An asphalt road quality evaluation method and system based on multi-data fusion

Abstract

This invention provides a method and system for assessing the quality of asphalt roads based on multi-data fusion. First, multi-source heterogeneous data of the asphalt road is acquired, including pavement surface image data, ground-penetrating radar scan data, and pavement physical parameter data. Then, the multi-source heterogeneous data is preprocessed, including time synchronization, spatial registration, and data standardization. Subsequently, features are extracted from each source data after preprocessing to obtain pavement surface feature vectors, pavement internal structure feature vectors, and pavement physical feature vectors. Then, based on a confidence-weighted strategy, the source feature vectors are fused at multiple levels to obtain a fused feature vector. Finally, the asphalt road quality is assessed based on the fused feature vector, and a quality grade assessment result is output. This invention achieves a comprehensive and accurate assessment of asphalt road quality through deep fusion of multi-source heterogeneous data.

Description

Technical Field

[0001] This invention relates to the field of road engineering and intelligent detection technology, and in particular to a method and system for assessing the quality of asphalt roads based on multi-data fusion. Background Technology

[0002] Asphalt roads are the most common type of road surface in urban and highway networks, and their quality directly affects driving safety and road lifespan. With the continuous growth of traffic volume and the frequent occurrence of extreme weather events, problems such as cracks, potholes, ruts, and loosening of asphalt pavements are becoming increasingly prominent. How to efficiently and accurately assess the quality of asphalt roads has become a core requirement for road maintenance and management.

[0003] Currently, asphalt road quality assessment mainly relies on manual inspections and single testing methods. Manual inspections are inefficient and highly subjective, making it difficult to quickly assess large areas of the road surface. Single testing methods, such as surface defect detection based on image recognition, internal structure detection based on ground-penetrating radar, and load-bearing capacity detection based on deflectometers, can only obtain information about one dimension of the road surface and cannot comprehensively reflect the road quality condition.

[0004] In the prior art, Chinese patent application CN120278586A proposes a road construction assessment method and system based on multi-source data fusion. This method collects vehicle driving data, construction equipment status data, and road condition data for the construction section, and integrates and assesses these data after cleaning, reduction, and standardization. However, this patent has the following shortcomings: First, its data fusion strategy only involves simple data integration, without providing specific fusion algorithms and weight allocation mechanisms, resulting in insufficient robustness of the fusion strategy; second, this scheme targets the assessment of the road construction phase, not the quality assessment of already operational roads, and the two differ fundamentally in data type, assessment indicators, and assessment objectives; third, it lacks processing for time synchronization and spatial registration between heterogeneous data, making effective fusion of multi-source data difficult to guarantee.

[0005] Furthermore, Chinese patent application CN119395273A proposes a non-destructive testing method and system for asphalt pavements. This method uses fiber optic grating sensors to collect pavement strain data, combining this data with a ground detection module and a region determination module for detection and early warning. However, this patent only uses single-source fiber optic grating sensor data, failing to acquire information on pavement surface defects and internal structural flaws, resulting in a limited evaluation dimension. Moreover, this solution lacks a multi-source data fusion mechanism; when sensor data deviates or fails, it lacks redundant verification and cross-validation methods, limiting the reliability of the evaluation results.

[0006] Therefore, there is an urgent need for a method and system for assessing the quality of asphalt roads that can integrate multi-source heterogeneous data and achieve comprehensive and accurate evaluation. Summary of the Invention

[0007] In view of the above problems, the purpose of this invention is to provide a method and system for assessing the quality of asphalt roads based on multi-data fusion. By fusing multi-source heterogeneous data such as pavement surface image data, ground-penetrating radar scan data, and pavement physical parameter data, a comprehensive and accurate assessment of the quality of asphalt roads can be achieved.

[0008] The first aspect of this invention provides a method for assessing the quality of asphalt roads based on multi-data fusion, comprising: Acquire multi-source heterogeneous data of asphalt roads, including road surface image data, ground-penetrating radar scan data, and road physical parameter data; The multi-source heterogeneous data is preprocessed, including time synchronization, spatial registration, and data standardization. Feature extraction was performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector, and pavement physical feature vector. Based on a credibility-weighted strategy, the source feature vectors are fused at multiple levels to obtain a fused feature vector. Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

[0009] In this scheme, the step of evaluating the quality of asphalt roads based on the fused feature vector and outputting the quality grade evaluation result includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality grade assessment results will be revised based on the results of the special testing and evaluation.

