Intelligent Analysis and Decision-making Method and Device for Road Damage Based on Driving Data

Abstract

The present invention discloses an intelligent analysis and decision-making method and device for road damage based on driving data, which relates to the technical field of road damage analysis, and includes: collecting driving data, obtaining driving data sets of K sections of roads, and performing multi-dimensional analysis to obtain driving coefficients of K sections of roads and busy degree distribution data of K sections of roads, identifying associated aggregation nodes to obtain M associated aggregation nodes, and combining with the road paving network to obtain a set of M associated aggregation road sections; perceiving the road surface state, determining a set of damage characteristics of M associated aggregation road sections, and performing damage derivative prediction to obtain M associated damage derivative coefficients; using a road damage maintenance identifier for analysis to obtain a set of M damage maintenance schemes. The present invention solves the technical problem that the prior art cannot effectively predict the damage expansion trend and generate maintenance decisions, resulting in the influence on the operation efficiency of the road system during road maintenance, and achieves the technical effect of improving the road maintenance efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of road damage analysis, and particularly to an intelligent analysis and decision-making method and device for road damage based on driving data. Background Art

[0002] With the acceleration of the urbanization process and the continuous growth of transportation demands, the scale and complexity of modern road networks are constantly increasing. However, with the increase in vehicle flow and the improvement of road usage intensity, the detection and maintenance of road damage are facing increasingly severe challenges. Traditional road damage detection methods mainly rely on manual inspections and regular maintenance, which have problems such as low detection efficiency, slow response speed, and inaccurate decision-making, and cannot effectively analyze road damage in complex traffic environments. Especially in areas with high traffic flow, road damage that cannot be repaired in time may lead to more serious safety hazards and traffic congestion, affecting the operation efficiency of the entire road system. Summary of the Invention

[0003] This application provides an intelligent analysis and decision-making method and device for road damage based on driving data, which are used to solve the technical problem that the prior art cannot effectively predict the expansion trend of damage and generate maintenance decisions, resulting in the impact on the operation efficiency of the road system during road maintenance.

[0004] In view of the above problems, this application provides an intelligent analysis and decision-making method and device for road damage based on driving data.

[0005] In the first aspect of this application, there is provided an intelligent analysis and decision-making method for road damage based on driving data, and the method includes:

[0006] Obtain the road paving network of the target area, where the road paving network includes K road sections and a set of section interaction nodes, and both the K road sections and the set of section interaction nodes have location identifiers; collect driving data of the K road sections using a sensor array within a preset analysis window to obtain K sets of section driving data; perform multi-dimensional analysis on the K sets of section driving data to obtain K section driving coefficients and K section busyness distribution data; identify associated aggregation nodes for the set of section interaction nodes based on the K section driving coefficients to obtain M associated aggregation nodes, and combine with the road paving network to obtain M sets of associated aggregation road sections, where the M associated aggregation nodes and the M sets of associated aggregation road sections correspond one by one; traverse the M sets of associated aggregation road sections for road surface state perception to determine M sets of damage characteristics of the associated aggregation road sections; perform damage derivative prediction based on the M sets of damage characteristics of the associated aggregation road sections to obtain M associated damage derivative coefficients; use a road damage maintenance identifier to analyze the K section busyness distribution data, the M associated damage derivative coefficients, and the M sets of damage characteristics of the associated aggregation road sections to obtain M sets of damage maintenance plans.

[0007] In the second aspect of the present application, there is provided a road damage intelligent analysis and decision-making device based on driving data, and the device includes:

[0008] A network acquisition module, which acquires the road paving network of the target area. Among them, the road paving network includes K road segments and a set of segment interaction nodes, and both the K road segments and the set of segment interaction nodes have location identifiers; a driving data acquisition module, which uses a sensor array to collect driving data on the K road segments within a preset analysis window to obtain K sets of segment driving data; a data analysis module, which performs multi-dimensional analysis on the K sets of segment driving data to obtain K segment driving coefficients and K segment busyness distribution data; an associated aggregation node recognition module, which identifies associated aggregation nodes for the set of segment interaction nodes based on the K segment driving coefficients, obtains M associated aggregation nodes, and combines with the road paving network to obtain M sets of associated aggregation road segments, where the M associated aggregation nodes and the M sets of associated aggregation road segments correspond one by one; a road surface state perception module, which traverses the M sets of associated aggregation road segments to perform road surface state perception to determine M sets of damage characteristics of the associated aggregation road segments; a damage derivation prediction module, which performs damage derivation prediction based on the M sets of damage characteristics of the associated aggregation road segments to obtain M associated damage derivation coefficients; a damage maintenance plan determination module, which uses a road damage maintenance identifier to analyze the K segment busyness distribution data, the M associated damage derivation coefficients, and the M sets of damage characteristics of the associated aggregation road segments to obtain M sets of damage maintenance plans.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application obtains the road paving network of the target area. Among them, the road paving network includes K road sections and a set of section interaction nodes. Both the K road sections and the set of section interaction nodes have location identifiers. In a preset analysis window, a sensor array is used to collect driving data for the K road sections to obtain K sets of section driving data. Multidimensional analysis is performed on the K sets of section driving data to obtain K section driving coefficients and K section busyness distribution data. Based on the K section driving coefficients, association aggregation node recognition is performed on the set of section interaction nodes to obtain M association aggregation nodes, and M sets of association aggregation road sections are obtained in combination with the road paving network. Among them, the M association aggregation nodes and the M sets of association aggregation road sections correspond one by one. Traverse the M sets of association aggregation road sections for road surface state perception to determine M sets of damage characteristics of the association aggregation road sections. Based on the M sets of damage characteristics of the association aggregation road sections, damage derivative prediction is performed to obtain M associated damage derivative coefficients. The road damage maintenance identifier is used to analyze the K section busyness distribution data, the M associated damage derivative coefficients, and the M sets of damage characteristics of the association aggregation road sections to obtain M sets of damage maintenance plans. The present invention solves the technical problem that the prior art cannot effectively predict the damage expansion trend and generate maintenance decisions, resulting in the impact on the operation efficiency of the road system during road maintenance. By obtaining the road paving network and driving data, performing multidimensional analysis to identify the busyness and damage conditions, combining the damage characteristics of the aggregated road sections for damage derivative prediction, and finally using the road damage maintenance identifier to analyze and generate maintenance plans, the technical effect of improving the road maintenance efficiency is achieved. Description of the Drawings

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0012] Figure 1 It is a schematic flowchart of the intelligent analysis and decision-making method for road damage based on driving data provided by the embodiments of this application;

[0013] Figure 2 It is a schematic structural diagram of the intelligent analysis and decision-making device for road damage based on driving data provided by the embodiments of this application.

