )<0, then let w=-1;

[0056] The modified error data is obtained by combining the modification amount corresponding to the error data with the error data, wherein the modified error data is obtained by the following formula:

[0057] X s =X e +ΔX

[0058] Among them, X s Indicates the corrected error data; X e represents the data value corresponding to the erroneous data; ΔX represents the modification amount corresponding to the erroneous data.

[0059] Preferably, the processing data model building unit is further used for:

[0060] The standardized carbon emission data is divided into a data set, and the data set is divided into a training set, a validation set and a test set;

[0061] Import the training set into the neural network learning machine for data training;

[0062] Among them, the data training process is to first forward propagate the training set, where the training set is propagated from low level to high level;

[0063] When the data result obtained by propagation is not within the standard propagation threshold, back propagation is performed, where back propagation is to propagate the error from the high level to the low level for training;

[0064] Among them, the process of propagation training is:

[0065] First, the weights of the training set are initialized. After the setting is completed, the training set is forward propagated through the convolution layer, downsampling layer, and fully connected layer to obtain the output value;

[0066] When the error is greater than the expected value, the error is passed back to the network, and the errors of the fully connected layer, downsampling layer, and convolutional layer are obtained in turn;

[0067] Among them, the error of each layer is the total error of the network; when the error is equal to or less than the expected value, the training is completed.

[0068] Preferably, the processing data model building unit is further used for:

[0069] After the training of the training set is completed, the validation set is used to evaluate the model hyperparameter performance, where the Bayesian optimization method is used to evaluate the hyperparameter performance;

[0070] Model hyperparameters include the learning rate of the validation set, batch size, number of hidden layers, number of hidden layer nodes, and regularization coefficient;

[0071] During the model hyperparameter performance evaluation process, if the threshold parameter of the model hyperparameter performance of the validation set reaches the critical threshold point, the evaluation is stopped;

[0072] According to the model hyperparameter performance evaluation results of the validation set, select the ensemble model with the best evaluation performance;

[0073] The ensemble model with the best performance is evaluated on the test set for performance prediction, and whether to adjust the performance index is determined based on the performance prediction results;

[0074] Among them, the performance indicators include the accuracy, recall, F1 score, and AUC-ROC of the integrated model;

[0075] Finally, an integrated learning model of standardized carbon emission data is obtained, and the integrated learning model is labeled as the target carbon emission data.

[0076] Preferably, the model data evaluation unit is further used for:

[0077] Retrieve urban planning carbon emission data from the database;

[0078] Perform curve data conversion on each type of carbon emission data in the retrieved urban planning carbon emission data and target carbon emission data;

[0079] Overlap the urban planning carbon emission data with the carbon emission curve data of the same type in the target carbon emission data;

[0080] After the curves overlap, a threshold assessment is performed on the target carbon emission data based on the overlap results;

[0081] The threshold assessment is to calculate the area of ​​non-overlapping areas in the curve overlap results of the urban planning carbon emission data and the carbon emission curve data of the same type in the target carbon emission data, and perform the threshold assessment based on the area of ​​non-overlapping areas;

[0082] The threshold assessment results are divided into low-risk emissions, medium-risk emissions, and high-risk emissions;

[0083] The threshold assessment results are marked as carbon emission decision data.

[0084] Preferably, the carbon emission assessment data decision unit is further used to:

[0085] Provide different levels of early warning based on carbon emission decision data;

[0086] The warning levels are level one, level two and level three;

[0087] Obtain different types of carbon emissions in different levels of early warning;

[0088] Formulate plans in different directions according to different types;

[0089] The formulated plan and threshold evaluation results are transmitted to the display terminal for display.

[0090] Compared with the prior art, the present invention has the following beneficial effects:

[0091] 1. The urban green planning system based on urban carbon emission integrated learning provided by the present invention divides the standardized carbon emission data into a reasonable data set to obtain a training set, a validation set, and a test set, which helps the model to conduct targeted learning and evaluation at different stages. By optimizing the learning rate, batch size, number of hidden layers, number of hidden layer nodes, regularization coefficient and other hyperparameters of the validation set, the performance of the model can be significantly improved. At the same time, the threshold parameters of the hyperparameter performance of the validation set model are set, and the evaluation is stopped when the critical threshold point is reached, avoiding unnecessary waste of computing resources.

