Road preventive maintenance intelligent decision-making method considering carbon emission influence

Abstract

The invention belongs to the technical field of road maintenance, and discloses an intelligent decision-making method for road preventive maintenance considering the influence of carbon emission, and the method comprises the steps: firstly building a maintenance measure library; then, real-time cost and transportation data are obtained, and the cost and carbon emission of each measure in the measure library are calculated and updated; thirdly, collecting historical maintenance work data, including road surface condition indexes and road section feature data, for training a road surface condition prediction model; utilizing the model to predict a future pavement condition index according to the current condition and the characteristic data of the road section; and when the prediction index reaches a preset threshold value, triggering a maintenance scheme generation process, and selecting an optimal measure combination from the maintenance measure library through an optimization algorithm. According to the method, resource waste caused by excessive maintenance is avoided, and structural damage and functional degradation of the pavement caused by insufficient maintenance are also prevented, so that efficient utilization of maintenance resources and minimization of environmental influence are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of highway maintenance. More specifically, the present invention relates to an intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions. Background Art

[0002] With the rapid development of transportation infrastructure and the increasing vehicle usage, the importance of highway maintenance has become increasingly prominent. Effective highway maintenance can not only extend the service life of the road, but also improve road safety, reduce vehicle operation costs, and lower environmental impacts. However, there are still many deficiencies in the existing highway preventive maintenance methods in achieving scientific, efficient, and environmentally friendly maintenance decisions.

[0003] First of all, traditional highway maintenance methods often adopt a mode of fixed time and fixed measures for maintenance. This static maintenance strategy cannot accurately reflect the dynamic changes in the actual condition of the road, and is prone to problems of over-maintenance or under-maintenance. Over-maintenance not only causes waste of resources, but also increases unnecessary traffic interruptions; while under-maintenance may lead to structural damage and serious functional degradation of the road, ultimately requiring higher repair costs. Secondly, the existing maintenance decision-making methods have deficiencies in considering environmental factors. With the increasing global attention to carbon emissions, as an important engineering activity, the carbon emission impact of highway maintenance cannot be ignored. However, most maintenance decision-making systems only focus on economic costs and fail to take carbon emissions as an important decision-making factor, which is not conducive to achieving sustainable maintenance strategies. Summary of the Invention

[0004] In order to overcome the problems of over-maintenance or under-maintenance easily caused by the existing technology, the present invention proposes an intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions to solve the above problems.

[0005] The present invention provides the following technical solutions: An intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions, comprising: Establish a maintenance measure library, which stores a number of maintenance measures, and the maintenance measures include measure numbers, measure details, and change amounts of condition indexes; Obtain real-time cost data, calculate the measure cost according to the real-time cost data and measure details, obtain real-time transportation data, and calculate the measure carbon emissions according to the real-time transportation data; add the calculated measure cost and measure carbon emissions to the corresponding maintenance measures in the maintenance measure library; Obtain a number of historical maintenance work data to form a historical maintenance work data set, and the historical maintenance work data includes the pavement condition indexes at two time intervals of the highway section and the characteristic data of the highway section; Train a pavement condition prediction model using a historical maintenance work dataset, where the pavement condition prediction model is used to predict the pavement condition index after a future time based on the current pavement condition index of a highway section and the highway section feature data; Obtain the current pavement condition index of a highway section and the feature data of the highway section. According to the current pavement condition index of the highway section and the corresponding feature data of the highway section, use the pre-trained pavement condition prediction model to predict the pavement condition index after a future time and record it as the predicted index; Monitor the predicted index, and when the predicted index is less than or equal to a preset index threshold, generate a maintenance plan; the generation of the maintenance plan includes selecting several maintenance measures from a maintenance measure library through an optimization method according to the predicted index and the preset index threshold to form a maintenance plan.

[0006] Preferably, the measure details include raw material details and construction details. The raw material details include the type of raw material and the corresponding usage amount, and the construction details include the man-hours, equipment type, and the corresponding operation time.