[0014] Description of the reference numerals: Network acquisition module 11, driving data acquisition module 12, data analysis module 13, association aggregation node recognition module 14, road surface state perception module 15, damage derivative prediction module 16, damage maintenance plan determination module 17. Detailed Embodiments

[0015] This application provides an intelligent analysis and decision-making method and device for road damage based on driving data, aiming to solve the technical problem that the prior art cannot effectively predict the damage expansion trend and generate maintenance decisions, resulting in the impact on the operation efficiency of the road system during road maintenance. By obtaining the road paving network and driving data, performing multi-dimensional analysis to identify the busyness and damage conditions, combining the damage characteristics of aggregated road sections for damage derivative prediction, and finally using a road damage maintenance identifier to analyze and generate a maintenance plan, the technical effect of improving road maintenance efficiency is achieved.

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

[0017] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, device, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0018] Embodiment 1, as Figure 1 shown, this application provides an intelligent analysis and decision-making method for road damage based on driving data, and the method includes:

[0019] Step S100: Obtain the road paving network of the target area, where the road paving network includes K road sections and a set of section interaction nodes, and both the K road sections and the set of section interaction nodes have location identifiers.

[0020] In the embodiments of the present application, first, the road data of the target area is collected through a geographic information system to determine the precise geographical coordinates and topological structure of each road in the target area, thereby forming a complete road network diagram. Next, the obtained road network diagram is used to divide the complete road network in the target area into multiple K road sections. These road sections are the basic units of the road network, independent of each other but connected through interaction nodes. Each road section is defined by the physical structure of the road or traffic rules, such as intersections and road junctions. The connection points between road sections are called section interaction nodes, which are the intersections, traffic light positions, or turning points of the road. By analyzing the road network, these interaction nodes are identified and assigned unique location identifiers to ensure their positioning in the entire network and subsequent analysis. Finally, geographical coordinates or other unique identifiers are assigned to each road section and interaction node.

[0021] Through the above process, a road paving network for the target area is obtained.

[0022] Step S200: Use a sensor array to collect driving data for the K road segments within a preset analysis window, and obtain K sets of driving data for the road segments.

[0023] In the embodiment of the present application, first, a fixed time period is set as the preset analysis window, such as one week or one month. Within this time range, through a sensor array arranged on vehicles or road infrastructure, continuous collection of driving data for the K road segments is carried out. The sensor array includes devices such as accelerometers, speedometers, and cameras, which can monitor the speed, flow, and type of vehicles in real time. Within the preset analysis window, detailed data of vehicles driving on each road segment are obtained through the sensor array, forming K sets of driving data for the roads. These data sets reflect the usage of each road segment, including the frequency of vehicle passage and speed changes.

[0024] Step S300: Perform multi-dimensional analysis on the K sets of driving data for the road segments to obtain K driving coefficients for the road segments and K data on the distribution of road segment busyness.

[0025] In the embodiment of the present application, first, the traffic flow data for each road segment is statistically analyzed to obtain the total number of vehicles passing through each road segment within a preset time window of one week or one month, thereby obtaining the traffic flow situation of the road segment. Then, the vehicle type data is analyzed to count the proportion of different types of vehicles, such as cars, trucks, buses, etc., because different vehicles have different degrees of wear on the road, especially heavy vehicles have a greater impact on the road surface. In addition, the vehicle speed data is analyzed by monitoring the average vehicle speed and speed fluctuations on each road segment to evaluate the traffic efficiency and smoothness of the road segment. By comprehensively analyzing these data, the driving coefficient for each road is calculated. At the same time, data on the distribution of busyness for each road segment is generated, showing the traffic density of the road at different time periods.

[0026] Through the above process, K driving coefficients for the road segments and K data on the distribution of road segment busyness are obtained.

[0027] Furthermore, in the method provided by the embodiment of the application, when performing multi-dimensional analysis on the K sets of driving data for the road segments to obtain K driving coefficients for the road segments and K data on the distribution of road segment busyness, it further includes:

[0028] Extract data from the K sets of road section traffic data according to the multi-dimensional traffic index matrix to obtain K sets of road section traffic volume data, K sets of road section vehicle type data, and K sets of road section vehicle speed data; perform single-day traffic volume concentration analysis and integrated traffic volume concentration analysis on the K sets of road section traffic volume data to obtain K sets of road section busyness distribution data and K sets of road section concentrated traffic volume respectively; perform integrated centralized identification on the K sets of road section vehicle type data and the K sets of road section vehicle speed data respectively to determine the K sets of road section concentrated vehicle type ratios and the K sets of road section concentrated vehicle speeds; perform comprehensive analysis on the K sets of road section concentrated traffic volume, the K sets of road section concentrated vehicle type ratios, and the K sets of road section concentrated vehicle speeds to obtain the K sets of road section traffic coefficients, where the K sets of road section traffic coefficients reflect the usage conditions of the K road sections.

[0029] In the embodiment of the present application, first, data is extracted from the K sets of road section traffic data according to the multi-dimensional traffic index matrix. The multi-dimensional traffic index matrix is an analysis tool. Through this matrix, the K sets of road section traffic data are classified according to different dimensions, and K sets of road section traffic volume data, K sets of road section vehicle type data, and K sets of road section vehicle speed data are extracted from the traffic data of the K road sections.

[0030] Next, analyze the traffic volume data. First, use time series clustering algorithms such as K-Means or DBSCAN to perform concentration analysis on the daily traffic volume, identify the traffic density at different time periods of each road section in a day, and obtain the single-day traffic volume data. Then, adopt time series analysis methods to perform integrated analysis through the cumulative traffic volume data to generate K sets of road section busyness distribution data and K sets of road section concentrated traffic volume.

[0031] After that, perform statistical analysis on the K sets of road section vehicle type data to determine the K sets of road section concentrated vehicle type ratios. This ratio reflects the proportion of different types of vehicles on each road section, such as the types of cars, trucks, buses, etc. At the same time, perform centralized identification on the K sets of road section vehicle speed data to determine the concentrated vehicle speeds of the K road sections. Similar to the process of traffic volume analysis, by analyzing the distribution of vehicle speeds on each road section within a certain time window, identify the area where the vehicle speed distribution is the most concentrated, which represents the concentrated vehicle speed of the road section. Through this process, the K sets of road section concentrated vehicle speeds are obtained.