[0092] 2. The urban green planning system based on integrated learning of urban carbon emissions provided by the present invention can objectively reflect the gap between the urban planned carbon emissions and the target carbon emissions by calculating the area of ​​non-overlapping regions, and divides the threshold assessment results into low-risk emissions, medium-risk emissions and high-risk emissions, and labels the results, which helps decision makers take corresponding measures according to different risk levels.

[0093] 3. The urban green planning system based on urban carbon emission integrated learning provided by the present invention can realize accurate monitoring of urban carbon emissions by carrying out early warnings of different degrees according to carbon emission decision data. This hierarchical early warning mechanism can trigger corresponding response measures in a timely manner according to different levels of carbon emissions, effectively avoid environmental problems caused by excessive carbon emissions, obtain different types of carbon emissions in early warnings of different degrees, and formulate plans in different directions accordingly, reflecting the idea of ​​flexible response and differentiated management. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 A schematic diagram of the city carbon emission integrated learning unit of the present invention;

[0095] Figure 2 It is a schematic diagram of the integrated learning process of urban carbon emissions of the present invention. DETAILED DESCRIPTION

[0096] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0097] In order to solve the problem in the existing technology that after obtaining the carbon emission data in the city, the data is not classified and processed according to the carbon emission type, resulting in poor quality of carbon emission data, please refer to Figure 1 and Figure 2 , this embodiment provides the following technical solutions:

[0098] The urban green planning system based on integrated learning of urban carbon emissions includes:

[0099] Carbon emission data acquisition unit, used for:

[0100] Retrieving different types of carbon emission data from the database, uniquely coding each type of carbon emission data, and marking the carbon emission data after the unique coding is completed as carbon emission data to be processed;

[0101] Carbon emission data processing unit, used for:

[0102] The carbon emission data to be processed are preprocessed, and after the data preprocessing is completed, standardized carbon emission data are obtained;

[0103] Processing data model building blocks for:

[0104] The standardized carbon emission data is used to construct an integrated learning model, and after the integrated learning model is constructed, the target carbon emission data is obtained;

[0105] Model data evaluation unit for:

[0106] Compare the target carbon emission data with the urban planning carbon emission data, conduct a threshold assessment on the target carbon emission data based on the data comparison results, and obtain the carbon emission decision data after the threshold assessment is completed;

[0107] Carbon emission assessment data decision unit, used for:

[0108] Targeted decision-making plans are formulated according to the evaluation scope of the carbon emission decision data, and the formulated plans and carbon emission decision data are transmitted to the display terminal for display of the data and plans.

[0109] Specifically, the carbon emission data acquisition unit implements standardized data processing by uniquely coding and timestamp-confirming each type of carbon emission data and generating a serial number using a hash algorithm. The automated data processing flow of the carbon emission data processing unit can greatly improve the efficiency of data processing. The integrated learning model of standardized carbon emission data obtained by the processing data model construction unit not only has high accuracy and generalization ability, but can also adjust and optimize performance indicators according to actual needs. The model data evaluation unit evaluates the difference between urban planning carbon emission data and target carbon emission data, so as to timely discover potential problems in urban planning. The carbon emission evaluation data decision unit transmits the formulated plan and threshold evaluation results to the display terminal for display, which helps to achieve efficient collaboration and information sharing.

[0110] Carbon emission data acquisition unit, including:

[0111] Carbon emission data classification module, used for:

[0112] Confirm different types of carbon emission data from the database;

[0113] The types of carbon emissions include transportation carbon emissions data, industrial carbon emissions data, building carbon emissions data, energy supply carbon emissions data, agricultural carbon emissions data, waste treatment carbon emissions data and indirect carbon emissions data;

[0114] Among them, transportation carbon emission data refers to the energy consumption data of private cars, public transportation, rail transportation, aviation, water transportation and various means of transportation; industrial carbon emission data refers to the greenhouse gas emission data in the industrial production process; building carbon emission data refers to the energy consumption data of heating, air conditioning, hot water supply and lighting systems in residential and public buildings; energy supply carbon emission data refers to the carbon emission data in the process of fossil energy mining, processing, transportation and use; agricultural carbon emission data refers to the greenhouse gas emission data generated by fertilizer use, animal husbandry, deforestation and land use change activities in agricultural activities; waste treatment carbon emission data refers to the greenhouse gas emission data generated by garbage incineration and landfill treatment methods; indirect carbon emission data refers to purchased electricity, heating and cooling, as well as other indirect emission data caused by internal urban activities but generated outside the jurisdiction.

