Road construction project carbon emission ratio toughness analysis method

Abstract

The invention provides a road construction project carbon emission ratio toughness analysis method, and belongs to the technical field of carbon emission, machine learning and big data analysis and mining, and the method comprises the steps: obtaining a historical change trend of a carbon emission ratio from historical data of a road construction project, and constructing a dynamic change model of the carbon emission ratio in combination with real-time monitoring data; determining the influence degree and the action mechanism of the interference factors on the carbon emission ratio dynamic change model by analyzing the interaction relationship between the carbon emission ratio and the interference factors; key fluctuation characteristics are extracted from the corrected change trend of the carbon emission ratio, and whether fluctuation of the carbon emission ratio exceeds an acceptable range or not is judged in combination with a preset threshold value; if the fluctuation of the carbon emission ratio exceeds a preset threshold value, performing noise reduction processing on the data by adopting an adaptive filtering algorithm to generate noise-reduced carbon emission ratio data; and training the segmented analysis result by adopting a random forest model in a machine learning algorithm to generate a prediction model of the carbon emission ratio.

Description

Technical Field

[0001] The present invention relates to the fields of carbon emissions, machine learning and big data analysis and mining technology, and in particular to a carbon emission ratio resilience analysis method for a road construction project. Background Art

[0002] In the road construction environmental impact assessment method and system based on life cycle analysis, the ratio resilience technology faces some technical problems and contradictions. First, when evaluating the fluctuation of the carbon emission ratio per unit mileage during road construction, it is necessary to consider various external interference factors, such as extreme weather conditions, changes in construction technology, fluctuations in raw material supply, etc. These interference factors are often uncertain and dynamic, which brings difficulties to the ratio resilience analysis. Secondly, the ratio resilience analysis requires a large amount of historical data and real-time monitoring data as support, but in actual road construction projects, due to the lack of a perfect data collection and transmission mechanism, the quality and availability of data are insufficient, which affects the accuracy and reliability of the ratio resilience analysis. Furthermore, there are differences in the characteristics and environmental conditions of different road construction projects. The ratio resilience analysis needs to set parameters and optimize models for specific projects, but there is currently a lack of a set of standardized ratio resilience analysis processes and specifications, and the comparability and consistency of results between different evaluation subjects need to be improved. Finally, the ratio resilience analysis involves complex mathematical models and calculations, which requires high professional capabilities of the evaluators. At the same time, the analysis process also consumes a lot of computing resources and time, which increases the cost and cycle of environmental impact assessment. Summary of the invention

[0003] The present invention provides a carbon emission ratio resilience analysis method for a road construction project, which mainly includes:

[0004] Obtain the historical change trend of carbon emission ratio from the historical data of road construction projects, and build a dynamic change model of carbon emission ratio by combining with real-time monitoring data;

[0005] In view of the uncertainty and dynamics of external interference factors, the time series analysis method is used to quantitatively model interference factors such as extreme weather conditions, construction process changes, and fluctuations in raw material supply, and generate a dynamic weight matrix of interference factors;

[0006] By analyzing the interaction between carbon emission ratio and interference factors, the influence degree and mechanism of interference factors on the dynamic change model of carbon emission ratio are determined;

[0007] According to the dynamic weight matrix and interaction relationship of interference factors, the dynamic change model of carbon emission ratio is corrected to obtain the corrected trend of carbon emission ratio;

[0008] Extract key fluctuation features from the modified carbon emission ratio change trend, and combine with the preset threshold to determine whether the fluctuation of the carbon emission ratio exceeds the acceptable range;

[0009] If the carbon emission ratio fluctuation exceeds the preset threshold, an adaptive filtering algorithm is used to reduce the noise of the data to generate the carbon emission ratio data after noise reduction;

[0010] Based on the de-noised carbon emission ratio data and the pre-established standardized analysis process, the carbon emission ratio is analyzed in sections to determine the changing pattern of the carbon emission ratio in each section;

[0011] The random forest model in the machine learning algorithm is used to train the segmented analysis results to generate a prediction model for the carbon emission ratio;

[0012] According to the results of the prediction model and the optimization requirements of parameter settings, the changing trend of the carbon emission ratio is dynamically adjusted to generate the final carbon emission ratio resilience analysis report;

[0013] The key information in the carbon emission ratio resilience analysis report is fed back to the dynamic weight matrix of interference factors to achieve adaptive updating of the dynamic weight matrix and improve the accuracy and reliability of the carbon emission ratio prediction model.

[0014] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0015] The present invention discloses a method for analyzing the resilience of carbon emission ratios in road construction projects. The method uses big data analysis technology to construct a dynamic change model of the carbon emission ratio by analyzing historical data and real-time monitoring data, and establishes a dynamic weight matrix for external interference factors such as extreme weather and construction process changes. The present invention innovatively combines time series analysis with machine learning to perform segmented analysis and prediction of the carbon emission ratio. The abnormal fluctuation data is denoised by an adaptive filtering algorithm, thereby improving the accuracy of the prediction model. Finally, the present invention generates a carbon emission ratio resilience analysis report and realizes adaptive updating of the dynamic weight matrix, effectively improving the accuracy and reliability of carbon emission management of road construction projects. The method provides an innovative solution for low-carbon management of road construction projects and has important practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a flow chart of a carbon emission ratio resilience analysis method for a road construction project.

[0017] Figure 2 It is a schematic diagram of a carbon emission ratio resilience analysis method for a road construction project of the present invention.

[0018] Figure 3It is another schematic diagram of a carbon emission ratio resilience analysis method for a road construction project of the present invention. DETAILED DESCRIPTION

[0019] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0020] like Figure 1 , Figure 2 and Figure 3 In this embodiment, a road construction project carbon emission ratio resilience analysis method may specifically include:

[0021] Step S101, obtaining the historical change trend of the carbon emission ratio from the historical data of the road construction project, and building a dynamic change model of the carbon emission ratio in combination with the real-time monitoring data.