[0032] After completing the independent analysis of traffic flow, vehicle type ratio, and vehicle speed data, it enters the stage of comprehensive analysis and calculation of driving coefficients. First, standardize the data for each dimension. Ensure that the traffic flow, vehicle type ratio, and vehicle speed data are comparable on the same scale through Z-score standardization or Min-Max normalization. Next, conduct a weighted comprehensive analysis of the traffic flow, vehicle type ratio, and vehicle speed data according to the preset weights. When calculating the driving coefficient for each road, multiply the standardized traffic flow, vehicle type ratio, and vehicle speed data by the corresponding weights to obtain the corresponding driving coefficients. Through this process, the driving coefficients for K sections of the road are obtained.

[0033] Furthermore, in the method provided by the application embodiment, for the set of traffic flow data of the K sections of the road, conduct a single-day traffic flow concentration analysis and an integrated traffic flow concentration analysis, respectively obtaining the traffic intensity distribution data of the K sections of the road and the concentrated traffic flow of the K sections of the road. It further includes:

[0034] Cluster the set of traffic flow data of the K sections of the road according to the data collection date to obtain the same-day traffic flow data clusters of the K sections of the road. Among them, each same-day traffic flow data cluster of a section of the road includes the traffic flow data of the section of the road collected at multiple time points within a data collection date; conduct a same-day traffic flow time period division on the same-day traffic flow data clusters of the K sections of the road to obtain the same-day traffic flow division node clusters of the K sections of the road; conduct node screening on the same-day traffic flow division node clusters of the K sections of the road to obtain the set of traffic flow division nodes of the K sections of the road; use the set of traffic flow division nodes of the K sections of the road to divide the same-day traffic flow data clusters of the K sections of the road and calculate the average value of the traffic flow data between two traffic flow division nodes of the sections of the road to obtain the traffic intensity distribution data of the K sections of the road.

[0035] In the embodiment of the present application, first cluster the set of traffic flow data of the K sections of the road according to the data collection date. Classify the traffic flow data of each section within the same date by using the timestamp clustering method to form the same-day traffic flow data clusters of the K sections of the road. Each cluster contains the traffic flow data at different time points within a certain day.

[0036] After obtaining the same-day data clusters, conduct further time period division on the same-day traffic flow data clusters of the K sections of the road through time series analysis. Use time series clustering algorithms such as K-Means or DBSCAN. According to the fluctuation pattern of the traffic flow, divide the traffic flow data of a day into multiple time period nodes to generate the same-day traffic flow division node clusters of the K sections of the road. These division nodes represent the key time periods of the traffic flow change in a day, such as the morning rush hour, evening rush hour, or low traffic volume period.

[0037] Next, node screening is performed on the same-day traffic flow data clusters of K sections to generate node clusters. To ensure that the divided nodes can represent typical traffic flow change periods, nodes with the highest frequency of occurrence in multiple dates are selected through frequency statistics. These nodes often reflect the typical characteristics of daily traffic flow fluctuations. After screening, a set of same-day traffic flow division nodes for K sections is generated.

[0038] After that, based on the set of same-day traffic flow division nodes for K sections, the data is segmented. Within the time interval between every two nodes, the average value of the traffic flow data is calculated. This step uses the segmented mean method to determine the traffic flow density in each time period by calculating the average traffic flow between two nodes. Finally, by summarizing and analyzing the traffic flow data of each time period, the traffic flow density distribution data for K sections is generated.

[0039] Furthermore, in the method provided by the application embodiment, when performing the same-day traffic flow period division on the same-day traffic flow data clusters of K sections to obtain the same-day traffic flow division node clusters of K sections, it further includes:

[0040] Extract the first-section same-day traffic flow data cluster from the same-day traffic flow data clusters of K sections, where the first-section same-day traffic flow data cluster includes multiple same-day traffic flow data sets for the section, and the multiple same-day traffic flow data sets correspond to multiple data collection dates; extract the first-section same-day traffic flow data set from the multiple same-day traffic flow data sets and sort them in chronological order from front to back to obtain the first-section same-day traffic flow data sequence; starting from the first-section same-day traffic flow data sequence, perform same-node data identification on the first-section same-day traffic flow data sequence according to a preset traffic flow bandwidth until the last first-section same-day traffic flow data in the first-section same-day traffic flow data sequence is identified, to obtain the first-section same-day traffic flow division node set; traverse the same-day traffic flow data clusters of K sections for the same-day traffic flow period division to obtain the same-day traffic flow division node clusters of K sections.

[0041] In the embodiment of the present application, first, the first-section same-day traffic flow data cluster is extracted from the same-day traffic flow data clusters of K sections. This data cluster includes the traffic flow data of the section on multiple different collection dates, and the data of each collection date forms an independent same-day traffic flow data set. Then, the first-section same-day traffic flow data set is extracted from these same-day traffic flow data sets and the data is sorted in chronological order. Here, the time series analysis method is used to sort the data according to the time stamps of each data point to generate the first-section same-day traffic flow data sequence.

[0042] After obtaining the same-day data sequence of the traffic flow of the first road segment, identify the key nodes in this sequence. Here, the method of preset traffic flow bandwidth is used. The traffic flow bandwidth refers to a set range of traffic flow changes, which is used to identify the time periods with obvious traffic flow changes in the data. By traversing the entire traffic flow data sequence and according to the preset traffic flow bandwidth, eligible nodes are detected through pattern recognition technology. This process is similar to finding traffic flow intervals that meet the set conditions in the data sequence. In these intervals, the traffic flow fluctuates greatly, which may indicate peak or trough periods of the day. For example, through time series clustering algorithms such as K-Means or DBSCAN, different time periods are clustered into different nodes according to the pattern of traffic flow fluctuations. By continuously traversing the data sequence until the last traffic flow data in the sequence is identified, a set of same-day division nodes of the traffic flow of the first road segment is generated.

[0043] Use the same process to process all K clusters of same-day traffic flow data of road segments. Traverse the data sets of each road segment and identify the same-node data of the traffic flow sequence according to the preset traffic flow bandwidth. The data of each road segment will go through time period division to form the corresponding set of nodes. Through multi-road segment time series analysis, the traffic flow change nodes of each road segment are identified at different time periods, and finally a cluster of same-day division nodes of the traffic flow of K road segments is generated.