[0115] Specifically, the main sources of urban carbon emissions, from transportation, industry to construction, energy supply, agriculture and waste treatment, as well as indirect emissions, ensure the comprehensive monitoring and assessment of urban carbon emissions. The subdivision of carbon emission data into different types of emission sources helps to identify the main emission sources and key emission links. Through in-depth analysis of various types of emission data, it is possible to more accurately locate emission reduction potential and key directions, improve the effectiveness and pertinence of emission reduction measures, and through integrated learning methods, make full use of the correlation and complementarity between various carbon emission data to improve the accuracy and reliability of data analysis and prediction. At the same time, integrated learning can also dig out the deep laws and knowledge hidden behind the data, and provide more scientific decision-making support for urban green planning.

[0116] Carbon emission data labeling module, used for:

[0117] Confirm each type of carbon emission data and uniquely code each type of carbon emission data;

[0118] Timestamp the acquisition time of each type of carbon emission data;

[0119] Using a hash algorithm to convert one or more field values ​​of each type of carbon emission data into a unique hash code, and using the hash code as a serial number;

[0120] The serial number and timestamp are uniquely coded to generate the carbon emission data, and the generated uniquely coded carbon emission data is marked as the carbon emission data to be processed.

[0121] Specifically, data standardization is achieved by uniquely coding and labeling each type of carbon emission data, confirming the timestamp, and using the hash algorithm to generate a serial number. This standardization not only improves the readability and manageability of the data, but also facilitates subsequent data analysis, processing, and integration. By uniquely coding and labeling each type of carbon emission data and confirming the timestamp of its acquisition time, the uniqueness and traceability of the data are ensured. The hash algorithm is used to convert one or more field values ​​of the carbon emission data into a unique hash code, and the hash code is used as a serial number to achieve data standardization and normalization. This not only improves the readability and manageability of the data, but also reduces the complexity and error rate of data processing. Since the carbon emission data has been uniquely coded and marked with a pending status, it can be processed and utilized more efficiently during data integration and learning.

[0122] The carbon emission data processing unit is also used for:

[0123] Preprocess the carbon emission data to be processed;

[0124] The data preprocessing process is to sequentially clean, integrate, convert, reduce dimension and verify the carbon emission data to be processed;

[0125] First, the carbon emission data to be processed is cleaned. Data cleaning is to correct or delete missing values, erroneous values ​​and duplicate values ​​in the data;

[0126] After data cleaning is completed, data integration is performed to merge data from different data sources in the carbon emission data to be processed into one data set;

[0127] After the data integration is completed, data conversion is performed to convert data of different dimensions in the carbon emission data to be processed into the same dimension;

[0128] After the data conversion is completed, data dimension reduction is performed. Data dimension reduction is to convert the carbon emission data to be processed into a new coordinate system through linear transformation, so that the first largest variance of any projection is on the first coordinate, the second largest variance is on the second coordinate, and so on;

[0129] After the data dimension reduction is completed, data verification is performed to detect the data range, logic and missing values ​​of the data after data cleaning, data integration, data conversion and data dimension reduction;

[0130] After data verification is completed, standardized carbon emission data is obtained.

[0131] Specifically, the data preprocessing process includes data cleaning, data integration, data conversion, data dimensionality reduction and data verification. These links work together on the carbon emission data to be processed, effectively improving the quality of the data. Through the automated data processing process, the efficiency of data processing can be greatly improved. It is no longer necessary to manually check and process the data one by one, but it can be done automatically through the program, which greatly saves time and labor costs. At the same time, the application of technologies such as hash algorithms has further improved the efficiency of data processing. Data cleaning removes missing values, erroneous values ​​and duplicate values, data integration merges data from different data sources, data conversion unifies the dimensions of the data, data dimensionality reduction reduces the redundancy and complexity of the data, and data verification ensures the accuracy and completeness of the data. These steps together improve the reliability and accuracy of the data.