[0022] The historical carbon emission ratio data and real-time carbon emission monitoring data of the road construction project were obtained, and the historical data were cleaned, denoised and normalized to obtain a standardized carbon emission ratio time series data set. The standardized carbon emission ratio data set was decomposed using a time series decomposition algorithm to obtain three components: the long-term trend, seasonal changes, and random fluctuations of the carbon emission ratio. Based on the decomposed long-term trend component of the carbon emission ratio, the gray prediction model was used to perform a trend extrapolation forecast of the carbon emission ratio to obtain the changing trend of the carbon emission ratio in the future. The formula for calculating the trend forecast value of the carbon emission ratio is as follows:

[0023] F t =βT t +(1-β)F t-1

[0024] Where Ft is the trend forecast value at time t, Tt is the actual trend value at the current moment, and β is the smoothing parameter.

[0025] According to the seasonal variation component of the carbon emission ratio obtained by decomposition, the seasonal autoregressive moving average model SARIMA is used to model the periodic changes and obtain the seasonal fluctuation pattern and prediction results. The trend prediction results and seasonal prediction results are combined to obtain a comprehensive carbon emission ratio prediction benchmark value, which is dynamically integrated with the real-time monitored carbon emission data to correct the prediction deviation in real time. The fused and optimized carbon emission ratio prediction model is backtested and parameterized, and the accuracy and stability of the model are evaluated by indicators such as root mean square error and mean absolute percentage error. The optimal dynamic prediction model for carbon emission ratio is determined and deployed in the production environment to obtain carbon emission monitoring data of road construction projects in real time, automatically update the prediction results, and provide decision support for carbon emission control.

[0026] Specifically, in order to obtain the historical change trend of carbon emission ratio, the historical data of road construction projects were first collected and preprocessed. Carbon emission data of 1,000 road construction projects from 2015 to 2020 were collected, including fossil fuel consumption during the construction phase, the use time of construction machinery, etc. Data cleaning and feature engineering were used to eliminate 5% of abnormal data, and key features such as energy consumption intensity and construction intensity were extracted. Then, the carbon emission ratio was modeled using the time series analysis algorithm ARIMA, and the parameters were fitted by the least squares method to obtain the historical change model of the carbon emission ratio with an average error of 3%. Then, 100 real-time monitoring devices were deployed to collect energy consumption data from the construction site every 15 minutes and transmit it to the cloud. After receiving the data in the cloud, the Kalman filter algorithm was used to fuse the data, and the current carbon emission ratio was predicted in real time in combination with the historical model, and displayed in real time through the dashboard. When the predicted carbon emission ratio exceeds the threshold, the system will automatically issue an early warning to prompt the project manager to take energy-saving and emission reduction measures. By dynamically tracking the changes in the carbon emission ratio, the construction process can be finely managed to achieve green construction.

[0027] Step S102, in view of the uncertainty and dynamics of external interference factors, a time series analysis method is used to quantitatively model interference factors such as extreme weather conditions, construction process changes, and raw material supply fluctuations, and generate a dynamic weight matrix of interference factors.

[0028] In view of the uncertainty and dynamics of external interference factors, historical data, including extreme weather conditions, construction process change records, raw material supply fluctuation data, etc., are obtained to construct a time series data set. Data preprocessing technology is used to clean and normalize the time series data to remove noise and outliers and improve data quality. Time series analysis algorithms, such as ARIMA and LSTM, are applied to establish a prediction model for interference factors. Through learning and training of historical data, the dynamic change law of interference factors is captured. According to the output results of the prediction model, the weight matrix of interference factors is dynamically adjusted to quantify the impact of different interference factors on engineering project management. The dynamic weight matrix is ​​combined with various aspects of engineering project management, such as schedule management, cost management, quality management, etc., to formulate targeted response strategies and adjustment measures. During the project implementation process, real-time data of external interference factors is continuously collected and updated, and the prediction model is learned and optimized online to continuously improve the accuracy and adaptability of the prediction. Regularly evaluate the execution effect of the solution, and adjust and improve the solution in a timely manner by comparing the actual project management performance with the expected goals, so as to improve the robustness and resilience of engineering project management.

[0029] Specifically, in view of the uncertainty and dynamics of external interference factors, the time series analysis method is used for quantitative modeling. First, for extreme weather conditions, by establishing an ARIMA model and using historical weather data from the past 10 years, extreme weather conditions that may occur in the next year, such as heavy rain and strong winds, are predicted. According to the probability and duration of their occurrence, a dynamic weight matrix is ​​generated, and a weight value of 3 is assigned. Secondly, for construction process changes, by introducing Markov chains, the frequency and amplitude of construction process changes in the past 5 years are analyzed, and the possible process changes in the next 6 months are predicted. According to the degree of their impact on the construction period, a dynamic weight matrix is ​​generated, and a weight value of 4 is assigned. Finally, for raw material supply fluctuations, by establishing a VAR model, using raw material price and supply data from the past 3 years, supply fluctuations that may occur in the next 3 months are predicted. According to the degree of their impact on costs, a dynamic weight matrix is ​​generated, and a weight value of 3 is assigned. By comprehensively considering the dynamic weight matrix of these three aspects, a dynamic weight matrix of total interference factors is generated, providing a quantitative basis for subsequent construction schedule optimization and risk control.

[0030] Step S103, by analyzing the interaction relationship between the carbon emission ratio and the interference factors, the influence degree and action mechanism of the interference factors on the dynamic change model of the carbon emission ratio are determined.