[0007] Preferably, the real-time cost data includes the unit price of each raw material, the unit price of each type of equipment, and the unit price of labor. The calculation formula for the measure cost is as follows: , In the formula, represents the measure cost, represents the number of raw material types, represents the th usage amount corresponding to the raw material, represents the th unit price of the raw material, represents the number of equipment types, represents the th operation time corresponding to the equipment type, represents the th unit price of the equipment type, represents the man-hours, represents the unit price of labor; The real-time transportation data includes the transportation distance of each raw material and the unit transportation carbon emission coefficient corresponding to the transportation method. The calculation formula for the measure carbon emission is as follows: , In the formula, represents the measure carbon emission, represents the original material type quantity, represents the th carbon emission coefficient of the raw material, represents the Carbon emission coefficient of the device represents the transportation distance of the th raw material, and represents the unit transportation carbon emission coefficient corresponding to the transportation mode of the

[0008] Preferably, the characteristic data of the highway section is obtained by performing characteristic processing on the corresponding pavement structure data, traffic volume data, and climate data of the highway section.

[0009] Preferably, the pavement structure data includes pavement thickness and pavement material type, the traffic volume data includes daily average traffic flow and heavy vehicle proportion, and the climate data includes monthly average temperature, monthly average humidity, and monthly precipitation; The characteristic processing includes: normalizing the pavement thickness data and converting the pavement material type into a numerical code; normalizing the daily average traffic flow data and converting the heavy vehicle proportion data into a decimal form; normalizing the monthly average temperature, monthly average humidity, and monthly precipitation data respectively; and combining the processed data items into a feature vector as the characteristic data corresponding to the highway section.

[0010] Preferably, training the pavement condition prediction model using the historical maintenance work dataset includes: Obtaining the historical maintenance work dataset and dividing the historical maintenance work dataset into a training set and a test set; Building a machine learning framework and defining an input layer and an output layer, setting a loss function and an optimizer, and constructing a machine learning model; Using the characteristic data of the highway section and the earlier one of the two pavement condition indices as the input layer data, and the later one of the pavement condition indices as the output layer data, and training the constructed machine learning model using the training set according to the set batch size and number of training epochs; Testing the trained machine learning model using the test set, and outputting the machine learning model that meets the preset accuracy as the pavement condition prediction model.

[0011] Preferably, selecting several maintenance measures from the maintenance measure library through an optimization method to form a maintenance plan includes: Setting the objective function , and the formula of the objective function is as follows:

[0012] In the formula, is the total maintenance cost, is the total carbon emission, is the reference cost, is the reference carbon emission. and are weight coefficients, and ; Set constraint conditions, which include: the sum of the prediction index and the change in the total condition index is not less than the index threshold, and the total maintenance cost does not exceed the budget ceiling; the total carbon emissions do not exceed the carbon emission ceiling; Generate an optimization problem according to the set objective function and constraint conditions and solve the optimization problem; the solution of the optimization problem includes: initializing a candidate solution set, calculating the objective function value and the satisfaction of the constraint conditions for each candidate solution; an iterative optimization process, continuously updating the candidate solution set until a preset termination condition is reached; selecting the solution that satisfies the constraint conditions and has the minimum objective function value from the final candidate solution set; and outputting the corresponding maintenance measure combination as the final maintenance plan.

[0013] Preferably, the initialization of the candidate solution set includes: Use the index threshold minus the prediction index to obtain the minimum change in the condition index; Sort all measures in the maintenance measure library from largest to smallest according to the change in the condition index, and initialize an empty candidate solution set; Set a counter , for from 1 to , where is a preset value greater than 1; Use a recursive method to select all possible combinations of measures from the sorted measure library For each combination, check whether the condition is satisfied: the total change in the condition index of measures is greater than the minimum change in the condition index, and the sum of the changes in the condition index of any measures is less than the minimum change in the condition index; add the combinations that meet the conditions to the candidate solution set. When the number of combinations in the candidate solution set reaches the preset value, stop the recursion to obtain the initialized candidate solution set.