[0044] Furthermore, the method provided by the application embodiment further includes:

[0045] Integrate the same-day data of multiple sets of same-day traffic flow data of the first road segment traffic flow cluster to obtain the integrated traffic flow of multiple road segments; calculate the mean value of the integrated traffic flow of multiple road segments to obtain the traffic flow mean value. Taking the traffic flow mean value as the iteration center, retrieve in the integrated traffic flow of multiple road segments according to the preset iteration step length to obtain the iterative integrated traffic flow of the road segment; determine whether the iteration density of the iterative integrated traffic flow of the road segment is greater than or equal to the iteration density of the iteration center. If so, update the iterative integrated traffic flow of the road segment as the iteration center and continue the iteration until the preset number of iterations is met. Take the integrated traffic flow of the road segment corresponding to the iteration center with the maximum iteration density during the iteration process as the target integrated traffic flow of the road segment; take the target integrated traffic flow of the road segment as the centralized traffic flow of the first road segment in the first road segment traffic flow cluster of the same day; perform centralized analysis of the integrated traffic flow of the K road segment traffic flow clusters of the same day to obtain the centralized traffic flow of the K road segments.

[0046] In the embodiment of the present application, first, same-day data integration is performed on multiple same-day traffic flow data sets in the same-day traffic flow data cluster of the first section. Each same-day data set represents traffic flow data at different time points within a specific date, recording the changes in traffic flow throughout the day in chronological order. Through data aggregation technology, these data sets from different dates are summarized to generate an integrated traffic flow set containing data for multiple days. For example, assume that the traffic flow data for a certain section over three days is 50 vehicles at 9 o'clock on the first day, 120 vehicles at 12 o'clock, 80 vehicles at 15 o'clock, 55 vehicles at 9 o'clock on the second day, 130 vehicles at 12 o'clock, 85 vehicles at 15 o'clock, 60 vehicles at 9 o'clock on the third day, 110 vehicles at 12 o'clock, and 90 vehicles at 15 o'clock. After aggregation, all these data are integrated into a whole to form an integrated traffic flow set, where the traffic flow at time points such as 9 o'clock and 12 o'clock respectively corresponds to the summary of traffic flows on different dates.

[0047] After completing the integration of the data, the average value of the integrated traffic flow of this section is calculated. Through statistical techniques, the average value of the traffic flow within all the collected dates is calculated, and this traffic flow average value reflects the overall traffic flow level of this section. After the average value calculation is completed, this traffic flow average value is used as the iteration center. The iteration center is the initial point for subsequent calculations. Starting from this average value, areas with dense traffic flow are gradually searched for in the data. An iterative search is carried out according to a preset iteration step size, and the step size defines the magnitude of the adjustment of the traffic flow value each time. For example, with a fixed traffic flow unit, such as the number of vehicles, as the step size, the data set area is gradually adjusted and searched for. Starting from the iteration center through an iterative algorithm, the integrated traffic flow data is gradually searched according to the preset step size. An iterative technique similar to the gradient search algorithm is used to locate the area with dense traffic flow by slightly adjusting the traffic flow value. In each iteration, the iteration density of this area is calculated. The density is calculated using kernel density estimation or histogram density estimation, and these methods can analyze the distribution concentration of traffic flow within a certain area. Through density calculation, it is judged whether the traffic flow data iterated currently is denser than the density of the initial iteration center. If it is found that the density of the integrated traffic flow in the iterated section is greater than or equal to the current iteration center density, the iteration result is updated as the new iteration center. In this way, the traffic flow distribution is gradually optimized to find the area with the densest traffic flow. The iteration process continues until the preset number of iterations is met or a specific termination condition is reached, such as the density no longer increasing significantly. In each iteration, the iteration density value is recorded and the optimal result is stored. When the iteration process ends, all the iteration results are compared, and the result with the maximum density is selected as the final integrated traffic flow of the target section. Subsequently, the integrated traffic flow of the target section is determined as the concentrated traffic flow of the first section in the same-day traffic flow data cluster of the first section.

[0048] Finally, the same analysis is performed on multiple road sections. For K road sections, the traffic flow data of each road section is processed in parallel, and the mean value calculation, iterative center setting, iterative search, and target traffic flow determination are respectively performed to obtain the traffic flow in the K road section concentration.

[0049] Step S400: Based on the driving coefficients of the K road sections, perform associated aggregation node recognition on the road section interaction node set to obtain M associated aggregation nodes, and combine with the road paving network to obtain M associated aggregation road section sets, where the M associated aggregation nodes and the M associated aggregation road section sets correspond one by one.

[0050] In the embodiment of the present application, when performing associated aggregation node recognition on the road section interaction node set based on the driving coefficients of the K road sections, first analyze the road section interaction node set based on the driving coefficients of the K road sections, and calculate the usage frequency of each node. By traversing the interaction nodes, generate multiple road section interaction node mapping road section sets to complete the first-level association, ensuring that each interaction node is mapped and associated with the corresponding road section. Then perform the second-level association through the location identifier and the road paving network to integrate adjacent nodes and road sections, generate the road section interaction node associated road section set, and reflect the stronger association between the nodes. According to the driving coefficient and the association information, perform weighted calculation according to the preset weight to obtain the driving frequency set of each node. Select the top M nodes with the driving frequency as the associated aggregation nodes, so as to identify M associated aggregation nodes. Finally, combine these aggregation nodes with the road network to form the corresponding M associated aggregation road section sets.

[0051] Further, in the method provided by the application embodiment, performing associated aggregation node recognition on the road section interaction node set based on the driving coefficients of the K road sections to obtain M associated aggregation nodes, and combining with the road paving network to obtain M associated aggregation road section sets, further includes:

[0052] Traverse the set of section interaction nodes for primary association to obtain multiple sets of mapped sections of section interaction nodes, where each set of mapped sections of section interaction nodes corresponds to a section interaction node and has a primary mapping identifier; perform secondary association on the multiple sets of mapped sections of section interaction nodes based on the location identifier and the road laying network to obtain multiple sets of associated sections of section interaction nodes, where the multiple sets of associated sections of section interaction nodes each have a secondary mapping identifier; perform weighted calculation on the basis of the K section driving coefficients, the multiple sets of mapped sections of section interaction nodes, and the sets of associated sections of section interaction nodes according to a preset weight distribution to obtain a set of driving frequencies of section interaction nodes; use the section interaction nodes corresponding to the top M driving frequencies of the set of driving frequencies of section interaction nodes as associated aggregation nodes to obtain M associated aggregation nodes; perform section aggregation based on the M associated aggregation nodes and the road laying network to obtain the M sets of associated aggregated sections.

[0053] In the embodiment of the present application, first, the set of section interaction nodes in the road network is traversed, and primary association is performed on each node. In this stage, geographic information system technology is used to analyze the road network in the target area to identify each section interaction node, which is the intersection point between sections, such as intersections or turning points. Using topological analysis technology, a set of mapped sections of section interaction nodes corresponding to each interaction node is generated. Each set contains an interaction node and its adjacent road sections, and a primary mapping identifier is assigned to indicate the basic association between the node and its connected sections.