[0031] The actual data of carbon emission ratio and the statistical data of each influencing factor in the historical period are obtained, and the data preprocessing is performed on the carbon emission ratio and each influencing factor to obtain the standardized time series data. The correlation analysis method is used to calculate the correlation coefficient matrix between the carbon emission ratio and each influencing factor, and the key influencing factors with a large correlation with the carbon emission ratio are screened out. According to the time series data of the carbon emission ratio and the key influencing factors, the VAR vector autoregression model is constructed. The influence degree and action path of the key influencing factors on the dynamic changes of the carbon emission ratio are analyzed by the impulse response function and variance decomposition methods. For the residual sequence fitted by the VAR model, the Granger causality test is used to further determine whether there is a causal relationship between the carbon emission ratio and the key influencing factors, and to determine the direction of action between each factor and the carbon emission ratio. Using machine learning algorithms such as support vector machine SVM or random forest, a nonlinear regression prediction model of the carbon emission ratio and key influencing factors is established, and the influence weight of each influencing factor on the carbon emission ratio is quantitatively evaluated through feature importance analysis. The analysis results of the comprehensive VAR model and machine learning model reveal the mechanism of action between key influencing factors and dynamic changes in carbon emission ratios from the perspectives of statistical association and causal mechanism, and construct a quantitative relationship model between carbon emission ratios and key factors. Based on the established carbon emission ratio change model, the impact of changes in key influencing factors on future carbon emission ratios under different scenarios is predicted, providing a quantitative evaluation basis and policy simulation tools for the formulation of carbon emission reduction countermeasures.

[0032] Specifically, first, historical data on carbon emission ratio and interference factors are collected, including monthly data for the past 10 years. Among them, the carbon emission ratio data is obtained by dividing the total carbon emissions by the total energy consumption, and the interference factors include economic growth rate, industrial structure, energy structure, etc. Then, the mathematical model between the carbon emission ratio and the interference factors is established using the multivariate regression analysis method. On this basis, the gray correlation between each interference factor and the carbon emission ratio is calculated by the gray correlation analysis method. The greater the correlation, the greater the impact of the factor on the carbon emission ratio. For example, the calculation results show that the gray correlation of the economic growth rate is 85, the industrial structure is 72, and the energy structure is 68, indicating that the economic growth rate has the greatest impact on the carbon emission ratio. Then, the system dynamics method is used to construct a dynamic change model of the carbon emission ratio to quantitatively analyze the mechanism of interference factors. The feedback relationship between various factors is considered in the model, such as economic growth will drive energy consumption and thus lead to an increase in carbon emissions. By adjusting the model parameters, the future trend of carbon emission ratio changes is predicted. Finally, a scenario analysis is carried out to set different scenarios for changes in interference factors, such as economic growth slowdown, industrial structure optimization, and clean energy structure, to evaluate their impact on the carbon emission ratio. The analysis results show that in the scenario where the economic growth rate slows by 1 percentage point, the carbon emission ratio decreases by 2% annually; in the scenario where the industrial structure is optimized, the carbon emission ratio decreases by 35%; in the scenario where the energy structure is clean, the carbon emission ratio decreases by 6%. Based on this, corresponding carbon emission reduction countermeasures and suggestions are proposed to provide a scientific basis for achieving the goals of carbon peak and carbon neutrality.

[0033] Step S104, according to the dynamic weight matrix and interaction relationship of the interference factors, the dynamic change model of the carbon emission ratio is corrected to obtain a corrected carbon emission ratio change trend.

[0034] According to the historical data of carbon emission related factors, the dynamic weight matrix of carbon emission factors is constructed, and the weight coefficient is obtained through machine learning algorithm training. The interaction relationship data between carbon emission factors are obtained, and the association rule mining algorithm is used to establish the interaction relationship model of carbon emission factors. The dynamic weight matrix is ​​integrated with the interaction relationship model to obtain a dynamic change model of carbon emission that comprehensively considers weights and interaction effects. For the original dynamic change model of carbon emission ratio, the correction term in the comprehensive dynamic change model is introduced to optimize and correct the carbon emission ratio prediction model. The random forest regression algorithm is used, and the optimized carbon emission ratio prediction model is obtained through training with a large amount of historical data based on the corrected dynamic change model of carbon emission ratio. Using the optimized carbon emission ratio prediction model, combined with the prediction data of carbon emission related factors under different scenarios in the future, the dynamic change trend of carbon emission ratio in different time ranges in the future is simulated and predicted. According to the predicted change trend of carbon emission ratio, the high-risk abnormal interval of carbon emission ratio is identified, and targeted carbon emission reduction optimization and control measures are formulated to promote the realization of carbon peak and carbon neutrality goals.

[0035] Specifically, first, the dynamic weight matrix of interference factors is defined. It is assumed that there are five interference factors, namely population, economy, energy structure, technology level and policies and regulations, and their influence weights on carbon emissions are 3, 25, 2, 15 and 1 respectively. Then, according to the interaction relationship between each factor, a system dynamics model is constructed, and the Vensim software is used for simulation to obtain the dynamic influence trend of interference factors on carbon emissions. Next, the dynamic weight matrix of interference factors is integrated with the dynamic change model of carbon emission ratio, and the weighted average method is used to correct the original model. For example, assuming that the carbon emission ratio in a certain year is 5%, and the weight vector of interference factors is [35, 22, 18, 16, 09], the corrected carbon emission ratio is 5% × (35 + 22 + 18 + 16 + 09) = 5%. Finally, the corrected carbon emission ratio is substituted into the dynamic change model to recalculate the carbon emission trend in the next few years. By comparing the results before and after the correction, it can be found that after considering the interference factors, the change trend of the carbon emission ratio is more gentle, which is closer to the actual situation. This method can effectively improve the accuracy of carbon emission forecasts and provide a scientific basis for formulating reasonable emission reduction policies.

[0036] Step S105 , extracting key fluctuation features from the corrected carbon emission ratio variation trend, and combining with a preset threshold value, determining whether the fluctuation of the carbon emission ratio exceeds an acceptable range.