[0014] The present invention provides an intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions, which has the following beneficial effects: By establishing a pavement condition prediction model, this method can dynamically predict future pavement conditions based on the current conditions and characteristic data of highway sections. This predictive ability enables maintenance decision-making to shift from passive response to proactive prevention, effectively avoiding the problems brought about by the traditional fixed-time and fixed-measurement mode. The system continuously monitors the prediction index and triggers the generation of maintenance plans in a timely manner when the pavement condition reaches the preset threshold, which not only avoids the waste of resources caused by over-maintenance but also prevents structural damage and serious functional degradation of the road due to insufficient maintenance, thus optimizing the utilization efficiency of maintenance resources. At the same time, both economic costs and carbon emissions are considered in the maintenance decision-making process, and optimization is carried out by setting an objective function that includes these two factors. This method overcomes the limitation of traditional maintenance decision-making systems that only focus on economic costs and incorporates environmental factors into the decision-making process. By adjusting the weight coefficients in the objective function, decision-makers can flexibly balance economic benefits and environmental protection according to actual needs, thus formulating more sustainable maintenance strategies. Brief Description of the Drawings

[0015] Figure 1 It is a schematic flow chart of an intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions according to the present invention. Detailed Embodiments

[0016] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0017] Embodiment 1 Please refer to Figure 1 , in this embodiment, an intelligent decision-making method for highway preventive maintenance considering the impact of carbon emissions includes: S1. Establish a maintenance measure library, which stores a number of maintenance measures, and the maintenance measures include measure numbers, measure details, and changes in condition indices; The measure details include raw material details and construction details. The raw material details include types of raw materials and corresponding usage amounts, and the construction details include man-hours, types of equipment, and corresponding operation times.

[0018] In this embodiment, the maintenance measure library is used to store the relevant information of various highway preventive maintenance measures. Its establishment process includes: First, assign a unique measure number to each measure to distinguish different maintenance measures. Second, record the details of each measure in detail. For example, for the pothole repair measure, record the details of the raw materials required for the measure, including the types of raw materials such as asphalt mixture, tack coat, and stone chips and the corresponding usage amounts. In addition, record the construction details, including the man-hours, types of equipment such as rollers, and the corresponding operation times. Finally, determine the change amount of the condition index corresponding to each maintenance measure, which is used to quantify the improvement degree of the measure on the road surface condition. The change amount of the condition index represents the expected increase value of the road surface condition index after implementing the corresponding maintenance measure. For example, if the change amount of the condition index of a certain maintenance measure is 5, it means that after implementing this measure, the road surface condition index is expected to increase by 5. This value can be obtained through historical data analysis or expert evaluation. It should be noted that the road surface condition index is a comprehensive index used to evaluate the overall condition of the highway pavement, usually using a scoring system from 0 to 100, where 100 represents the best state of the road surface and 0 represents that the road surface is completely damaged. Then store all this information in the database to obtain the maintenance measure library.

[0019] S2. Obtain real-time cost data, calculate the measure cost according to the real-time cost data and measure details, obtain real-time transportation data, and calculate the measure carbon emissions according to the real-time transportation data; add the calculated measure cost and measure carbon emissions to the corresponding maintenance measure in the maintenance measure library; The real-time cost data includes the unit price of each raw material, the unit price of each type of equipment, and the unit price of labor. The calculation formula for the measure cost is as follows: , In the formula, represents the measure cost, represents the number of raw material types, represents the usage amount corresponding to the th raw material, represents the unit price of the th raw material, represents the number of equipment types, represents the operation time corresponding to the th type of equipment, represents the unit price of the th type of equipment, represents the man-hours, represents the unit price of labor; , In the formula, represents the carbon emissions of the measures, and represents the original quantity of raw material types, represents the carbon emission coefficient of the th raw material, represents the carbon emission coefficient of the th type of equipment, represents the transportation distance of the th raw material, represents the unit transportation carbon emission coefficient corresponding to the transportation mode of the th raw material.