[0054] After the primary association is completed, proximity analysis technology is further used to perform secondary association on these section interaction nodes. In this process, combined with the location identifiers of each node, the location coordinate data in the spatial database is used to analyze the distance between nodes through spatial connection, and nodes with a close topological relationship are identified. Through analysis tools such as adjacency matrices, multiple sets of associated sections of section interaction nodes are aggregated. These sets show a deeper association between nodes and are assigned secondary mapping identifiers to distinguish them from the primary association sets. This step ensures the identification of the closeness between adjacent nodes and sections through distance calculation and spatial relationship analysis.

[0055] After completing the first-level and second-level associations, all interaction nodes are further weighted calculated in combination with the K road section traffic coefficients, that is, the traffic flow, vehicle type ratio, speed and other data collected from the sensor array. The weights are preset by technical experts and are allocated according to the importance of previously defined traffic indicators. For example, the influence of dimensions such as road section traffic flow, vehicle type ratio and speed will be preset according to different traffic management strategies or road usage. The weighted average method is used to calculate the traffic frequency of the road section interaction nodes by weighted synthesis of the preset weights of each dimension. This value reflects the activity level of each interaction node in the road network. Through this process, the traffic frequency set of the road section interaction nodes is obtained.

[0056] After obtaining the set of traffic frequency of the interactive nodes of the road section, the quick sort algorithm is used to sort these nodes in descending order of traffic frequency, and the nodes in the first M positions in the sorting result are selected as the associated aggregation nodes. These nodes represent the most active and critical interactive nodes in the traffic network, which are areas with large traffic flow or more intersections.

[0057] Finally, based on these associated aggregated nodes and combined with the road paving network, clustering algorithms such as K-Means clustering or hierarchical clustering are used to aggregate the road segments connected to these nodes. The clustering algorithm groups the associated nodes and road segments into a group based on the similarities of the nodes and road segments, such as the driving frequency, and generates M associated aggregated road segment sets, which usually represent the areas with the busiest traffic flow.

[0058] Step S500: traverse the M associated aggregated road section sets to sense the road surface condition and determine the M associated aggregated road section damage feature sets.

[0059] In an embodiment of the present application, first, an acceleration sensor installed on the road infrastructure is used to sense the road surface state of M associated aggregated sections. The acceleration sensor can detect the vibration and deformation of the road itself in real time. To ensure the accuracy of the data, the collected acceleration data is low-pass filtered to help remove invalid parts of the data. Next, the filtered data is analyzed in the time domain to extract features reflecting road damage, such as the amplitude, frequency, and duration of vibration. Through time domain analysis, the damage pattern of the road surface is identified, such as vibration peaks caused by cracks or potholes.

[0060] After feature extraction is completed, a pre-trained support vector machine (SVM) model is used for damage classification. This SVM model is trained with a large number of known road surface damage datasets and can classify road surface damage into different types, such as cracks, potholes, or road surface wear, based on the vibration feature data collected in real time. The SVM classifier is pre-trained, and its training data includes multiple known road surface damage samples. These sample data come from historical data covering different types of road surface damage characteristics, such as information on cracks, potholes, and road surface wear detected by acceleration sensors. By using these training data, the SVM model can learn the patterns of each damage feature and accurately classify road surface damage in real-time applications. The training data includes the eigenvalue of the acceleration signal and the corresponding labels, such as the labels of different damage types.

[0061] Finally, a corresponding set of road section damage characteristics is generated for each road section according to the classification results. This set of characteristics includes the damage type, damage severity, and location information of the damage on the road section. All this information is summarized to generate M associated aggregated road section damage characteristic sets.

[0062] Step S600: Based on the M associated aggregated road section damage characteristic sets, perform damage derivative prediction to obtain M associated damage derivative coefficients.

[0063] In the embodiment of the present application, based on the M associated aggregated road section damage characteristic sets, first, a damage derivative prediction recognizer is used to analyze the damage characteristics of each road section to generate M initial damage derivative coefficient sets. These initial coefficients reflect the potential trend and speed of road section damage expansion. Then, these initial damage derivative coefficients are averaged to eliminate the influence of abnormal data or extreme situations, and finally M associated damage derivative coefficients are generated.

[0064] Furthermore, in the method provided by the embodiment of the application, based on the M associated aggregated road section damage characteristic sets and the M associated aggregated road section sets, performing damage derivative prediction to obtain M associated damage derivative coefficients further includes:

[0065] The damage derivative prediction recognizer is trained based on a loss function, where the loss function is:

[0066] ;

[0067] Where is the training loss amount, is the i-th output value of the damage derivative prediction recognizer, n is the number of samples in training, and n is an integer greater than or equal to 1. is the i-th expected output value of the damage-derived prediction recognizer during training; the damage-derived prediction recognizer is used to perform damage-derived analysis on the M associated aggregated road section damage features to obtain M initial damage-derived coefficient sets; the M initial damage-derived coefficient sets are averaged to obtain the M associated damage-derived coefficients.

[0068] In the embodiment of the present application, first, data related to road section damage is obtained from the historical database, including the historical damage feature set and the corresponding historical derived coefficient set. Among them, the historical damage feature set includes the specific forms and change information of cracks and potholes. The historical derived coefficients quantify the expansion speed and trend of the damage features. For example, the width by which a certain crack expands each month, or the depth by which a certain pothole gradually increases.

[0069] Next, the historical damage feature set and the corresponding historical derived coefficient set are used to train based on a time series model to obtain a damage-derived prediction recognizer. During the training process, the model is gradually optimized through a loss function to reduce the error between the model prediction value and the actual value, so that the model can more accurately predict the future damage-derived trend. After the model training is completed, the trained damage-derived prediction recognizer is used to analyze the damage features of the M associated aggregated road sections. Based on these damage features, the recognizer predicts the future expansion trend of each type of damage and outputs M initial damage-derived coefficient sets.

[0070] Finally, to ensure the accuracy of the prediction, the M initial damage-derived coefficient sets are averaged. By averaging the various initial damage-derived coefficients of each road section, outliers and data with large fluctuations are eliminated to obtain M associated damage-derived coefficients.

[0071] Step S700: The road damage maintenance recognizer is used to analyze the K road section busyness distribution data, the M associated damage-derived coefficients, and the M associated aggregated road section damage feature sets to obtain M damage maintenance plan sets.