[0132] Specifically, data cleaning involves correcting or deleting erroneous data, including:

[0133] Extract erroneous data that occurs during data cleaning;

[0134] Extracting two pieces of data information adjacent to the collection time of the erroneous data as the data information at the previous time and the data information at the next time;

[0135] Extract data cleaning history records with corrected error data;

[0136] The error degree coefficient corresponding to the error data is obtained by using the previous data information and the subsequent data information in combination with the data cleaning history record and the corrected error data; wherein the error degree coefficient is obtained by the following formula:

[0137]

[0138] Among them, R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X x and X h They represent the data values ​​corresponding to the previous data information and the subsequent data information respectively; n represents the number of corrected error data in the data cleaning history record; X i represents the corrected data value corresponding to the corrected erroneous data in the i-th data cleaning history record; X yi Indicates the data value corresponding to the corrected erroneous data in the i-th data cleaning history record;

[0139] Comparing the error degree coefficient corresponding to the error data with a preset degree coefficient threshold;

[0140] When the error degree coefficient corresponding to the erroneous data exceeds a preset degree coefficient threshold, the erroneous data is deleted;

[0141] When the error degree coefficient corresponding to the erroneous data does not exceed a preset degree coefficient threshold, the erroneous data is corrected.

[0142] The technical effect of the above technical solution is: by identifying erroneous data in the cleaning process, extracting adjacent data information, and using historical cleaning records to evaluate the degree of error, the technical solution can more accurately identify and process erroneous data, thereby improving the overall quality of the data. The solution not only relies on the current erroneous data itself, but also combines the information of adjacent data points and the erroneous data and its correction in the historical cleaning records. This multi-dimensional consideration makes the processing of erroneous data more intelligent and meticulous.

[0143] By setting the error degree coefficient and the degree coefficient threshold, the technical solution can flexibly adjust the strictness of error handling according to the actual application scenarios and requirements, enhancing its adaptability in different data sets and application scenarios. Automatically assessing the error degree and deciding whether to delete or correct the erroneous data based on the threshold reduces the need for manual review and intervention and improves the efficiency of data processing. By recording and analyzing the erroneous data and its correction in the historical cleaning records, the technical solution helps to continuously optimize the process and rules of data cleaning and further improve the accuracy and efficiency of data processing.

[0144] In summary, this technical solution effectively improves data quality, enhances the intelligence and automation level of data processing, and improves the flexibility and efficiency of the data cleaning process through a refined erroneous data processing strategy.

[0145] Specifically, when the error degree coefficient corresponding to the erroneous data does not exceed a preset degree coefficient threshold, the erroneous data is corrected, including:

[0146] Extracting a plurality of correct data before the collection time of the erroneous data, wherein the value range of the correct data is 50-100;

[0147] Extracting the subsequent data information corresponding to the erroneous data;

[0148] The modification amount corresponding to the erroneous data is obtained by using the multiple correct data collected before the time and the data information corresponding to the erroneous data at the time after the time; wherein the modification amount corresponding to the erroneous data is obtained by the following formula:

[0149]

[0150] Among them, ΔX represents the modification amount corresponding to the erroneous data; R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X hrepresents the data value corresponding to the data information at the later time; m represents the number of correct data corresponding to the previous collection time; X j represents the data value corresponding to the correct data before the jth acquisition time; w represents the bit coefficient, when (X e -X h )>0, then let w=1; when (X e -X h )<0, then let w=-1;

[0151] The modified error data is obtained by combining the modification amount corresponding to the error data with the error data, wherein the modified error data is obtained by the following formula:

[0152] X s =X e +ΔX

[0153] Among them, X s Indicates the corrected error data; X e represents the data value corresponding to the erroneous data; ΔX represents the modification amount corresponding to the erroneous data.

[0154] The technical effect of the above technical solution is: the solution calculates a reasonable modification amount (ΔX) by comprehensively considering the correct data points before and after the erroneous data (multiple correct data before the collection time and data information after the collection time), as well as the relationship between them and the erroneous data. This correction method based on multiple reference points can more accurately adjust the erroneous data to make it closer to the true value.

[0155] By introducing the bit coefficient w, the scheme can dynamically adjust the direction of correction according to the relative size relationship between the erroneous data and the subsequent data information (i.e., the sign of (Xe-Xh)). When the erroneous data is larger than the subsequent data, the correction amount will reduce the erroneous data; otherwise, the erroneous data will be increased, ensuring the correctness of the correction direction. By correcting the erroneous data, the scheme helps to reduce outliers in the data set and improve the overall consistency and reliability of the data. This is crucial for subsequent data analysis and mining work, because conclusions based on consistent and reliable data sets are often more accurate and reliable. The entire correction process is highly automated, reducing the need for human intervention. This not only improves the efficiency of data processing, but also reduces the risk of human error. Automated processing also enables the scheme to be applied to large-scale data sets, further leveraging its advantages.