[0010] In this scheme, the multi-level fusion of the source feature vectors based on the credibility weighting strategy to obtain the fused feature vector includes: Calculate the confidence weight of each source feature vector separately; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector. This scheme also includes: Acquire external correlation data, including meteorological data and historical traffic flow data for the area where the road is located; The confidence-weighted strategy is used to perform multi-level fusion of the source feature vectors to obtain a fused feature vector, including: The external correlation data is used as an auxiliary fusion layer and jointly fused with the source feature vectors to obtain the fused feature vector. The meteorological data is used to correct the weighting of temperature-sensitive physical parameters, and the historical traffic flow data is used to correct the evaluation benchmark for road wear-related indicators.

[0011] This plan also includes: When extreme weather conditions or special operating conditions are detected, switch to emergency assessment mode; in the emergency assessment mode, adjust the confidence weight allocation strategy of each source feature vector to reduce the weight of data sources that are more affected by extreme conditions. The trigger threshold for the quality warning threshold is simultaneously lowered to improve the sensitivity of early warning for potential quality risks. In this scheme, the multi-level fusion of the source feature vectors based on a credibility-weighted strategy to obtain a fused feature vector includes: Identify inherent missing patterns in each source data, including image occlusion areas, radar signal attenuation intervals, and sensor sampling blind spots; The inherent missing patterns are used as complementary signals to assign enhanced weights from other data sources to the missing regions during the fusion process; Based on the enhanced weights, cross-source compensation fusion is performed on the source feature vectors to obtain the fused feature vector.

[0012] A second aspect of the present invention provides an asphalt road quality assessment system based on multi-data fusion, comprising a memory and a processor. The memory includes a program for an asphalt road quality assessment method based on multi-data fusion. When the processor executes the program for the asphalt road quality assessment method based on multi-data fusion, it performs the following steps: Acquire multi-source heterogeneous data of asphalt roads, including road surface image data, ground-penetrating radar scan data, and road physical parameter data; The multi-source heterogeneous data is preprocessed, including time synchronization, spatial registration, and data standardization. Feature extraction was performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector, and pavement physical feature vector. Based on a credibility-weighted strategy, the source feature vectors are fused at multiple levels to obtain a fused feature vector. Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

[0013] In this scheme, the step of evaluating the quality of asphalt roads based on the fused feature vector and outputting the quality grade evaluation result includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality level assessment results are revised based on the results of specialized testing and evaluation. In this scheme, the multi-level fusion of the source feature vectors based on a credibility-weighted strategy to obtain a fused feature vector includes: Calculate the confidence weight of each source feature vector separately; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector.

[0014] A third aspect of the present invention provides a computer-readable storage medium comprising a program for an asphalt road quality assessment method based on multi-data fusion, wherein when the program is executed by a processor, it implements the steps of the asphalt road quality assessment method based on multi-data fusion as described in any of the preceding claims.

[0015] This invention discloses a method and system for assessing asphalt road quality based on multi-data fusion. By fusing heterogeneous data from multiple sources, including pavement surface image data, ground-penetrating radar scan data, and pavement physical parameter data, and after preprocessing, feature extraction, and multi-level fusion with confidence weighting, a comprehensive and accurate assessment of asphalt road quality is achieved. This method overcomes the shortcomings of insufficient assessment dimensions from a single data source, ensures the reliability of the fusion results through cross-validation and weighted arbitration mechanisms, and exhibits robust adaptive capabilities under extreme conditions and in scenarios with missing data, providing a scientific basis for asphalt road maintenance decisions. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope.

[0017] Figure 1 A flowchart of an asphalt road quality assessment method based on multi-data fusion according to the present invention is shown; Figure 2 A flowchart of a threshold-guided quality assessment correction method provided by an embodiment of the present invention is shown; Figure 3A flowchart of a multimodal cross-validation fusion method provided by an embodiment of the present invention is shown; Figure 4 A block diagram of an asphalt road quality assessment system based on multi-data fusion according to the present invention is shown. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Unless otherwise defined, all terms (including technical and scientific terms) used in embodiments of this invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in a common dictionary shall be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as being interpreted in an idealized or highly formalized sense, unless expressly defined in this embodiment of the invention.