[0072] In the embodiment of the present application, the road damage maintenance recognizer is a model trained through historical data. During training, first, the damage features, busyness distribution data, and maintenance records of historical roads are obtained from the historical database, and these data are in one-to-one correspondence. Through these data, a convolutional neural network is used to train the road damage maintenance recognizer, enabling the road damage maintenance recognizer to learn how to formulate maintenance strategies based on the road section damage situation, busyness, and damage expansion trend.

[0073] After training, the road damage maintenance recognizer analyzes the current busy degree distribution data of K sections, M associated damage derivation coefficients, and M associated aggregated section damage feature sets. The busy degree distribution data provides information on the traffic flow distribution of each section at different time periods. The damage derivation coefficient reflects the possibility and speed of future damage expansion of each section. The damage feature set contains specific damage information of each section, such as cracks, potholes, etc. By analyzing these data, the road damage maintenance recognizer comprehensively evaluates the damage status and traffic conditions of each section, and finally generates M damage maintenance plan sets.

[0074] For example, the busy degree distribution data of section A shows that the traffic flow is relatively large during the morning and evening rush hours on weekdays, and the traffic volume is relatively small at night and on weekends. According to the associated damage derivation coefficient, the crack expansion speed of this section is relatively fast, and it is expected that the crack width may increase by 20% within the next month. The current damage feature set indicates that there are multiple cracks and minor potholes on section A. The total length of the cracks is about 10 meters, and the depth of the potholes does not exceed 5 centimeters. Based on these data, since the damage expansion coefficient of section A is relatively high and the crack expansion speed is relatively fast, the maintenance plan generated by the road damage maintenance recognizer recommends giving priority to dealing with this section to prevent further deterioration of the damage.

[0075] Combined with the busy degree data, it is recommended to carry out repair work at night or on weekends to avoid affecting traffic during the rush hours. The recognizer recommends filling the cracks in a timely manner and dealing with the relatively shallow potholes to extend the service life of the section.

[0076] In the embodiments of this application, in summary, the embodiments of this application have at least the following technical effects:

[0077] This application obtains the road paving network of the target area. Among them, the road paving network includes K road sections and a set of section interaction nodes. Both the K road sections and the set of section interaction nodes have location identifiers. The driving data of the K road sections are collected by using a sensor array within a preset analysis window to obtain K sets of section driving data. Multidimensional analysis is performed on the K sets of section driving data to obtain K section driving coefficients and K section busyness distribution data. Based on the K section driving coefficients, the set of section interaction nodes is used to identify associated aggregation nodes, obtaining M associated aggregation nodes, and combining with the road paving network to obtain M sets of associated aggregation road sections. Among them, the M associated aggregation nodes and the M sets of associated aggregation road sections correspond one by one. The pavement state of the M sets of associated aggregation road sections is sensed by traversing to determine M sets of damage characteristics of the associated aggregation road sections. Damage derivative prediction is performed based on the M sets of damage characteristics of the associated aggregation road sections to obtain M associated damage derivative coefficients. The road damage maintenance identifier is used to analyze the K section busyness distribution data, the M associated damage derivative coefficients, and the M sets of damage characteristics of the associated aggregation road sections to obtain M sets of damage maintenance plans. The present invention solves the technical problem that the prior art cannot effectively predict the damage expansion trend and generate maintenance decisions, resulting in the influence on the operation efficiency of the road system during road maintenance. By obtaining the road paving network and driving data, performing multidimensional analysis to identify the busyness and damage conditions, combining the damage characteristics of the aggregated road sections for damage derivative prediction, and finally using the road damage maintenance identifier to analyze and generate maintenance plans, the technical effect of improving the road maintenance efficiency is achieved.

[0078] Embodiment 2. Based on the same inventive concept as the method for intelligent analysis and decision-making of road damage based on driving data in the foregoing embodiment, as Figure 2 shown, this application provides a device for intelligent analysis and decision-making of road damage based on driving data. The device in the embodiment of this application and the method embodiment are based on the same inventive concept. Among them, the device includes:

[0079] A network acquisition module 11, where the network acquisition module 11 acquires the road paving network of the target area. Among them, the road paving network includes K road sections and a set of section interaction nodes, and both the K road sections and the set of section interaction nodes have location identifiers; a driving data acquisition module 12, where the driving data acquisition module 12 uses a sensor array to collect driving data for the K road sections within a preset analysis window to obtain K sets of section driving data; a data analysis module 13, where the data analysis module 13 performs multi-dimensional analysis on the K sets of section driving data to obtain K section driving coefficients and K section busyness distribution data; an associated aggregation node recognition module 14, where the associated aggregation node recognition module 14 recognizes associated aggregation nodes for the set of section interaction nodes based on the K section driving coefficients to obtain M associated aggregation nodes, and combines with the road paving network to obtain M sets of associated aggregation road sections, where the M associated aggregation nodes and the M sets of associated aggregation road sections correspond one by one; a road surface state perception module 15, where the road surface state perception module 15 traverses the M sets of associated aggregation road sections to perform road surface state perception to determine M sets of damage characteristics of the associated aggregation road sections; a damage derivation prediction module 16, where the damage derivation prediction module 16 performs damage derivation prediction based on the M sets of damage characteristics of the associated aggregation road sections to obtain M associated damage derivation coefficients; a damage maintenance plan determination module 17, where the damage maintenance plan determination module 17 uses a road damage maintenance identifier to analyze the K section busyness distribution data, the M associated damage derivation coefficients, and the M sets of damage characteristics of the associated aggregation road sections to obtain M sets of damage maintenance plans.

[0080] Further, the device is also used to implement the following functions:

[0081] Extract data from the K sets of section driving data according to a multi-dimensional driving index matrix to obtain K sets of section traffic flow data, K sets of section vehicle type data, and K sets of section vehicle speed data; perform single-day traffic flow concentration analysis and integrated traffic flow concentration analysis on the K sets of section traffic flow data to obtain K section busyness distribution data and K section concentrated traffic flows respectively; perform integrated concentration recognition on the K sets of section vehicle type data and the K sets of section vehicle speed data respectively to determine K section concentrated vehicle type ratios and K section concentrated vehicle speeds; perform comprehensive analysis on the K section concentrated traffic flows, the K section concentrated vehicle type ratios, and the K section concentrated vehicle speeds to obtain the K section driving coefficients, where the K section driving coefficients reflect the usage conditions of the K road sections.