[0156] The parameters in this scheme (such as the bit coefficient w) can be adjusted and optimized according to specific application scenarios to meet the processing requirements of different data sets and error types. This flexibility makes the scheme have a wide range of application prospects.

[0157] In summary, this technical solution effectively improves the accuracy and efficiency of data correction through a sophisticated erroneous data correction strategy, enhances the consistency and reliability of data, and provides strong support for subsequent data analysis and mining work.

[0158] In order to solve the problem in the existing technology that when building a model with urban carbon emission data, there is no better selection of integrated learning model for the construction process, which leads to the reduction of the efficiency of integrated learning of carbon emission data, please refer to Figure 1 and Figure 2 , this embodiment provides the following technical solutions:

[0159] Processes data model building blocks and is also used to:

[0160] The standardized carbon emission data is divided into a data set, and the data set is divided into a training set, a validation set and a test set;

[0161] Import the training set into the neural network learning machine for data training;

[0162] Among them, the data training process is to first forward propagate the training set, where the training set is propagated from low level to high level;

[0163] When the data result obtained by propagation is not within the standard propagation threshold, back propagation is performed, where back propagation is to propagate the error from the high level to the low level for training;

[0164] Among them, the process of propagation training is:

[0165] First, the weights of the training set are initialized. After the setting is completed, the training set is forward propagated through the convolution layer, downsampling layer, and fully connected layer to obtain the output value;

[0166] When the error is greater than the expected value, the error is passed back to the network, and the errors of the fully connected layer, downsampling layer, and convolutional layer are obtained in turn;

[0167] Among them, the error of each layer is the total error of the network; when the error is equal to or less than the expected value, the training is completed.

[0168] After the training of the training set is completed, the validation set is used to evaluate the model hyperparameter performance, where the Bayesian optimization method is used to evaluate the hyperparameter performance;

[0169] Model hyperparameters include the learning rate of the validation set, batch size, number of hidden layers, number of hidden layer nodes, and regularization coefficient;

[0170] During the model hyperparameter performance evaluation process, if the threshold parameter of the model hyperparameter performance of the validation set reaches the critical threshold point, the evaluation is stopped;

[0171] According to the model hyperparameter performance evaluation results of the validation set, select the ensemble model with the best evaluation performance;

[0172] The ensemble model with the best performance is evaluated on the test set for performance prediction, and whether to adjust the performance index is determined based on the performance prediction results;

[0173] Among them, the performance indicators include the accuracy, recall, F1 score, and AUC-ROC of the integrated model;

[0174] Finally, an integrated learning model of standardized carbon emission data is obtained, and the integrated learning model is labeled as the target carbon emission data.

[0175] Specifically, by dividing the standardized carbon emission data into a reasonable data set, we can obtain a training set, a validation set, and a test set, which helps the model to conduct targeted learning and evaluation at different stages. The training set is used for model learning and optimization, the validation set is used for hyperparameter performance evaluation, and the test set is used for final performance prediction. This division method improves the efficiency and effectiveness of data use. The model training process adopts a combination of forward propagation and back propagation. Through layer-by-layer processing of convolutional layers, downsampling layers, and fully connected layers, in-depth learning and feature extraction of carbon emission data are achieved. At the same time, by setting a standard propagation threshold and timely back propagation training for results that do not meet expectations, the performance of the model is further optimized. Using the Bayesian optimization method to evaluate the performance of the model hyperparameters, the optimal hyperparameter combination can be found more efficiently. By optimizing the hyperparameters of the validation set, such as the learning rate, batch size, number of hidden layers, number of hidden layer nodes, and regularization coefficient, the performance of the model can be significantly improved. At the same time, the threshold parameters of the validation set model hyperparameter performance are set, and the evaluation is stopped when the critical threshold point is reached, avoiding unnecessary waste of computing resources. In the performance evaluation process, multiple performance indicators including accuracy, recall rate, F1 score, AUC-ROC, etc. are used to conduct a comprehensive and accurate evaluation of the model performance.