[0020] The terms "first," "second," and similar terms used in the embodiments of this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "a," "one," or "the" do not indicate a quantity limitation, but rather indicate the presence of at least one. Similarly, terms such as "including" or "comprising" mean that the element or object preceding the word encompasses the element or object listed after the word and its equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The steps preceding or following the steps in the methods of the embodiments of this invention are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them. Additionally, the functional modules in the various embodiments of this invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0021] Figure 1 A flowchart of an asphalt road quality assessment method based on multi-data fusion according to the present invention is shown.

[0022] like Figure 1 As shown, the first aspect of this invention discloses a method for assessing the quality of asphalt roads based on multi-data fusion, the method comprising: S102, acquire multi-source heterogeneous data of asphalt road, the multi-source heterogeneous data including road surface image data, ground penetrating radar scan data and road physical parameter data; S104, preprocess the multi-source heterogeneous data, the preprocessing including time synchronization, spatial registration and data standardization; S106, feature extraction is performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector and pavement physical feature vector; S108, Multi-level fusion of the source feature vectors is performed based on a confidence weighting strategy to obtain a fused feature vector; S110, Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

[0023] It should be noted that in this embodiment, the multi-source heterogeneous data covers three sources: road surface images, ground-penetrating radar scans, and deflection, temperature, and humidity data. Image data is collected by vehicle-mounted cameras or drones to identify visible defects such as cracks and potholes; radar data is acquired from vehicle-mounted ground-penetrating radar to detect internal voids and interlayer bonding issues; physical parameters are measured on-site by deflectometers and various sensors to reflect bearing capacity and temperature and humidity conditions. The preprocessing stage involves three steps: first, aligning data from different sampling frequencies onto the same timeline; second, unifying the coordinates of each source to the same road surface grid; and finally, normalization to eliminate dimensional differences. For feature extraction, the image side focuses on texture and shape, the radar side captures reflected waves and anomalous signals, and the physical parameter side calculates statistics and trends. During fusion, weights are assigned based on the historical accuracy, current signal-to-noise ratio, and data completeness of each source. The weighted feature vectors are then mapped to four levels (excellent, good, average, and poor) according to a preset grading standard. This process effectively addresses the issue of insufficient coverage from a single data source.

[0024] Figure 2 A flowchart of a threshold-guided quality assessment correction method provided by an embodiment of the present invention is shown.

[0025] like Figure 2 As shown in the embodiment of the present invention, the step of evaluating the quality of asphalt roads based on the fused feature vector and outputting the quality grade evaluation result specifically includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality grade assessment results are revised based on the results of specialized testing and evaluation. It should be noted that in this embodiment, the quality grading thresholds are pre-set critical values ​​for various dimensions of indicators based on road grade, design life, and maintenance standards, including crack density thresholds, rut depth thresholds, deflection value thresholds, and internal defect area ratio thresholds. When a certain dimension indicator exceeds its corresponding quality warning threshold, it indicates that the pavement performance represented by that dimension has approached or exceeded its safe service limit. At this time, the dimension indicator is marked as an abnormal indicator, and the specialized testing and evaluation process is automatically triggered. The specialized testing and evaluation process includes: for the type of damage corresponding to the abnormal indicator, calling the corresponding professional testing algorithm for refined analysis. For example, when the crack density indicator exceeds the threshold, a high-resolution image analysis algorithm is called to identify the crack type and quantify its severity; when the deflection value indicator exceeds the threshold, a structural bearing capacity analysis model is called to estimate the remaining life. Based on the refined results of the specialized testing and evaluation, the initial quality grade assessment results are revised to ensure that the assessment results accurately reflect the actual quality condition of the pavement. By employing the aforementioned technical means, we can achieve timely early warning and precise correction of potential quality risks, thus avoiding the problem of overlooking serious defects due to overly lenient threshold settings in conventional assessments.

[0026] Figure 3 A flowchart of a multimodal cross-validation fusion method provided by an embodiment of the present invention is shown.

[0027] like Figure 3 As shown in the embodiment of the present invention, the multi-level fusion of the source feature vectors based on the confidence weighting strategy to obtain the fused feature vector specifically includes: calculating the confidence weight of each source feature vector respectively; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector.