[0037] Obtain the corrected carbon emission ratio change trend data, analyze and process the data to obtain at least one key fluctuation feature; obtain a preset acceptable fluctuation threshold of the carbon emission ratio, and compare the key fluctuation feature with the threshold; if the key fluctuation feature exceeds the threshold, determine that the carbon emission ratio fluctuation exceeds the acceptable range, and output warning information; for the carbon emission ratio fluctuation that exceeds the acceptable range, obtain the corresponding carbon emission source data, analyze the source data, and determine the main emission source causing the fluctuation; based on the main emission source obtained by the analysis, obtain the corresponding historical emission data, and predict the carbon emission ratio change trend of the emission source in the future through a time series analysis algorithm; compare the predicted future carbon emission ratio change trend with the preset threshold again to determine whether the carbon emission ratio of the emission source is still at risk of exceeding the acceptable range; if the judgment result is that there is a risk of exceeding, output a warning prompt, and generate corresponding carbon emission control strategies and suggestions to guide relevant departments to reduce carbon emissions.

[0038] Specifically, in order to extract key fluctuation features from the trend of the revised carbon emission ratio, the wavelet analysis method can be used to perform multi-scale decomposition on the time series data to extract fluctuation information in different frequency ranges. First, the revised carbon emission ratio data is preprocessed to remove missing values ​​and outliers, and normalized to scale the data range to the [0,1] interval. Then, a suitable wavelet basis function (such as Daubechies wavelet) is selected to perform a 5-layer wavelet decomposition on the data to obtain the approximate coefficients and detail coefficients at different scales. Next, the energy ratio of the detail coefficients at each scale is calculated, that is, the sum of the squares of the detail coefficients at each layer is divided by the total energy to obtain the relative size of the fluctuation intensity in each frequency range. According to the preset threshold (such as 2), it is judged whether the energy ratio of each scale exceeds the threshold. If it exceeds, it indicates that there is a significant fluctuation in the frequency range. Finally, the extracted key fluctuation features are compared with the preset acceptable range to determine whether the fluctuation of the carbon emission ratio exceeds the control range. For example, if the energy ratio of high-frequency fluctuations in a certain period exceeds 3 and lasts for more than 3 months, it can be determined as an abnormal fluctuation, and further analysis of the cause and measures need to be taken. By real-time monitoring of the changing trend of the carbon emission ratio and combining it with wavelet analysis and threshold judgment, abnormal fluctuations can be discovered in a timely manner, providing a basis for formulating reasonable carbon emission control strategies.

[0039] Step S106: If the carbon emission ratio fluctuation exceeds a preset threshold, an adaptive filtering algorithm is used to perform noise reduction processing on the data to generate noise-reduced carbon emission ratio data.

[0040] According to the preset carbon emission ratio fluctuation threshold, the carbon emission ratio data is obtained in real time to determine whether the carbon emission ratio fluctuation exceeds the threshold; if the carbon emission ratio fluctuation exceeds the preset threshold, the Kalman filter algorithm is used to perform adaptive noise reduction processing on the obtained carbon emission ratio data; in the process of Kalman filter algorithm processing, by establishing a state space model of the carbon emission ratio data, the filter parameters are adaptively estimated to achieve optimized noise reduction of the carbon emission ratio data; after noise reduction processing, the noise-reduced and optimized carbon emission ratio data is obtained, and the data is used as the final calculation result of the carbon emission ratio; according to the noise-reduced and optimized carbon emission ratio data, it is determined whether the carbon emission ratio is within a reasonable range, and if it exceeds the reasonable range, the early warning mechanism is triggered; by statistically analyzing the carbon emission ratio data for multiple consecutive time periods, the long-term change trend of the carbon emission ratio is determined to provide a decision-making reference for carbon emission control; the carbon emission ratio data is correlated with the production process parameters, and the principal component analysis method is used to determine the correlation between each parameter and the carbon emission ratio, find out the key influencing factors, optimize the production process, and reduce carbon emissions.

[0041] Specifically, in practical applications, the preset threshold of carbon emission ratio fluctuation is first set to ±5%. When it is detected that the fluctuation range of carbon emission ratio in a certain period of time exceeds this threshold, the system will automatically trigger the adaptive filtering algorithm. Based on the principle of Kalman filtering, the algorithm dynamically adjusts the filtering parameters by adaptively estimating and updating the carbon emission ratio data to achieve the best noise reduction effect. Specifically, the algorithm first preprocesses the data to remove obvious outliers and noise interference. Then, by establishing a state space model of the carbon emission ratio, the optimal state variables and covariance matrix are recursively estimated using historical data and current observations. On this basis, the Kalman gain is adaptively adjusted to correct and smooth the carbon emission ratio at the current moment. After algorithm processing, the generated denoised carbon emission ratio data can more accurately reflect the actual emission level and provide reliable data support for subsequent carbon emission analysis and decision-making. At the same time, through data integration with the enterprise production management system, the denoised carbon emission ratio data can also be fed back to the production scheduling and optimization module in real time to guide the enterprise to conduct low-carbon production regulation and realize dynamic monitoring and intelligent management of carbon emissions.

[0042] Step S107, based on the denoised carbon emission ratio data and in combination with a pre-established standardized analysis process, the carbon emission ratio is analyzed in sections to determine the changing pattern of the carbon emission ratio in each section.

[0043] The original data of carbon emission ratio is obtained, and the original data is preprocessed, including denoising and outlier processing, to obtain the preprocessed carbon emission ratio data. According to the pre-established standardized analysis process, the preprocessed carbon emission ratio data is standardized to eliminate the influence of different dimensions and orders of magnitude. The segmented clustering algorithm, such as the K-Means algorithm, is used to divide the standardized carbon emission ratio data into several segments, and the data characteristics in each segment are similar. For each segment, the distribution characteristics of the internal carbon emission ratio data are analyzed, including statistics such as mean, variance, and skewness, to determine the overall change trend of the data in the segment. The mean values ​​of the carbon emission ratios of adjacent segments are compared. If the difference in the mean values ​​of adjacent segments exceeds the preset threshold, it is determined that the carbon emission ratio between adjacent segments has changed significantly, and the direction and magnitude of the change are recorded. Based on the change trend of the carbon emission ratio in each segment and the change between adjacent segments, a time series analysis method, such as the ARIMA model, is used to establish a quantitative model for the change of the carbon emission ratio to reveal the overall change law of the carbon emission ratio. The changing trends of the carbon emission ratios in each segment and the overall changing rules are compiled into an analysis report to provide data support for subsequent carbon emission management and decision-making.