[0020] In this embodiment, first, the latest raw material unit price, equipment unit price, and labor unit price can be obtained through real-time data interfaces with material suppliers, equipment leasing companies, and human resources departments. These data may be updated daily or weekly according to market conditions. Then, combined with the measure details stored in the maintenance measure library, the cost of each maintenance measure is calculated. For example, for the pothole repair measure, it may be necessary to calculate the material costs of asphalt mixture, tack coat, and stone chips, the usage costs of equipment such as rollers, and the labor costs of repair workers. At the same time, through the data interface with the logistics company, the real-time transportation distance data of various raw materials is obtained. Combined with the pre-set unit transportation carbon emission coefficients of different transportation modes (such as trucks, trains, etc.), the carbon emissions of each maintenance measure are calculated. This calculation process takes into account the carbon emissions of the raw materials themselves, the carbon emissions generated by equipment use, and the carbon emissions during the transportation of raw materials. It should be noted that the carbon emission coefficients of raw materials and equipment are fixed values determined in advance according to the life cycle assessment method. Finally, the calculated measure costs and carbon emissions are updated to the corresponding maintenance measure records in the maintenance measure library for use in subsequent decision-making processes with the latest data.

[0021] S3. Obtain a number of historical maintenance work data to form a historical maintenance work data set, where the historical maintenance work data includes the pavement condition indices at two time intervals of the highway section and the characteristic data of the highway section; The characteristic data of the highway section is obtained by performing characteristic processing on the corresponding pavement structure data, traffic volume data, and climate data of the highway section The pavement structure data includes pavement thickness and pavement material type, the traffic volume data includes average daily traffic flow and heavy vehicle ratio, and the climate data includes monthly average temperature, monthly average humidity, and monthly precipitation; The feature extraction process includes: normalizing the road surface thickness data, converting the road surface material type into a numerical code; normalizing the daily average traffic volume data, converting the heavy vehicle proportion data into decimal form; normalizing the monthly average temperature, monthly average humidity, and monthly precipitation data respectively; and combining the processed data items into a feature vector as the feature data corresponding to the highway section.

[0022] In this embodiment, the process of obtaining and processing historical maintenance work data can be as follows: First, extract historical maintenance records from the database of the highway management department. Each record contains the pavement condition index of two time points of the highway section (such as the pavement condition index of two time points six months apart), and the relevant feature data of the highway section. The feature data includes pavement structure data, traffic volume data, and climate data. Specifically, the pavement structure data includes the road surface thickness (such as 20 cm) and the road surface material type (such as asphalt concrete); the traffic volume data includes the daily average traffic volume (such as 10,000 vehicles per day) and the heavy vehicle proportion (such as 15%); the climate data includes the monthly average temperature (such as 25 °C), the monthly average humidity (such as 60%), and the monthly precipitation (such as 100 mm). Next, perform feature extraction on these raw data. The road surface thickness data is converted to the range of 0 - 1 through the min-max normalization method; the road surface material type is converted into a numerical code, such as asphalt concrete is coded as 1, cement concrete is coded as 2, etc.; the daily average traffic volume is also normalized; the heavy vehicle proportion is converted into decimal form (such as 0.15); the monthly average temperature, humidity, and precipitation are also normalized respectively. It should be noted that the normalization process helps to eliminate the dimensional differences between different features, enabling subsequent machine learning models to better process these features. Finally, combine the processed data items in a fixed order into a vector as the feature data of the highway section. For example, a feature vector such as [0.8, 1, 0.6, 0.15, 0.7, 0.5, 0.3] may be formed, where each value corresponds to the processed road surface thickness, material type, traffic volume, heavy vehicle proportion, temperature, humidity, and precipitation respectively. This processing method enables the data of different highway sections to have the same feature representation, facilitating subsequent machine learning model training and prediction.