[0028] It should be noted that in this embodiment, the confidence weight is between 0 and 1, and the sum of all sources equals 1. This value is determined by combining three aspects of information: historical detection accuracy records, current data quality scores, and manually set expert confidence levels. In actual operation, conflicts may occasionally arise between the sources, such as an image indicating a crack while radar determines the interior is intact. In such cases, a cross-validation process is used: first, the consistency of each source and its alignment with surrounding data are checked, and then weighted arbitration is performed. The arbitration rule is that if the weight of a source is greater than a set advantage threshold, its value is directly accepted; if the weights of several sources are similar, a weighted vote is taken. Finally, the arbitration result replaces the original contradictory values ​​and is filled into the fusion vector. In actual testing, this mechanism can resolve most contradictions.

[0029] According to an embodiment of the present invention, before acquiring the multi-source heterogeneous data of asphalt roads, the method further includes: Acquire external correlation data, including meteorological data and historical traffic flow data for the area where the road is located; The confidence-weighted strategy is used to perform multi-level fusion of the source feature vectors to obtain a fused feature vector, including: The external correlation data is used as an auxiliary fusion layer and jointly fused with the source feature vectors to obtain the fused feature vector. The meteorological data is used to correct the weighting of temperature-sensitive physical parameters, and the historical traffic flow data is used to correct the evaluation benchmark for road wear-related indicators.

[0030] It should be noted that in this embodiment, the external correlation data mainly refers to two categories: meteorological data and traffic flow. Meteorological data includes ambient temperature, precipitation, and ultraviolet radiation intensity. These parameters are directly linked to asphalt performance; for example, high temperatures soften asphalt, exacerbating rutting, while low temperatures easily induce thermal shrinkage cracks. Therefore, a linkage layer is added to the fusion logic: when the temperature is high, the weight of temperature-sensitive indicators such as deflection and rutting depth is automatically increased, and vice versa. Historical traffic flow records the cumulative axle load count, which is an important reference for judging pavement fatigue. During maintenance assessments, road sections with high traffic volume are usually the focus of attention.

[0031] During the assessment process, the assessment benchmarks for road wear-related indicators (such as rut ​​depth and crack density) are adjusted based on historical traffic flow data. For road sections with high traffic volume, the allowable deviations in the assessment benchmarks are appropriately relaxed, while those for road sections with low traffic volume are tightened. By introducing external correlated data as an auxiliary fusion layer, cross-system information collaboration can be achieved, making the assessment results more environmentally adaptable and condition-specific.

[0032] According to an embodiment of the present invention, the method further includes: When extreme weather conditions or special operating conditions are detected, switch to emergency assessment mode; in the emergency assessment mode, adjust the confidence weight allocation strategy of each source feature vector to reduce the weight of data sources that are more affected by extreme conditions. Simultaneously lower the trigger threshold of the quality warning threshold to improve the sensitivity of early warning of potential quality risks.

[0033] It should be noted that in this embodiment, extreme weather and special operating conditions use two different processing paths. Image quality deteriorates significantly during heavy rain and snowstorms. Rain and snow obscure the image, and insufficient lighting drastically reduces its reliability. In such cases, the system automatically lowers the weight of the image source while increasing the weight of radar and deflection data, as these are less affected by weather. The emergency mode also includes a threshold reduction action: the warning threshold is temporarily lowered. For example, if the crack density trigger line is 5 lines per 100 square meters in normal mode, it is reduced to 3 lines in an emergency, prioritizing avoiding false alarms over missing risk points. Similar dynamic weight adjustment logic applies to special operating conditions such as road construction areas and major events.

[0034] According to an embodiment of the present invention, the multi-level fusion of the source feature vectors based on a confidence-weighted strategy to obtain a fused feature vector specifically includes: Identify inherent missing patterns in each source data, including image occlusion areas, radar signal attenuation intervals, and sensor sampling blind spots; The inherent missing patterns are used as complementary signals to assign enhanced weights from other data sources to the missing regions during the fusion process; Based on the enhanced weights, cross-source compensation fusion is performed on the source feature vectors to obtain the fused feature vector.