[0176] In order to solve the problem that the existing technology does not use the city carbon emission data for further risk assessment and early warning judgment, which leads to reduced subsequent processing capacity, please refer to Figure 1 and Figure 2 , this embodiment provides the following technical solutions:

[0177] The model data evaluation unit is also used for:

[0178] Retrieve urban planning carbon emission data from the database;

[0179] Perform curve data conversion on each type of carbon emission data in the retrieved urban planning carbon emission data and target carbon emission data;

[0180] Overlap the urban planning carbon emission data with the carbon emission curve data of the same type in the target carbon emission data;

[0181] After the curves overlap, a threshold assessment is performed on the target carbon emission data based on the overlap results;

[0182] The threshold assessment is to calculate the area of ​​non-overlapping areas in the curve overlap results of the urban planning carbon emission data and the carbon emission curve data of the same type in the target carbon emission data, and perform the threshold assessment based on the area of ​​non-overlapping areas;

[0183] The threshold assessment results are divided into low-risk emissions, medium-risk emissions, and high-risk emissions;

[0184] The threshold assessment results are marked as carbon emission decision data.

[0185] Specifically, by converting the urban planning carbon emission data and the target carbon emission data into curve data and overlapping them, the differences and similarities between the two can be intuitively displayed. This graphical representation makes the data comparison more intuitive and easy to understand, which helps decision makers quickly grasp the changing trends and differences of carbon emission data. It is a quantitative and accurate method to use the area of ​​non-overlapping areas in the curve overlap results for threshold assessment. By calculating the area of ​​non-overlapping areas, the gap between urban planning carbon emissions and target carbon emissions can be objectively reflected. The threshold assessment results are divided into low-risk emissions, medium-risk emissions and high-risk emissions, and the results are marked, which helps decision makers take corresponding measures according to different risk levels. By evaluating the differences between urban planning carbon emission data and target carbon emission data, potential problems in urban planning can be discovered in a timely manner, such as high emission areas and unreasonable emission trends.

[0186] The carbon emission assessment data decision unit is also used for:

[0187] Provide different levels of early warning based on carbon emission decision data;

[0188] The warning levels are level one, level two and level three;

[0189] Obtain different types of carbon emissions in different levels of early warning;

[0190] Formulate plans in different directions according to different types;

[0191] The formulated plan and threshold evaluation results are transmitted to the display terminal for display.

[0192] Specifically, according to the carbon emission decision data, different levels of early warning (level 1, level 2, level 3) can be carried out to achieve accurate monitoring of urban carbon emissions. This hierarchical early warning mechanism can trigger corresponding response measures in time according to different levels of carbon emissions, effectively avoid environmental problems caused by excessive carbon emissions, obtain different types of carbon emissions in different levels of early warning, and formulate plans in different directions accordingly, which reflects the idea of ​​flexible response and differentiated management. Formulating more targeted solutions for different types and levels of carbon emission problems will help improve governance effects and resource utilization efficiency. Through integrated learning algorithms, a large amount of carbon emission data is analyzed and mined to extract valuable information and knowledge, providing a scientific basis for early warning and plan formulation. The formulated plans and threshold assessment results are transmitted to the display terminal for display, which helps to achieve efficient collaboration and information sharing. All relevant departments and personnel can understand the carbon emission situation and governance plan in real time, communicate and cooperate in a timely manner, and jointly promote the implementation of urban green planning.