[0023] S4. Use the historical maintenance work data set to train a pavement condition prediction model, where the pavement condition prediction model is used to predict the pavement condition index after a future time based on the current pavement condition index of the highway section and the feature data of the highway section; The process of using the historical maintenance work data set to train the pavement condition prediction model includes: Obtain the historical maintenance work data set, and divide the historical maintenance work data set into a training set and a test set; Build a machine learning framework, define the input layer and output layer, set the loss function and optimizer, and construct a machine learning model; Use the feature data of the road section and the earlier pavement condition index among the two pavement condition indexes as the input layer data, and the later pavement condition index as the output layer data. Use the training set to train the constructed machine learning model according to the set batch size and number of training epochs; Use the test set to test the trained machine learning model, and output the machine learning model that meets the preset accuracy as the pavement condition prediction model.

[0024] In this embodiment, the training process of the pavement condition prediction model can be as follows: First, randomly divide the obtained historical maintenance work data set into a training set and a test set according to a ratio of 8:2. Then, use a deep learning framework (such as TensorFlow or PyTorch) to build a neural network model. The input layer of the model includes the feature data of the road section and the current pavement condition index, and the output layer is the predicted future pavement condition index. In the model structure, a multi-layer fully connected neural network or a long short-term memory network (LSTM) and other structures suitable for time series prediction may be used. The loss function is selected as the mean squared error (MSE), and the Adam algorithm is used as the optimizer. During the training process, use the feature data of the road section in the training set and the pavement condition index at an earlier time point as the input, and the pavement condition index at a later time point as the target output. Set an appropriate batch size (such as 32 or 64) and number of training epochs (such as 100 or 200 epochs), and start model training. After training is completed, use the test set to evaluate the model performance. If the prediction accuracy of the model on the test set reaches the preset threshold (such as the mean absolute error is less than 5), it is considered that the model meets the requirements and can be used as the final pavement condition prediction model. If the requirements are not met, it is necessary to optimize by adjusting the model structure, hyperparameters or increasing training data until the model performance meets the requirements.

[0025] S5. Obtain the current pavement condition index of the road section and the feature data of the road section. According to the current pavement condition index of the road section and the corresponding feature data of the road section, use the pre-trained pavement condition prediction model to predict the pavement condition index after a future time and record it as the prediction index; S6. Monitor the prediction index, and when the prediction index is less than or equal to the preset index threshold, generate a maintenance plan; the generation of the maintenance plan includes selecting several maintenance measures from the maintenance measure library through an optimization method according to the prediction index and the preset index threshold to form a maintenance plan.

[0026] The selection of several maintenance measures from the maintenance measure library through an optimization method to form a maintenance plan includes: Set the objective function , the objective function has the following formula:

[0027] In the formula, is the total maintenance cost, is the total carbon emissions, is the reference cost, is the reference carbon emissions, and are weight coefficients, and ; Set the constraint conditions, which include: the prediction index plus the change in the total condition index is not less than the index threshold, and the total maintenance cost does not exceed the budget ceiling; the total carbon emissions do not exceed the carbon emission ceiling; Generate an optimization problem based on the set objective function and constraint conditions and solve the optimization problem; solving the optimization problem includes: initializing a candidate solution set, calculating the objective function value and the satisfaction of the constraint conditions for each candidate solution; an iterative optimization process, continuously updating the candidate solution set until a preset termination condition is reached; selecting the solution that satisfies the constraint conditions and has the minimum objective function value from the final candidate solution set; and outputting the corresponding maintenance measure combination as the final maintenance plan.

[0028] The initialization of the candidate solution set includes: Use the index threshold minus the prediction index to obtain the minimum change in the condition index; Sort all the measures in the maintenance measure library in descending order of the change in the condition index, and initialize an empty candidate solution set; Set a counter , for from 1 to , where is a preset value greater than 1; Use the recursive method to select all possible combinations of measures from the sorted measure library For each combination, check whether it satisfies the condition: the total change in the condition index of the measures is greater than the minimum change in the condition index, and the sum of the changes in the condition index of any measures is less than the minimum change in the condition index; add the combinations that meet the conditions to the candidate solution set.