[0035] It's important to note that in this embodiment, each data source has its blind spots. Cameras become blurry when obstructed by water, snow, or tree shadows, creating image blind areas; ground-penetrating radar experiences significant signal attenuation when encountering high-water-content strata or underground metal objects; and sensors themselves have blind spots due to sampling frequency and installation location. Traditionally, these missing data points are simply discarded as noise. However, this approach treats the missing data as complementary signals. If a source lacks data in a certain area, the weights of other sources in that area are automatically increased to compensate. For example, if an image is largely obscured by trees along a road, the radar and sensor data for that section are weighted more heavily, and the cross-source compensation still provides sufficient evaluation data. This approach has been tested in real-world engineering projects, demonstrating a significant improvement in the stability of conclusions in scenarios with incomplete data.

[0036] According to an embodiment of the present invention, the method further includes: Receive user assessment intent instructions, which include at least one of routine inspection assessment, specific disease assessment, and maintenance decision assessment; According to the evaluation intent instruction, the corresponding target evaluation strategy is matched from the preset evaluation strategy library. The target evaluation strategy includes the required data source combination, feature extraction algorithm and fusion weight configuration. The data collection and evaluation process is dynamically orchestrated according to the stated target evaluation strategy.

[0037] It should be noted that in this embodiment, the evaluation intent determines the arrangement of the entire process. Routine inspections emphasize speed and breadth, utilizing all data sources but selecting algorithms with medium precision, since high-frequency inspections cannot run the heaviest calculations every time. Specialized disease assessments target a specific disease, selecting only the data sources most closely related to that disease, allowing for the use of high-precision algorithms. Maintenance decisions require a broader perspective, considering both current conditions and historical trends, thus employing a combination of full-source high-precision assessments and historical comparisons. The strategy library pre-configures templates corresponding to various intents, including three core elements: data source combination, feature algorithm type, and weight parameter configuration. Upon receiving an intent instruction, the corresponding template is automatically applied, and subsequent processes are arranged.

[0038] According to an embodiment of the present invention, acquiring multi-source heterogeneous data of asphalt roads includes: Obtain resource constraint information for the current data acquisition, wherein the resource constraint information includes at least one of the following: number of available sensors, data transmission bandwidth, and computing power; The resource constraint information is converted into a scheduling priority signal, and the collection frequency and accuracy level of each data source are determined based on the scheduling priority signal. The multi-source heterogeneous data is collected according to the specified collection frequency and accuracy level.

[0039] It should be noted that in this embodiment, equipment resources, communication bandwidth, and computing power are often scarce in the field. The number of sensors that can be installed on the vehicle platform is limited, the bandwidth of the mobile network is unstable, and the computing power at the edge is a hard constraint. Faced with these limitations, the scheduling layer has implemented several rules: when there are not enough sensors, priority is given to the channels with the highest contribution; when bandwidth is tight, the transmission frequency of secondary data sources is reduced or the accuracy level is lowered; when computing power is insufficient, a less complex feature extraction algorithm is used, sacrificing some accuracy for real-time performance. The combined effect of these rules is that even with limited hardware conditions, an acceptable level of evaluation quality can be maintained.

[0040] According to an embodiment of the present invention, after the output quality level evaluation result, the method further includes: generating a data acquisition optimization instruction in reverse based on the quality level evaluation result; The data acquisition optimization command is used to adjust the acquisition parameters of each data source, including acquisition location, acquisition density, and acquisition frequency. The data acquisition optimization instructions are fed back to the data acquisition terminal so that subsequent data acquisition processes can focus on covering the identified weak quality areas.

[0041] It should be noted that in this embodiment, the previous approach involved the acquisition end simply collecting data and the evaluation end simply calculating it, with each side operating independently without a feedback loop. The current design adds a reverse guidance mechanism: once a road segment is found to have weak quality indicators in the evaluation, the system immediately generates an acquisition optimization command and sends it to the front end to adjust the acquisition density and frequency for that area. For example, the image acquisition interval for road segments with excessive crack density is reduced from 50 meters to 10 meters, and the radar scan section spacing is increased from 100 meters to 20 meters. This way, data acquisition is no longer blindly covering the entire area but dynamically focusing based on the evaluation results, resulting in a significant improvement in overall efficiency.

[0042] According to an embodiment of the present invention, after determining the final conclusion of the quality indicator through weighted arbitration, the method further includes: The final conclusion is subjected to a consistency check, which includes comparing the final conclusion with the independent evaluation results of each source feature vector. When the final conclusion is inconsistent with more than half of the independent evaluation results, the final conclusion is marked as pending review and a manual review prompt message is generated. When the final conclusion is consistent with the results of each independent evaluation, the final conclusion is marked as credible.

