An intelligent control system for the whole process of asphalt mixture construction

Abstract

The application provides an intelligent control system for the whole process of asphalt mixture construction, and belongs to the technical field of intelligent control, which comprises process set acquisition, fine determination, optimal search, sequence construction and construction improvement modules. The construction site map is first analyzed to match the construction process set, the construction direction is monitored in real time, and the fine construction set is collected. The construction difference set is optimized based on the ant colony algorithm to predict the construction effectiveness. The construction deviation generation element sequence is extracted, and finally the improved sub-process is iteratively generated in combination with the simulation model to realize the intelligent control of the whole construction process. The application improves the construction control accuracy and construction efficiency, reduces the construction rework rate and cost, and ensures the standardization of the asphalt mixture construction quality.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to an intelligent control system for the entire process of asphalt mixture construction. Background Technology

[0002] Asphalt mixture construction is a core process in road engineering construction and maintenance. Construction quality directly determines the road's performance and service life; therefore, precise control throughout the entire construction process is crucial. Currently, asphalt mixture construction control generally relies on manual methods combined with local parameter monitoring. This approach only allows for post-construction verification of single parameters such as construction direction and paving temperature, failing to provide real-time prediction of parameter changes and construction results at each stage. Furthermore, it cannot dynamically optimize subsequent construction processes based on parameter differences during construction. Consequently, if parameter deviations occur, construction strategies cannot be adjusted promptly, easily leading to construction results that do not meet standard requirements. This not only increases rework rates but also raises construction costs, making it difficult to guarantee the quality stability of the entire asphalt mixture construction process. Additionally, the lack of reliance on industrial internet platforms for cloud-based data aggregation, cross-device interoperability, and end-to-end collaborative control hinders the use of industrial internet platforms for remote and intensive construction management.

[0003] Therefore, this invention proposes an intelligent control system for the entire construction process of asphalt mixtures. Summary of the Invention

[0004] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, in order to solve the technical problems mentioned above.

[0005] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, comprising: The process set acquisition module is used to acquire the construction site map of the area to be constructed, and obtain the corresponding construction process set according to the standard construction requirements of each connected area in the construction site map; The precision determination module is used to monitor the actual construction direction of each stage during the initial construction period when construction is carried out according to the construction process set, and compare it with the standard construction direction of the standard construction process of the corresponding stage. If the operation direction is inconsistent, it will remind to stop construction. If the operation direction is consistent, obtain the detailed construction set for each actual construction moment between the first end time of the initial construction period and the second end time of the first designated construction anchor point in the corresponding stage. The detailed construction set includes construction environment parameters, asphalt mixture construction parameters, construction coverage area and coverage effect map of the construction coverage area. The optimal search module, based on the ant colony algorithm, performs optimal search on the fine difference set at each adjacent time point between the first end time of the initial construction period and the second end time of the first designated construction anchor point, the first difference set between the fine construction set at the next adjacent time point and the fine construction set at the second end time point, and the second difference set between the fine construction set at the next adjacent time point and the standard construction set at the first designated construction anchor point. The optimal difference set is then input into a pre-built predictive analysis model for the same stage to obtain the predicted performance map after the completion of the corresponding stage. The fine difference set is the set of parameter differences of the fine construction sets at two adjacent actual construction times. The sequence construction module is used to mark the element of the next time step of the corresponding adjacent time step as 0 if the similarity between the predicted effect map and the final effect map of the corresponding standard construction process is greater than a preset similarity threshold. Otherwise, extract the difference effect set based on the final effect map of the corresponding stage from the predicted effect map, use it as a new element for the next time moment, and generate an element sequence from the first end time of the initial construction period to the second end time of the first specified construction anchor point. The construction improvement module is used to generate an improved sub-process for the next specified construction anchor point based on the element sequence, combined with the standard sub-process from the first specified construction anchor point to the next specified construction anchor point and the standard as-built drawing of the next specified construction anchor point, and to perform construction control on the corresponding connected area according to the improved sub-process until the construction is completed.