[0082] Further, the device is also used to implement the following functions:

[0083] Cluster the traffic flow data sets of the K sections according to the data collection date to obtain the traffic flow same-day data clusters of the K sections. Among them, each traffic flow same-day data cluster of a section includes the traffic flow data of the section collected at multiple time points within a data collection date; perform same-day traffic flow period division on the traffic flow same-day data clusters of the K sections to obtain the K traffic flow same-day division node clusters; perform node screening on the K traffic flow same-day division node clusters to obtain the K traffic flow division node sets; use the K traffic flow division node sets to perform data division on the traffic flow same-day data clusters of the K sections and calculate the average value of the traffic flow data between two traffic flow division nodes to obtain the traffic flow busyness distribution data of the K sections.

[0084] Further, the device is also used to implement the following functions:

[0085] Extract the first traffic flow same-day data cluster from the traffic flow same-day data clusters of the K sections. Among them, the first traffic flow same-day data cluster of a section includes multiple traffic flow same-day data sets, and the multiple traffic flow same-day data sets correspond to multiple data collection dates; extract the first traffic flow same-day data set from the multiple traffic flow same-day data sets and sort them in the order from front to back in time to obtain the first traffic flow same-day data sequence; use the first traffic flow same-day data sequence as the starting point and perform same-node data identification on the first traffic flow same-day data sequence according to the preset traffic flow bandwidth until the last first traffic flow same-day data in the first traffic flow same-day data sequence is identified to obtain the first traffic flow same-day division node set; traverse the traffic flow same-day data clusters of the K sections to perform same-day traffic flow period division to obtain the K traffic flow same-day division node clusters.

[0086] Further, the device is also used to implement the following functions:

[0087] Integrate the same-day data of multiple same-day traffic data sets of the traffic flow of the first section to obtain the integrated traffic flow of multiple sections; calculate the mean value of the integrated traffic flow of multiple sections to obtain the traffic flow mean value. Using the traffic flow mean value as the iteration center, retrieve in the integrated traffic flow of multiple sections according to the preset iteration step size to obtain the iterative integrated traffic flow of the section; determine whether the iteration density of the iterative integrated traffic flow of the section is greater than or equal to the iteration density of the iteration center. If so, update the iterative integrated traffic flow of the section as the iteration center and continue the iteration until the preset number of iterations is satisfied. Take the integrated traffic flow of the section corresponding to the iteration center with the maximum iteration density during the iteration process as the target integrated traffic flow of the section; take the target integrated traffic flow of the section as the centralized traffic flow of the first section of the same-day traffic data cluster of the first section; perform centralized analysis of the integrated traffic flow on the K same-day traffic data clusters of the traffic flow of the section to obtain the centralized traffic flow of the K sections.

[0088] Further, the device is also used to implement the following functions:

[0089] Traverse the set of section interaction nodes for first-level association to obtain multiple sets of mapped sections of section interaction nodes, where each set of mapped sections of section interaction nodes corresponds to a section interaction node and has a first-level mapping identifier; perform second-level association on the multiple sets of mapped sections of section interaction nodes based on the location identifier and the road paving network to obtain multiple sets of associated sections of section interaction nodes, where the multiple sets of associated sections of section interaction nodes each have a second-level mapping identifier; perform weighted calculation according to the K section driving coefficients, multiple sets of mapped sections of section interaction nodes, and the sets of associated sections of section interaction nodes according to the preset weight distribution to obtain the set of driving frequencies of section interaction nodes; take the section interaction nodes corresponding to the driving frequencies of the section interaction nodes in the top M positions in the set of driving frequencies of section interaction nodes as the associated aggregation nodes to obtain M associated aggregation nodes; perform section aggregation based on the M associated aggregation nodes and the road paving network to obtain the M sets of associated aggregated sections.

[0090] Further, the device is also used to implement the following functions:

[0091] Train the damage-derived prediction recognizer based on the loss function, where the loss function is:

[0092] ;

[0093] Where is the training loss amount, is the i-th output value of the damage-derived prediction recognizer, n is the number of samples in the training, and n is an integer greater than or equal to 1. It is the i-th expected output value of the injury-derived prediction recognizer during training; the injury-derived prediction recognizer is used to perform injury-derived analysis on the M associated aggregated section injury features to obtain M initial injury-derived coefficient sets; the M initial injury-derived coefficient sets are averaged to obtain the M associated injury-derived coefficients.

[0094] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0096] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A road damage intelligent analysis and decision-making method based on driving data, characterized in that: The method comprises: Acquire a road paving network of a target area, wherein the road paving network includes K road segments and a set of road segment interaction nodes, wherein the K road segments and the set of road segment interaction nodes both have location identifiers; Using a sensor array to collect driving data on the K road sections within a preset analysis window to obtain a set of driving data on the K road sections; Performing multi-dimensional analysis on the K road section traffic data sets to obtain the K road section traffic coefficients and K road section busyness distribution data; Based on the K road section driving coefficients, the road section interaction node set is identified with associated aggregation nodes to obtain M associated aggregation nodes, and M associated aggregation road section sets are obtained in combination with the road paving network, wherein the M associated aggregation nodes correspond to the M associated aggregation road section sets one by one; Traversing the M associated aggregated road section sets to sense road conditions and determine the M associated aggregated road section damage feature sets; Perform damage derivative prediction based on the M associated aggregated road section damage feature sets to obtain M associated damage derivative coefficients; Using a road damage maintenance identifier, the busyness distribution data of the K road sections, the M associated damage derivative coefficients and the M associated aggregated road section damage feature sets are analyzed to obtain a set of M damage maintenance solutions; Among them, based on the K road section driving coefficients, the road section interaction node set is identified with associated aggregation nodes to obtain M associated aggregation nodes, and the M associated aggregation road section sets are obtained in combination with the road paving network, including: Traversing the road segment interaction node set to perform primary association, and obtaining multiple road segment interaction node mapping road segment sets, wherein each road segment interaction node mapping road segment set corresponds to a road segment interaction node and has a primary mapping identifier; Performing secondary association on the plurality of road segment interaction node mapping road segment sets based on the location identifier and the road paving network to obtain a plurality of road segment interaction node associated road segment sets, wherein the plurality of road segment interaction node associated road segment sets respectively have secondary mapping identifiers; A set of driving frequency of the segment interaction nodes is obtained by performing weighted calculation according to the K segment driving coefficients, the multiple segment interaction node mapping segment sets and the segment interaction node associated segment set according to a preset weight distribution; The road segment interaction nodes corresponding to the road segment interaction node driving frequencies located in the first M positions in the road segment interaction node driving frequency set are used as associated aggregation nodes to obtain M associated aggregation nodes; Road segments are aggregated based on the M associated aggregation nodes and the road paving network to obtain the M associated aggregation road segment sets.