[0043] It should be noted that while weighted arbitration works in most scenarios in this embodiment, there are a few exceptions that warrant extra attention. If a data source has an unreasonably high weight for some reason, the arbitration result will be skewed by this single source's conclusion, even if the other two sources give opposing opinions. To address this, a consistency verification step is added: the final conclusion is compared one by one with each independent evaluation result. If more than half of the independent results do not match the final conclusion, this conclusion is marked as pending review, and a notification containing details of the conflict and the conclusions of each party is generated and sent to a human for confirmation. Only those that pass the verification are marked as reliable. This additional step, while increasing computational overhead, effectively mitigates the risk of deviation in extreme cases.

[0044] Please see Figure 4 The present invention also provides an asphalt road quality assessment system 4 based on multi-data fusion. The system includes a memory 401 and a processor 402. The memory includes a program for an asphalt road quality assessment method based on multi-data fusion. When the processor executes the program for an asphalt road quality assessment method based on multi-data fusion, it performs the following steps: Acquire multi-source heterogeneous data of asphalt roads, including road surface image data, ground-penetrating radar scan data, and road physical parameter data; The multi-source heterogeneous data is preprocessed, including time synchronization, spatial registration, and data standardization. Feature extraction was performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector, and pavement physical feature vector. Based on a credibility-weighted strategy, the source feature vectors are fused at multiple levels to obtain a fused feature vector. Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

[0045] It should be noted that, in this embodiment, in addition to storing the main program of the asphalt road quality assessment method, the memory also undertakes the responsibility of persisting various intermediate data, including the original multi-source heterogeneous data, preprocessing intermediate products, feature vectors of each stage, fusion vectors, threshold parameter tables for grading, and the final assessment conclusions. The processor sequentially drives the four main calculation stages—data preprocessing, feature extraction, multi-level fusion, and quality assessment—according to program instructions. Functionally, the system is divided into five parts: a data acquisition interface, a preprocessing module, a feature extraction module, a fusion calculation module, and an assessment output module. These modules operate collaboratively in a pipelined manner. The memory and processor together constitute a complete integrated hardware and software deployment.

[0046] According to an embodiment of the present invention, the step of evaluating the quality of asphalt roads based on the fused feature vector and outputting a quality grade evaluation result specifically includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality grade assessment results will be revised based on the results of the special testing and evaluation.

[0047] It should be noted that in this embodiment, the processor checks each indicator of the fused feature vector dimension by dimension during quality assessment. Once any indicator is found to have exceeded a pre-stored warning threshold, the corresponding specialized subroutine is immediately invoked for refined analysis, and the original assessment conclusion is revised based on the results of the specialized analysis. This secondary verification is mainly to prevent coarse-grained fusion assessment from giving a judgment with large deviations in certain boundary conditions.

[0048] According to an embodiment of the present invention, the multi-level fusion of the source feature vectors based on a confidence-weighted strategy to obtain a fused feature vector specifically includes: Calculate the confidence weight of each source feature vector separately; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector.

[0049] It should be noted that in this embodiment, the fusion calculation module operates in two steps: first, it calculates the credibility weight distribution for this round based on the historical accuracy performance and current data quality score of each source; then, it checks each indicator for inconsistencies between conclusions from multiple data sources. If a contradiction is found, the weighted arbitration process mentioned earlier is used to determine the final value of the indicator. After these two steps are completed, a consistent fusion vector is output for downstream evaluation.

[0050] The present invention also provides a computer-readable storage medium having a computer program stored thereon, the computer-readable storage medium including a method program for assessing the quality of asphalt roads based on multi-data fusion, wherein when the method program is executed by a processor, it implements the steps of the method program for assessing the quality of asphalt roads based on multi-data fusion as described above.

[0051] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for assessing the quality of asphalt roads based on multi-data fusion, characterized in that, The method includes: S102, acquire multi-source heterogeneous data of asphalt road, the multi-source heterogeneous data including road surface image data, ground penetrating radar scan data and road physical parameter data; S104, preprocess the multi-source heterogeneous data, the preprocessing including time synchronization, spatial registration and data standardization; S106, feature extraction is performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector and pavement physical feature vector; S108, Multi-level fusion of the source feature vectors is performed based on a confidence weighting strategy to obtain a fused feature vector; S110, Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

2. The method for assessing the quality of asphalt roads based on multi-data fusion according to claim 1, characterized in that, The process of assessing the quality of asphalt roads based on the fused feature vector and outputting a quality grade assessment result specifically includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality grade assessment results will be revised based on the results of the special testing and evaluation.