[0044] Specifically, the carbon emission ratio data is first denoised, the wavelet transform method is used to decompose the data into 5 layers, and the soft threshold method is used to denoise the high-frequency components, and the threshold is set to 2. Then, according to the pre-established standardized analysis process, the denoised carbon emission ratio data is segmented according to 0-20%, 20%-40%, 40%-60%, 60%-80%, and 80%-100%. Then, the statistical characteristic values ​​such as the mean, variance, and slope of the carbon emission ratio in each segment are calculated, and the carbon emission ratio change curve of each segment is fitted using the least squares method to obtain the change law equation of each segment. For example, the carbon emission ratio change equation of the 0-20% segment is y=15x+05, indicating that the carbon emission ratio of this segment shows a slow upward trend, and the carbon emission ratio change equation of the 20%-40% segment is y=-2x+6, indicating that the carbon emission ratio of this segment shows a downward trend, and so on. Finally, the change law of each segment is combined to judge the change trend of the overall carbon emission ratio, providing data support for relevant decisions.

[0045] Step S108, using the random forest model in the machine learning algorithm to train the segmented analysis results to generate a prediction model for the carbon emission ratio.

[0046] Obtain historical data related to carbon emissions, including carbon emissions and energy consumption structure of various industries and regions, and construct a training data set. Preprocess the training data, including data cleaning, outlier processing, missing value filling, etc., to ensure data quality. Use segmented analysis methods to group data according to different dimensions such as industry, region, and energy type to obtain multiple data subsets. For each data subset, use the random forest algorithm to build a carbon emission ratio prediction model, and optimize the model performance by setting appropriate parameters such as the number of decision trees and feature sampling ratio. Integrate the random forest models trained on each data subset to obtain the final carbon emission ratio prediction model. Use the holdout method and other methods to evaluate the prediction model, calculate indicators such as accuracy, recall, and F1 value, and determine whether the model performance meets the requirements. Apply the trained carbon emission ratio prediction model to new input data, determine the corresponding model according to attributes such as industry and region, and output the predicted carbon emission ratio results.

[0047] Specifically, based on the carbon emission data of the time series, the ARIMA model is used for sliding window analysis. Each window size is set to 30 days, with a step size of 1 day. The data in the window is tested for stationarity and the optimal ARIMA (p, d, q) parameters are selected. The model coefficients are obtained by maximum likelihood estimation, and the model fitting effect is evaluated using the AIC criterion. For each segmented ARIMA model, its coefficients, residuals and other features are extracted and used as input features of the random forest model together with the average carbon emissions of the segment. 500 historical data are collected as training sets, and each data contains 20-dimensional ARIMA model features. The random forest model uses 100 decision trees and constructs subsamples through bootstrap sampling with replacement. When splitting a node, 5 features are randomly selected from the 20 features, and their Gini indexes are compared to select the optimal splitting features and thresholds. Each decision tree grows to a maximum depth of 10 and is trained with average carbon emissions as the target variable. For new time series data, segmented ARIMA modeling is also performed, and features are extracted and input into the random forest model. Each decision tree independently predicts carbon emissions, and finally the average of the decision tree prediction results is taken as the final carbon emission ratio prediction value. After 10-fold cross-validation, the prediction error of the model is controlled within 5%, which can accurately predict the trend of carbon emission intensity changes in the future.

[0048] Step S109, based on the results of the prediction model and in combination with the parameter setting optimization requirements, dynamically adjust the changing trend of the carbon emission ratio to generate a final carbon emission ratio resilience analysis report.

[0049] Obtain the output results of the prediction model, and determine the initial change trend of the carbon emission ratio based on the results; establish a parameter adjustment rule library for dynamically adjusting the change trend of the carbon emission ratio according to the parameter setting optimization requirements; use a fuzzy control algorithm to integrate the prediction model results with the parameter adjustment rules to dynamically adjust the change trend of the carbon emission ratio; during the change of the carbon emission ratio, monitor the data of various indicators in real time, and use the anomaly detection algorithm to determine whether trend adjustment is needed; if an anomaly is detected, select the corresponding adjustment strategy from the parameter adjustment rule library based on the anomaly, and dynamically optimize the change trend of the carbon emission ratio; after the dynamic adjustment process is completed, comprehensively analyze the changes in the carbon emission ratio at each stage to generate a carbon emission ratio resilience assessment result; summarize the prediction model results, dynamic adjustment process data and resilience assessment results, and output the final carbon emission ratio resilience analysis report.

[0050] Specifically, first, a carbon emission ratio prediction model is established based on historical data. The ARIMA time series model is used to set parameters p=2, d=1, q=2 to predict the carbon emission ratio in the next five years. The prediction results show that the carbon emission ratio will increase at an annual rate of 5% in the next five years. Then, based on the prediction results, combined with the constraint that the carbon emission ratio does not exceed 5%, the dynamic programming algorithm is used to optimize the carbon emission ratio. The objective function of the algorithm is to minimize the total carbon emissions, and the state transition equation is that the difference between the current carbon emission ratio and the carbon emission ratio of the previous year does not exceed 2%. Through the algorithm calculation, the optimal carbon emission ratios for the next five years are 8%, 0%, 2%, 3% and 5% respectively. Finally, a carbon emission ratio resilience analysis report is generated. The report points out that in the next five years, the carbon emission ratio can be controlled within 5%, and it will grow steadily year by year, with strong resilience. At the same time, the report also gives a series of suggestions for improving carbon emission resilience, such as strengthening the research and development of energy-saving and emission reduction technologies, optimizing the industrial structure, and accelerating the substitution of clean energy. Through the above measures, it is expected to further reduce the carbon emission ratio and achieve the goal of carbon neutrality.