[0193] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0194] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. The urban green planning system based on urban carbon emission integrated learning is characterized by: include: Carbon emission data acquisition unit, used for: Retrieving different types of carbon emission data from the database, uniquely coding each type of carbon emission data, and marking the carbon emission data after the unique coding is completed as carbon emission data to be processed; Carbon emission data processing unit, used for: The carbon emission data to be processed are preprocessed, and after the data preprocessing is completed, standardized carbon emission data are obtained; Processing data model building blocks for: The standardized carbon emission data is used to construct an integrated learning model, and after the integrated learning model is constructed, the target carbon emission data is obtained; Model data evaluation unit for: Compare the target carbon emission data with the urban planning carbon emission data, conduct a threshold assessment on the target carbon emission data based on the data comparison results, and obtain the carbon emission decision data after the threshold assessment is completed; Carbon emission assessment data decision unit, used for: Formulate targeted decision-making plans based on the assessment scope of carbon emission decision data, and transmit the formulated plans and carbon emission decision data to the display terminal for display of data and plans; The carbon emission data processing unit is also used for: Preprocess the carbon emission data to be processed; The data preprocessing process is to sequentially clean, integrate, convert, reduce dimension and verify the carbon emission data to be processed; First, the carbon emission data to be processed is cleaned. Data cleaning is to correct or delete missing values, erroneous values ​​and duplicate values ​​in the data; After data cleaning is completed, data integration is performed to merge data from different data sources in the carbon emission data to be processed into one data set; After the data integration is completed, data conversion is performed to convert data of different dimensions in the carbon emission data to be processed into the same dimension; After the data conversion is completed, data dimension reduction is performed. Data dimension reduction is to convert the carbon emission data to be processed into a new coordinate system through linear transformation, so that the first largest variance of any projection is on the first coordinate, the second largest variance is on the second coordinate, and so on; After the data dimension reduction is completed, data verification is performed to detect the data range, logic and missing values ​​of the data after data cleaning, data integration, data conversion and data dimension reduction; After data verification is completed, standardized carbon emission data is obtained; Data cleaning corrects or deletes erroneous data, including: Extract erroneous data that occurs during data cleaning; Extracting two pieces of data information adjacent to the collection time of the erroneous data as the data information at the previous time and the data information at the next time; Extract data cleaning history records with corrected error data; The error degree coefficient corresponding to the error data is obtained by using the previous data information and the subsequent data information in combination with the data cleaning history record and the corrected error data; wherein the error degree coefficient is obtained by the following formula: Among them, R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X x and X h They represent the data values ​​corresponding to the previous data information and the subsequent data information respectively; n represents the number of corrected error data in the data cleaning history record; X i represents the corrected data value corresponding to the corrected erroneous data in the i-th data cleaning history record; X yi Indicates the data value corresponding to the corrected erroneous data in the i-th data cleaning history record; Comparing the error degree coefficient corresponding to the error data with a preset degree coefficient threshold; When the error degree coefficient corresponding to the erroneous data exceeds a preset degree coefficient threshold, the erroneous data is deleted; When the error degree coefficient corresponding to the erroneous data does not exceed the preset degree coefficient threshold, the erroneous data is corrected; When the error degree coefficient corresponding to the erroneous data does not exceed the preset degree coefficient threshold, the erroneous data is corrected, including: Extracting a plurality of correct data before the collection time of the erroneous data, wherein the value range of the correct data is 50-100; Extracting the subsequent data information corresponding to the erroneous data; The modification amount corresponding to the erroneous data is obtained by using the multiple correct data collected before the time and the data information corresponding to the erroneous data at the time after the time; wherein the modification amount corresponding to the erroneous data is obtained by the following formula: Among them, ΔX represents the modification amount corresponding to the erroneous data; R represents the error degree coefficient; X e Indicates the data value corresponding to the error data; X h represents the data value corresponding to the data information at the later time; m represents the number of correct data corresponding to the previous collection time; X j represents the data value corresponding to the correct data before the jth acquisition time; w represents the bit coefficient, when (X e -X h )>0, then let w=1; when (X e -X h )<0, then let w=-1; in the case of R=1, X e -X h =0 The modified error data is obtained by combining the modification amount corresponding to the error data with the error data, wherein the modified error data is obtained by the following formula: Among them, X s Indicates the corrected error data; X e represents the data value corresponding to the erroneous data; ΔX represents the modification amount corresponding to the erroneous data.

2. The urban green planning system based on urban carbon emission integrated learning according to claim 1 is characterized by: The carbon emission data acquisition unit comprises: Carbon emission data classification module, used for: Confirm different types of carbon emission data from the database; The types of carbon emissions include transportation carbon emissions data, industrial carbon emissions data, building carbon emissions data, energy supply carbon emissions data, agricultural carbon emissions data, waste treatment carbon emissions data and indirect carbon emissions data; Among them, transportation carbon emission data refers to the energy consumption data of private cars, public transportation, rail transportation, aviation, water transportation and various means of transportation; industrial carbon emission data refers to the greenhouse gas emission data in the industrial production process; building carbon emission data refers to the energy consumption data of heating, air conditioning, hot water supply and lighting systems in residential and public buildings; energy supply carbon emission data refers to the carbon emission data in the process of fossil energy mining, processing, transportation and use; agricultural carbon emission data refers to the greenhouse gas emission data generated by fertilizer use, animal husbandry, deforestation and land use change activities in agricultural activities; waste treatment carbon emission data refers to the greenhouse gas emission data generated by garbage incineration and landfill treatment methods; indirect carbon emission data refers to purchased electricity, heating and cooling, as well as other indirect emission data caused by internal urban activities but generated outside the jurisdiction.