3. The method for assessing the quality of asphalt roads based on multi-data fusion according to claim 1, characterized in that, The multi-level fusion of the source feature vectors based on the credibility weighting strategy to obtain the fused feature vector specifically includes: Calculate the confidence weight of each source feature vector separately; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector.

4. The method for assessing the quality of asphalt roads based on multi-data fusion according to claim 1, characterized in that, Before acquiring the multi-source heterogeneous data of asphalt roads, the method further includes: Acquire external correlation data, including meteorological data and historical traffic flow data for the area where the road is located; The confidence-weighted strategy is used to perform multi-level fusion of the source feature vectors to obtain a fused feature vector, including: The external correlation data is used as an auxiliary fusion layer and jointly fused with the source feature vectors to obtain the fused feature vector. The meteorological data is used to correct the weighting of temperature-sensitive physical parameters, and the historical traffic flow data is used to correct the evaluation benchmark for road wear-related indicators.

5. The method for assessing the quality of asphalt roads based on multi-data fusion according to claim 2, characterized in that, The method further includes: When extreme weather conditions or special operating conditions are detected, switch to emergency assessment mode; In the emergency assessment mode, the confidence weight allocation strategy of each source feature vector is adjusted to reduce the weight of data sources that are more affected by extreme conditions. Simultaneously lower the trigger threshold of the quality warning threshold to improve the sensitivity of early warning of potential quality risks.

6. The method for assessing the quality of asphalt roads based on multi-data fusion according to claim 1, characterized in that, The multi-level fusion of the source feature vectors based on the credibility weighting strategy to obtain the fused feature vector specifically includes: Identify inherent missing patterns in each source data, including image occlusion areas, radar signal attenuation intervals, and sensor sampling blind spots; The inherent missing patterns are used as complementary signals to assign enhanced weights from other data sources to the missing regions during the fusion process; Based on the enhanced weights, cross-source compensation fusion is performed on the source feature vectors to obtain the fused feature vector.

7. An asphalt road quality assessment system based on multi-data fusion, characterized in that, The system includes a memory and a processor. The memory includes a program for an asphalt road quality assessment method based on multi-data fusion. When the processor executes the program for an asphalt road quality assessment method based on multi-data fusion, it performs the following steps: Acquire multi-source heterogeneous data of asphalt roads, including road surface image data, ground-penetrating radar scan data, and road physical parameter data; The multi-source heterogeneous data is preprocessed, including time synchronization, spatial registration, and data standardization. Feature extraction was performed on each of the preprocessed source data to obtain the pavement surface feature vector, pavement internal structure feature vector, and pavement physical feature vector. Based on a credibility-weighted strategy, the source feature vectors are fused at multiple levels to obtain a fused feature vector. Asphalt road quality is assessed based on the fused feature vector, and the quality level assessment result is output.

8. The asphalt road quality assessment system based on multi-data fusion according to claim 7, characterized in that, The process of assessing the quality of asphalt roads based on the fused feature vector and outputting a quality grade assessment result specifically includes: The fused feature vector is compared with a preset quality grading threshold. When any dimension indicator in the fused feature vector exceeds the corresponding quality warning threshold, the dimension indicator is marked as an abnormal indicator, and a special detection and evaluation process for the abnormal indicator is triggered. The quality grade assessment results will be revised based on the results of the special testing and evaluation.

9. The asphalt road quality assessment system based on multi-data fusion according to claim 7, characterized in that, The multi-level fusion of the source feature vectors based on the credibility weighting strategy to obtain the fused feature vector specifically includes: Calculate the confidence weight of each source feature vector separately; When at least two source feature vectors output contradictory conclusions for the same quality indicator, cross-validation is performed based on the credibility weights of each source, and the final conclusion for the quality indicator is determined through weighted arbitration. The final conclusion is used as the corresponding dimension value of the fused feature vector.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer-readable storage medium includes a program for an asphalt road quality assessment method based on multi-data fusion. When the program is executed by a processor, it implements the steps of the asphalt road quality assessment method based on multi-data fusion as described in any one of claims 1 to 6.

Images