[0051] Step S1010, feeding back key information in the carbon emission ratio resilience analysis report to the dynamic weight matrix of interference factors, realizing adaptive updating of the dynamic weight matrix, and improving the accuracy and reliability of the carbon emission ratio prediction model.

[0052] Obtain the carbon emission ratio resilience analysis report and extract the key information in the report, including the main influencing factors and their weights, sensitivity analysis results, etc. According to the extracted key information, determine the type and scope of the interference factors, initialize the corresponding dynamic weight matrix, set the initial value of the weight and the adjustment strategy. Input the carbon emission ratio data into the prediction model to obtain the preliminary prediction results, compare them with the actual carbon emission ratio, and calculate the prediction error. According to the size of the prediction error, determine whether the dynamic weight matrix needs to be updated. If the error exceeds the preset threshold, the matrix adaptive update mechanism is triggered. Through optimization algorithms such as gradient descent, adjust the weight values ​​of each interference factor in the dynamic weight matrix to minimize the prediction error and obtain the updated weight matrix. Apply the updated dynamic weight matrix to the carbon emission ratio prediction model, re-predict, and obtain the optimized prediction results. Evaluate the prediction accuracy and reliability after optimization. If the requirements are met, output the final prediction results. If not, return to the fourth step to continue iterative optimization until the expected effect is achieved.

[0053] Specifically, the key information in the carbon emission ratio resilience analysis report is extracted, mainly including the degree of influence of various interference factors on the carbon emission ratio and the changing trend of the carbon emission ratio under different scenarios. Natural language processing technology is used to perform semantic analysis on the report content, identify key information such as interference factors and degree of influence, and quantify them into numerical form. For example, for every 1 percentage point increase in economic growth, the carbon emission ratio will increase by 5%; for every 1 percentage point decrease in the proportion of coal in the energy structure, the carbon emission ratio will decrease by 3%, etc. Then, these quantified key information are fed back to the dynamic weight matrix of the interference factors, and the weight coefficients of each interference factor are adjusted in real time through iterative optimization algorithms, such as the gradient descent method. In each iteration, the back propagation algorithm is used to calculate the adjustment amount of the weight coefficient based on the error between the predicted value and the actual value of the carbon emission ratio at the previous moment, and the dynamic weight matrix is ​​updated. After multiple iterative optimizations, the dynamic weight matrix can adaptively reflect the real-time impact of various interference factors on the carbon emission ratio, and the accuracy and reliability of the prediction model are significantly improved. For example, when the downward pressure on the economy increases, the dynamic weight matrix will automatically increase the weight coefficient of economic factors and reduce the weight coefficient of other factors, thereby more accurately predicting the changing trend of the carbon emission ratio. At the same time, methods such as sensitivity analysis can also be used to evaluate the impact of different interference factors on the prediction results, further optimize the dynamic weight matrix, and improve the robustness of the prediction model.

[0054] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. For example, the above features are replaced with the technical features with similar functions disclosed in the present application (but not limited to) to form a technical solution.

Claims

1. A carbon emission ratio resilience analysis method for a road construction project, characterized in that: The method comprises: Obtain the historical change trend of carbon emission ratio from the historical data of road construction projects, and build a dynamic change model of carbon emission ratio by combining with real-time monitoring data; In view of the uncertainty and dynamics of external interference factors, time series analysis methods are used to quantitatively model interference factors such as extreme weather conditions, construction process changes, and fluctuations in raw material supply, and generate a dynamic weight matrix of interference factors; By analyzing the interaction between carbon emission ratio and interference factors, the influence degree and mechanism of interference factors on the dynamic change model of carbon emission ratio are determined; According to the dynamic weight matrix and interaction relationship of interference factors, the dynamic change model of carbon emission ratio is corrected to obtain the corrected trend of carbon emission ratio; Extract key fluctuation features from the modified carbon emission ratio change trend, and combine with the preset threshold to determine whether the fluctuation of the carbon emission ratio exceeds the acceptable range; If the carbon emission ratio fluctuation exceeds the preset threshold, an adaptive filtering algorithm is used to reduce the noise of the data to generate the carbon emission ratio data after noise reduction; Based on the de-noised carbon emission ratio data and the pre-established standardized analysis process, the carbon emission ratio is analyzed in sections to determine the changing pattern of the carbon emission ratio in each section; The random forest model in the machine learning algorithm is used to train the segmented analysis results to generate a prediction model for the carbon emission ratio; According to the results of the prediction model and the optimization requirements of parameter settings, the changing trend of the carbon emission ratio is dynamically adjusted to generate the final carbon emission ratio resilience analysis report; The key information in the carbon emission ratio resilience analysis report is fed back to the dynamic weight matrix of interference factors to achieve adaptive updating of the dynamic weight matrix and improve the accuracy and reliability of the carbon emission ratio prediction model.