3. The urban green planning system based on urban carbon emission integrated learning according to claim 2 is characterized by: The carbon emission data acquisition unit further includes: Carbon emission data labeling module, used for: Confirm each type of carbon emission data and uniquely code each type of carbon emission data; Timestamp the acquisition time of each type of carbon emission data; Using a hash algorithm to convert one or more field values ​​of each type of carbon emission data into a unique hash code, and using the hash code as a serial number; The serial number and timestamp are uniquely coded to generate the carbon emission data, and the generated uniquely coded carbon emission data is marked as the carbon emission data to be processed.

4. The urban green planning system based on urban carbon emission integrated learning according to claim 3 is characterized by: The processing data model building unit is further used for: The standardized carbon emission data is divided into a data set, and the data set is divided into a training set, a validation set and a test set; Import the training set into the neural network learning machine for data training; Among them, the data training process is to first forward propagate the training set, where the training set is propagated from low level to high level; When the data result obtained by propagation is not within the standard propagation threshold, back propagation is performed, where back propagation is to propagate the error from the high level to the low level for training; Among them, the process of propagation training is: First, the weights of the training set are initialized. After the setting is completed, the training set is forward propagated through the convolution layer, downsampling layer, and fully connected layer to obtain the output value; When the error is greater than the expected value, the error is passed back to the network, and the errors of the fully connected layer, downsampling layer, and convolutional layer are obtained in turn; Among them, the error of each layer is the total error of the network; when the error is equal to or less than the expected value, the training is completed.

5. The urban green planning system based on urban carbon emission integrated learning according to claim 4 is characterized by: The processing data model building unit is further used for: After the training of the training set is completed, the validation set is used to evaluate the model hyperparameter performance, where the Bayesian optimization method is used to evaluate the hyperparameter performance; Model hyperparameters include the learning rate of the validation set, batch size, number of hidden layers, number of hidden layer nodes, and regularization coefficient; During the model hyperparameter performance evaluation process, if the threshold parameter of the model hyperparameter performance of the validation set reaches the critical threshold point, the evaluation is stopped; According to the model hyperparameter performance evaluation results of the validation set, select the ensemble model with the best evaluation performance; The ensemble model with the best performance is evaluated on the test set for performance prediction, and whether to adjust the performance index is determined based on the performance prediction results; Among them, the performance indicators include the accuracy, recall, F1 score, and AUC-ROC of the integrated model; Finally, an integrated learning model of standardized carbon emission data is obtained, and the integrated learning model is labeled as the target carbon emission data.

6. The urban green planning system based on urban carbon emission integrated learning according to claim 5 is characterized by: The model data evaluation unit is further used for: Retrieve urban planning carbon emission data from the database; Perform curve data conversion on each type of carbon emission data in the retrieved urban planning carbon emission data and target carbon emission data; Overlap the urban planning carbon emission data with the carbon emission curve data of the same type in the target carbon emission data; After the curves overlap, a threshold assessment is performed on the target carbon emission data based on the overlap results; The threshold assessment is to calculate the area of ​​non-overlapping areas in the curve overlap results of the urban planning carbon emission data and the carbon emission curve data of the same type in the target carbon emission data, and perform the threshold assessment based on the area of ​​non-overlapping areas; The threshold assessment results are divided into low-risk emissions, medium-risk emissions, and high-risk emissions; The threshold assessment results are marked as carbon emission decision data.

7. The urban green planning system based on urban carbon emission integrated learning according to claim 6 is characterized by: The carbon emission assessment data decision unit is also used to: Provide different levels of early warning based on carbon emission decision data; The warning levels are level one, level two and level three; Obtain different types of carbon emissions in different levels of early warning; Formulate plans in different directions according to different types; The formulated plan and threshold evaluation results are transmitted to the display terminal for display.

Images