[0029] In this embodiment, the process of road condition prediction and maintenance plan generation is as follows: First, the current road condition index and characteristic data of a specific highway section are obtained through on-site inspection or sensor data. These data are input into a pre-trained road condition prediction model to obtain the predicted road condition index at a future point in time (half a year in this example). The system continuously monitors this predicted index. When it drops below a preset index threshold (such as 70 points), the maintenance plan generation process is triggered. The generation of the maintenance plan adopts a multi-objective optimization method, considering two objectives: cost and carbon emissions. The reference cost and reference carbon emissions in the objective function can be set based on historical data or budgets, and the weight coefficients are determined according to the preferences of the decision-makers. For example, with a slightly greater emphasis on cost, they can be set to 0.6 and 0.4 respectively. The constraint conditions ensure that the generated maintenance plan can improve the road condition to an acceptable level while not exceeding the budget and carbon emission limits. During the optimization process, first, the minimum required change in the condition index is calculated. For example, if the predicted index is 65 and the index threshold is 80, the minimum change is 15. Then, the measures in the maintenance measure library are sorted in descending order of the change in the condition index, and different combinations of measures are generated using a recursive method. For example, combinations such as crack sealing and pothole repair for asphalt pavements, and pavement milling and resurfacing may be generated. For each combination, its total cost, total carbon emissions, and total change in the condition index are calculated to check whether the constraint conditions are met. Combinations that meet the conditions are added to the candidate solution set. The candidate solution set is continuously iteratively updated through an optimization algorithm (such as a genetic algorithm or particle swarm optimization) until a preset number of iterations is reached or no better solution can be found for several consecutive times. Finally, the solution with the minimum objective function value is selected from the candidate solution set as the final maintenance plan. It should be noted that this method can achieve a balance between cost and environmental impact, generate an economical and environmentally friendly maintenance plan, and contribute to the sustainable development of highway maintenance.

[0030] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only one way, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0031] As described above, the above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

[0032] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions, characterized in that: include: Establishing a maintenance measure library, wherein the maintenance measure library stores a plurality of maintenance measures, wherein the maintenance measures include measure numbers, measure details, and condition index changes; Obtain real-time cost data, calculate the measure cost based on the real-time cost data and measure details, obtain real-time transportation data, calculate the measure carbon emissions based on the real-time transportation data; add the calculated measure cost and measure carbon emissions to the corresponding maintenance measure in the maintenance measure library; Acquire a number of historical maintenance work data to form a historical maintenance work data set, wherein the historical maintenance work data includes a highway section interval Two road condition indices and characteristic data of highway sections over time; The road condition prediction model is trained using the historical maintenance work dataset. The road condition prediction model is used to predict the future road condition index based on the current road condition index of the highway segment and the highway segment characteristic data. pavement condition index after time; Obtain the current road condition index of the highway section and the characteristic data of the highway section, and use the pre-trained road condition prediction model to predict the future road condition index and the characteristic data corresponding to the highway section. The road condition index after a certain time is recorded as the predicted index; Monitor the prediction index and generate a maintenance plan when the prediction index is less than or equal to a preset index threshold; the maintenance plan generation includes selecting a number of maintenance measures from a maintenance measure library through an optimization method according to the prediction index and the preset index threshold to form a maintenance plan.

2. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 1 is characterized in that: The measure details include raw material details and construction details. The raw material details include raw material types and corresponding usage amounts. The construction details include man-hours, equipment types and corresponding operation time.

3. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 2 is characterized in that: The real-time cost data includes the unit price of each raw material, the unit price of each type of equipment and the unit price of labor. The calculation formula of the measure cost is as follows: ， In the formula, represents the cost of the measure, Indicates the number of raw material types, Indicates The usage of the corresponding raw materials, Indicates The unit price of the raw materials, Indicates the number of device types. Indicates The operation time corresponding to the equipment of this type, Indicates Unit price of the equipment, Indicates man-hours. Indicates the labor unit price; The real-time transportation data includes the transportation distance of each raw material and the unit transportation carbon emission coefficient corresponding to the transportation mode. The calculation formula of the carbon emission of the measure is as follows: ， In the formula, Indicates the carbon emissions of the measure, Number of material types, Indicates Carbon emission coefficient of raw materials, Indicates Carbon emission coefficient of this type of equipment, Indicates The transportation distance of raw materials, Indicates The unit transportation carbon emission coefficient corresponding to the transportation mode of the raw materials.

4. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 1 is characterized in that: The characteristic data of the highway section is obtained by characterizing the corresponding pavement structure data, traffic volume data and climate data of the highway section.

5. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 4 is characterized in that: The pavement structure data includes pavement thickness and pavement material type, the traffic volume data includes daily average vehicle flow and heavy vehicle ratio, and the climate data includes monthly average temperature, monthly average humidity and monthly precipitation; The characterization processing includes: normalizing the road thickness data and converting the road material type into a numerical code; normalizing the daily average vehicle flow data and converting the heavy vehicle ratio data into a decimal form; normalizing the monthly average temperature, monthly average humidity and monthly precipitation data respectively; and merging the processed data into a feature vector as the feature data corresponding to the highway section.

6. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 5 is characterized in that: The method of using the historical maintenance work data set to train the road condition prediction model includes: Obtain a historical maintenance work data set, and divide the historical maintenance work data set into a training set and a test set; Build a machine learning framework and define the input layer and output layer, set the loss function and optimizer, and build a machine learning model; The characteristic data of the highway section and the road condition index with the earlier time as the input layer data, and the later time as the output layer data, are used to train the constructed machine learning model using the training set according to the set batch size and number of training rounds; The trained machine learning model is tested using the test set, and a machine learning model that meets the preset accuracy is output as a road condition prediction model.

7. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 1 is characterized in that: The method of selecting a plurality of maintenance measures from the maintenance measure library through the optimization method to form a maintenance plan includes: Setting the objective function , the objective function The formula is as follows: In the formula, is the total maintenance cost, is the total carbon emissions, For reference cost, For reference carbon emissions, and is the weight coefficient, and ; Setting constraints, wherein the constraints include: the predicted index plus the total condition index change is not less than the index threshold, the total maintenance cost does not exceed the budget upper limit; the total carbon emissions do not exceed the carbon emission upper limit; Generate an optimization problem and solve it according to the set objective function and constraints; solving the optimization problem includes: initializing a set of candidate solutions, calculating the objective function value and constraint satisfaction for each candidate solution; iterating the optimization process, continuously updating the candidate solution set until the preset termination condition is reached; selecting a solution that satisfies the constraints and has the smallest objective function value from the final set of candidate solutions; and outputting the corresponding maintenance measure combination as the final maintenance plan.

8. The intelligent decision-making method for preventive highway maintenance considering the impact of carbon emissions according to claim 7 is characterized in that: The initialization candidate solution set includes: Use the index threshold to subtract the predicted index to obtain the minimum condition index change; Sort all measures in the maintenance measure library from large to small according to the change in condition index, and initialize an empty candidate solution set; Set a counter ,for From 1 to ,in is a preset value greater than 1; Using a recursive approach, select from the sorted measure library All possible combinations of measures For each combination, check whether the condition is met: The total condition index change of each measure is greater than the minimum condition index change. The sum of the changes in the condition index of the measures is less than the minimum change in the condition index; the combination that meets the conditions is added to the candidate solution set; When the number of combinations in the candidate solution set reaches a preset value, the recursion is stopped and an initialized candidate solution set is obtained.

Images