2. The method according to claim 1, characterized in that The method of obtaining the historical change trend of the carbon emission ratio from the historical data of the road construction project and combining it with the real-time monitoring data to construct a dynamic change model of the carbon emission ratio includes: Obtain historical carbon emission ratio data and real-time carbon emission monitoring data of road construction projects, clean, denoise and normalize the historical data to obtain a standardized carbon emission ratio time series data set; The standardized carbon emission ratio data set is decomposed using a time series decomposition algorithm to obtain three components of the carbon emission ratio: long-term trend, seasonal variation, and random fluctuation. According to the decomposed long-term trend component of the carbon emission ratio, the gray prediction model is used to perform trend extrapolation prediction of the carbon emission ratio to obtain the changing trend of the carbon emission ratio in the future. The formula for calculating the trend prediction value of the carbon emission ratio is as follows: F t =βT t +(1-β)F t-1 In the formula, Ft is the trend forecast value at time t, Tt is the actual trend value at the current time, and β is the smoothing parameter; According to the seasonal variation component of the carbon emission ratio obtained by decomposition, the seasonal autoregressive moving average model SARIMA is used to model the periodic variation and obtain the seasonal fluctuation law and prediction results. Combine trend forecast results and seasonal forecast results to obtain a comprehensive carbon emission ratio forecast benchmark value, and dynamically integrate it with real-time monitored carbon emission data to correct forecast deviations in real time; The optimized carbon emission ratio prediction model was backtested and its parameters were tuned. The accuracy and stability of the model were evaluated through indicators such as root mean square error and mean absolute percentage error. Determine the optimal dynamic prediction model for carbon emission ratio, deploy it in the production environment, obtain carbon emission monitoring data of road construction projects in real time, automatically update the prediction results, and provide decision support for carbon emission control.

3. The method according to claim 1, characterized in that In view of the uncertainty and dynamics of external interference factors, the time series analysis method is used to quantitatively model interference factors such as extreme weather conditions, construction process changes and raw material supply fluctuations, and generate a dynamic weight matrix of interference factors, including: In view of the uncertainty and dynamics of external interference factors, historical data, including extreme weather conditions, construction process change records, and raw material supply fluctuation data, are obtained to construct a time series data set; Use data preprocessing technology to clean and normalize time series data, remove noise and outliers, and improve data quality; Apply time series analysis algorithms to establish a prediction model for interference factors, and capture the dynamic changes of interference factors through learning and training of historical data; According to the output results of the prediction model, the weight matrix of interference factors is dynamically adjusted to quantify the impact of different interference factors on engineering project management; Combine the dynamic weight matrix with various aspects of project management, such as schedule management, cost management and quality management, to formulate targeted response strategies and adjustment measures; During the project implementation, we continuously collect and update real-time data of external interference factors, conduct online learning and optimization of the prediction model, and continuously improve the accuracy and adaptability of the prediction; Regularly evaluate the implementation effect of the solution, and by comparing the actual project management performance with the expected goals, timely adjust and improve the plan to improve the robustness and adaptability of engineering project management.

4. The method according to claim 1, characterized in that The method analyzes the interaction between the carbon emission ratio and the interference factors to determine the influence degree and mechanism of the interference factors on the dynamic change model of the carbon emission ratio, including: Obtain the actual data of carbon emission ratio and statistical data of various influencing factors in the historical period, perform data preprocessing on carbon emission ratio and various influencing factors respectively, and obtain standardized time series data; The correlation analysis method is used to calculate the correlation coefficient matrix between the carbon emission ratio and various influencing factors, and the key influencing factors with a high correlation with the carbon emission ratio are screened out; Based on the time series data of carbon emission ratio and key influencing factors, a VAR vector autoregression model is constructed. The impact degree and action path of key influencing factors on the dynamic changes of carbon emission ratio are analyzed through impulse response function and variance decomposition method. For the residual sequence of VAR model fitting, Granger causality test is used to further determine whether there is a causal relationship between the carbon emission ratio and key influencing factors, and to determine the direction of action between each factor and the carbon emission ratio; Using machine learning algorithms, a nonlinear regression prediction model of carbon emission ratio and key influencing factors was established. Through feature importance analysis, the weight of each influencing factor on the carbon emission ratio was quantitatively evaluated. By integrating the analysis results of the VAR model and the machine learning model, the mechanism of action between key influencing factors and the dynamic changes of carbon emission ratios is revealed from the perspectives of statistical association and causal mechanism, and a quantitative relationship model between carbon emission ratios and key factors is constructed; Based on the established carbon emission ratio change model, the impact of changes in key influencing factors on future carbon emission ratios under different scenarios is predicted, providing a quantitative evaluation basis and policy simulation tools for formulating carbon emission reduction measures.

5. The method according to claim 1, characterized in that The dynamic change model of the carbon emission ratio is corrected according to the dynamic weight matrix and interaction relationship of the interference factors to obtain the corrected trend of the carbon emission ratio, including: Based on the historical data of carbon emission related factors, a dynamic weight matrix of carbon emission factors is constructed, and the weight coefficients are obtained through machine learning algorithm training; Obtain the interaction relationship data between carbon emission factors, use association rule mining algorithm to establish the interaction relationship model of carbon emission factors; The dynamic weight matrix is ​​integrated with the interaction relationship model to obtain a dynamic change model of carbon emissions that comprehensively considers weights and interaction effects; In view of the original dynamic change model of carbon emission ratio, the correction term in the comprehensive dynamic change model is introduced to optimize and correct the carbon emission ratio prediction model; The random forest regression algorithm is used as the basis for the modified carbon emission ratio dynamic change model, and the optimized carbon emission ratio prediction model is obtained through training with a large amount of historical data; Using the optimized carbon emission ratio prediction model, combined with the prediction data of carbon emission related factors under different future scenarios, simulate and predict the dynamic change trend of carbon emission ratio within different time ranges in the future; According to the predicted trend of changes in the carbon emission ratio, identify the high-risk abnormal range of the carbon emission ratio, formulate targeted carbon emission reduction optimization and regulatory measures, and promote the realization of carbon peak and carbon neutrality goals.