2. The road damage intelligent analysis and decision-making method based on driving data according to claim 1, characterized in that: Performing a multi-dimensional analysis on the K road section driving data set to obtain the K road section driving coefficients and K road section busyness distribution data, including: Extracting data from the K road section driving data sets according to the multidimensional driving index matrix to obtain K road section traffic flow data sets, K road section vehicle type data sets and K road section vehicle speed data sets; Performing a single-day traffic flow centralized analysis and an integrated traffic flow centralized analysis on the K road section traffic flow data sets, and obtaining the busyness distribution data of the K road sections and the centralized traffic flow of the K road sections respectively; Integrate and centrally identify the K road section vehicle type data sets and the K road section vehicle speed data sets respectively, and determine the concentrated vehicle type ratios and concentrated vehicle speeds of the K road sections; A comprehensive analysis is performed on the concentrated traffic flow of the K road sections, the concentrated vehicle type ratio of the K road sections and the concentrated vehicle speed of the K road sections to obtain the driving coefficients of the K road sections, wherein the driving coefficients of the K road sections reflect the usage of the K road sections.

3. The road damage intelligent analysis and decision-making method based on driving data according to claim 2 is characterized in that: The traffic flow data set of the K sections is subjected to a single-day traffic flow centralized analysis and an integrated traffic flow centralized analysis to obtain the busyness distribution data of the K sections and the centralized traffic flow of the K sections, respectively, including: Clustering the K road section traffic flow data sets according to the data collection date to obtain K road section traffic flow same-day data clusters, wherein each road section traffic flow same-day data cluster includes the road section traffic flow data collected at multiple time points within one data collection date; The K road section traffic flow same-day data clusters are divided into traffic flow time periods on the same day to obtain K road section traffic flow same-day divided node clusters; Perform node screening on the K road section traffic flow same-day partition node clusters to obtain K road section traffic flow partition node sets; The K road section traffic flow division node sets are used to divide the K road section traffic flow same-day data clusters and the traffic flow data between two road section traffic flow division nodes are averaged to obtain the K road section busyness distribution data.

4. The road damage intelligent analysis and decision-making method based on driving data according to claim 3 is characterized in that: The K road section traffic flow same-day data clusters are divided into traffic flow time periods on the same day to obtain K road section traffic flow same-day division node clusters, including: Extracting a first road section traffic flow same-day data cluster from the K road section traffic flow same-day data clusters, wherein the first road section traffic flow same-day data cluster includes multiple road section traffic flow same-day data sets, and the multiple road section traffic flow same-day data sets correspond to multiple data collection dates; Extracting a first road section traffic flow same-day data set from the plurality of road section traffic flow same-day data sets, and sorting them in chronological order to obtain a first road section traffic flow same-day data sequence; Taking the first road section traffic flow same-day data sequence as the starting point, identifying the same-node data of the first road section traffic flow same-day data sequence according to the preset traffic flow bandwidth, until the last first road section traffic flow same-day data in the first road section traffic flow same-day data sequence is identified, and obtaining the first road section traffic flow same-day division node set; The K road section traffic flow same-day data clusters are traversed to divide the traffic flow on the same day into time periods, and the K road section traffic flow same-day division node clusters are obtained.

5. The road damage intelligent analysis and decision-making method based on driving data according to claim 4 is characterized in that: include: Performing same-day data integration on multiple road section traffic flow same-day data sets of the first road section traffic flow same-day data cluster to obtain multiple road section integrated traffic flows; Calculate the average of the integrated traffic flow of the multiple road sections to obtain the average of the traffic flow, take the average of the traffic flow as the iteration center, search the integrated traffic flow of the multiple road sections according to the preset iteration step length, and obtain the integrated traffic flow of the iterated road section; Determine whether the iteration density of the iterative section integrated traffic flow is greater than or equal to the iteration density of the iteration center. If so, update the iterative section integrated traffic flow to the iteration center, continue iterating until the preset number of iterations is met, and take the section integrated traffic flow corresponding to the iteration center corresponding to the maximum iteration density during the iteration process as the target section integrated traffic flow; The integrated traffic flow of the target road section is used as the concentrated traffic flow of the first road section of the first road section traffic flow data cluster on the same day; The K road sections' traffic flow data clusters on the same day are integrated and analyzed to obtain the concentrated traffic flow of the K road sections.

6. The road damage intelligent analysis and decision-making method based on driving data according to claim 1, characterized in that: Based on the M associated aggregated road section damage feature sets and the M associated aggregated road section sets, damage derivative prediction is performed to obtain M associated damage derivative coefficients, including: The damage derivative prediction identifier is obtained based on loss function training, wherein the loss function is: ； in, is the training loss, is the i-th output value of the damage derivative prediction identifier, n is the number of samples in training and is an integer greater than or equal to 1, is the i-th expected output value of the damage derivative prediction identifier in training; Using a damage derivative prediction identifier to perform damage derivative analysis on the damage characteristics of the M associated aggregated road sections to obtain M initial damage derivative coefficient sets; The M initial damage derivative coefficient sets are averaged to obtain the M associated damage derivative coefficients.

7. The intelligent road damage analysis and decision-making device based on driving data is characterized by: The device is used to execute the road damage intelligent analysis and decision-making method based on driving data according to any one of claims 1 to 6, comprising: A network acquisition module, wherein the network acquisition module acquires a road paving network of a target area, wherein the road paving network includes K road segments and a set of road segment interaction nodes, wherein the K road segments and the set of road segment interaction nodes both have location identifiers; A driving data collection module, wherein the driving data collection module collects driving data on the K road sections using a sensor array within a preset analysis window to obtain a set of driving data on the K road sections; A data analysis module, wherein the data analysis module performs multi-dimensional analysis on the K road section driving data set to obtain the K road section driving coefficients and K road section busyness distribution data; An associated aggregation node identification module, wherein the associated aggregation node identification module performs associated aggregation node identification on the road segment interaction node set based on the K road segment driving coefficients to obtain M associated aggregation nodes, and obtains M associated aggregation road segment sets in combination with the road paving network, wherein the M associated aggregation nodes correspond to the M associated aggregation road segment sets one by one; A road surface state perception module, wherein the road surface state perception module traverses the M associated aggregated road section sets to perceive the road surface state and determines a damage feature set of the M associated aggregated road sections; A damage derivative prediction module, wherein the damage derivative prediction module performs damage derivative prediction based on the M associated aggregated road section damage feature sets to obtain M associated damage derivative coefficients; A damage maintenance plan determination module uses a road damage maintenance identifier to analyze the K road section busyness distribution data, the M associated damage derivative coefficients and the M associated aggregated road section damage feature sets to obtain M damage maintenance plan sets.

Images