6. The method according to claim 1, characterized in that The extracting of key fluctuation features from the modified carbon emission ratio change trend and combining with a preset threshold to determine whether the fluctuation of the carbon emission ratio exceeds an acceptable range include: Obtaining the corrected carbon emission ratio change trend data, analyzing and processing the data to obtain at least one key fluctuation feature; Obtaining a preset acceptable fluctuation threshold of the carbon emission ratio, and comparing the key fluctuation feature with the threshold; If the key fluctuation feature exceeds the threshold, it is determined that the carbon emission ratio fluctuation exceeds the acceptable range, and a warning message is output; For carbon emission ratio fluctuations that exceed an acceptable range, obtain corresponding carbon emission source data, analyze the source data, and determine the main emission sources causing the fluctuations; According to the main emission sources obtained through analysis, the corresponding historical emission data are obtained, and the trend of carbon emission ratio changes of the emission sources in the future is predicted through time series analysis algorithm; Compare the predicted future carbon emission ratio change trend with the preset threshold again to determine whether the carbon emission ratio of the emission source still has the risk of exceeding the acceptable range; If the judgment result is that there is an excess risk, an early warning prompt will be output and corresponding carbon emission control strategies and suggestions will be generated to guide relevant departments to reduce carbon emissions.

7. The method according to claim 1, characterized in that If the carbon emission ratio fluctuation exceeds the preset threshold, an adaptive filtering algorithm is used to perform noise reduction processing on the data to generate noise-reduced carbon emission ratio data, including: According to the preset carbon emission ratio fluctuation threshold, the carbon emission ratio data is obtained in real time to determine whether the carbon emission ratio fluctuation exceeds the threshold; If the carbon emission ratio fluctuation exceeds the preset threshold, the Kalman filter algorithm is used to perform adaptive noise reduction processing on the acquired carbon emission ratio data; In the process of Kalman filter algorithm processing, the state space model of carbon emission ratio data is established and the filter parameters are adaptively estimated to achieve optimized noise reduction of carbon emission ratio data; After the noise reduction process, the noise-reduced and optimized carbon emission ratio data is obtained, and the data is used as the final calculation result of the carbon emission ratio; According to the carbon emission ratio data after noise reduction optimization, determine whether the carbon emission ratio is within a reasonable range. If it exceeds the reasonable range, trigger the early warning mechanism; By conducting statistical analysis on carbon emission ratio data for multiple consecutive time periods, we can determine the long-term trend of carbon emission ratios and provide decision-making references for carbon emission control; The carbon emission ratio data is correlated with the production process parameters, and the principal component analysis method is used to determine the correlation between each parameter and the carbon emission ratio, find out the key influencing factors, optimize the production process, and reduce carbon emissions.

8. The method according to claim 1, characterized in that The carbon emission ratio data after noise reduction is combined with a pre-established standardized analysis process to perform segmented analysis on the carbon emission ratio to determine the change pattern of the carbon emission ratio in each segment, including: Obtaining raw data of carbon emission ratios, and preprocessing the raw data, including denoising and outlier processing, to obtain preprocessed carbon emission ratio data; According to the pre-established standardized analysis process, the pre-processed carbon emission ratio data is standardized to eliminate the impact of different dimensions and orders of magnitude; Using a segmented clustering algorithm, such as the K-Means algorithm, the standardized carbon emission ratio data is divided into several segments, and the data characteristics within each segment are similar; For each segment, analyze the distribution characteristics of the carbon emission ratio data within it, including mean, variance and skewness, to determine the overall change trend of the data within the segment; Compare the mean values ​​of carbon emission ratios of adjacent segments. If the difference between the mean values ​​of adjacent segments exceeds a preset threshold, it is determined that a significant change has occurred in the carbon emission ratio between adjacent segments, and the direction and magnitude of the change are recorded. By integrating the changing trends of carbon emission ratios within each segment and the changes between adjacent segments, a quantitative model of carbon emission ratio changes is established using time series analysis methods, such as the ARIMA model, to reveal the overall changing pattern of carbon emission ratios; The changing trends of the carbon emission ratios in each segment and the overall changing rules are compiled into an analysis report to provide data support for subsequent carbon emission management and decision-making.

9. The method according to claim 1, characterized in that: The random forest model in the machine learning algorithm is used to train the segmented analysis results to generate a prediction model for the carbon emission ratio, including: Obtain historical data related to carbon emissions, including carbon emissions and energy consumption structure of various industries and regions, and build a training data set; Preprocess the training data, including data cleaning, outlier processing, and missing value filling, to ensure data quality; Using segmented analysis methods, the data is grouped according to different industries, regions and energy types to obtain multiple data subsets; For each data subset, a carbon emission ratio prediction model was constructed using the random forest algorithm, and the model performance was optimized by setting appropriate parameters; The random forest models trained on each data subset are integrated to obtain the final carbon emission ratio prediction model; Use the holdout method to evaluate the prediction model, calculate the accuracy, recall, and F1 value, and determine whether the model performance meets the requirements; Apply the trained carbon emission ratio prediction model to new input data, determine the corresponding model according to the industry and region, and output the predicted carbon emission ratio results.

10. The method according to claim 1, characterized in that According to the results of the prediction model, combined with the parameter setting optimization requirements, the changing trend of the carbon emission ratio is dynamically adjusted to generate the final carbon emission ratio resilience analysis report, including: Obtain the output results of the prediction model and determine the initial change trend of the carbon emission ratio based on the results; In response to the needs of parameter setting optimization, a parameter adjustment rule library is established to dynamically adjust the trend of carbon emission ratio changes; Using fuzzy control algorithm, the prediction model results are integrated with parameter adjustment rules to dynamically adjust the trend of carbon emission ratio; During the change of carbon emission ratio, various indicator data are monitored in real time, and the anomaly detection algorithm is used to determine whether trend adjustment is needed; If an anomaly is detected, the corresponding adjustment strategy is selected from the parameter adjustment rule library according to the anomaly to dynamically optimize the trend of carbon emission ratio changes; After the dynamic adjustment process is completed, the carbon emission ratio changes in each stage are integrated to generate the carbon emission ratio resilience assessment results; The prediction model results, dynamic adjustment process data and resilience assessment results are summarized to output the final carbon emission ratio resilience analysis report.

Images