[0006] Preferably, the process set acquisition module includes: The vector analysis submodule is used to perform grayscale processing and edge extraction on the acquired construction site map, and analyze the regional attribute vector of each connected region. The regional attribute vector contains several construction properties of the corresponding connected region and the confidence level of each construction property. The matching submodule is required to input the regional attribute vector into a pre-built vector analysis model to obtain the standard construction requirements of the corresponding connected region, and to match the construction process set of the corresponding connected region from the preset requirements-process database.

[0007] Preferably, the vector analysis submodule includes: The grayscale and edge processing unit is used to perform grayscale processing and edge extraction on the construction site map to obtain a first closed edge region and a first unclosed edge region. The complete coordinates and feature data of the first closed edge region are retained. The remaining image region after removing the first closed edge region is magnified and contrast enhanced once. The image after magnification and enhancement is further processed with grayscale and edge extraction to obtain a second closed edge region and a second unclosed edge region. The feature extraction and annotation unit is used to extract the edge grayscale features of the second unclosed edge region and annotate them on the edge line of the matched region; The initial feature acquisition unit is used to obtain the initial common features of each edge point on the edge line of the region based on the first nearest distance point set between the region edge line and the boundary line of the first closed edge region, and the second nearest distance point set between the region edge line and the boundary line of the second closed edge region. The feature pair generation unit is used to perform normal probability analysis on all the initial common features of the edge line of the region to obtain the common gray-scale features of the corresponding edge line of the region, and to generate feature pairs by combining the edge gray-scale features of the corresponding edge line of the region, wherein the edge gray-scale features are the breakpoint extension trend features and contour smoothness features of the corresponding edge line of the region. A global search unit is used to perform a global search from the first nearest distance point set and the second nearest distance point set according to the features, to obtain a first point that matches the initial common features of the first point of the region edge line and a second point that matches the initial common features of the last point of the region edge line. The first point and the second point are located on the same continuous boundary line of the original closed edge region. The matching degree of the initial common features is calculated based on the feature cosine similarity. Points with a matching degree of the initial common features greater than a preset matching threshold are regarded as matching points. From the matching points, the point with the largest matching degree with the first point is regarded as the first point and the point with the largest matching degree with the last point is regarded as the second point. Connect the first point to the first point, connect the last point to the second point, and incorporate the continuous boundary line segments of the original closed edge region between the first point and the second point into the outline to form a new closed edge region containing the edge lines of the corresponding region, wherein each closed edge region is a construction connected region; The alignment processing unit is used to convert the regional construction drawing of each connected region into R-channel, G-channel, and B-channel respectively, extract the feature set of each channel, and perform alignment processing on the feature set of each channel to obtain several intersection features and several non-intersection features. Among them, the overlapping features of two or more channels are regarded as intersection features. The confidence level determination unit is used to match the construction properties corresponding to each feature from a preset attribute-property lookup table; When the feature is an intersection feature, the confidence level of the corresponding construction nature is obtained based on the construction nature, feature shape and number of overlapping channels of the intersection feature; When the features are non-intersecting, the corresponding confidence level is matched from the historical database based on the construction nature of the non-intersecting features.

[0008] Preferably, the initial feature acquisition unit includes: The sub-unit is used to acquire the first grayscale representation of each first nearest distance line in the first nearest distance point set, and the first auxiliary grayscale representation of a first circle constructed with the first nearest distance as the diameter and the point on the boundary line of the first nearest distance line located on the first closed edge region as the center. Simultaneously, the second grayscale representation of each second nearest distance line in the second nearest distance point set is obtained, as well as the second auxiliary grayscale representation of a second circle constructed with the second nearest distance as the diameter and the point on the boundary line of the second nearest distance line located in the second closed edge region as the center; The relationship determination subunit is used to determine the positional relationship between the first circle and the second circle corresponding to each edge point on the edge line of the region. When the positional relationship is an intersection relationship, the intersection representation of the first grayscale representation and the second grayscale representation, as well as the grayscale representation of the intersection region, are extracted as the initial common features of the corresponding edge points; When the positional relationship is non-intersecting, the intersection of the first grayscale representation and the second grayscale representation, as well as the intersection of the first auxiliary grayscale representation and the second auxiliary grayscale representation, are extracted as the initial common features of the corresponding edge points.

[0009] Preferably, the optimal search module includes: The normalization submodule is used to normalize the construction feature dimension of the fine difference set, the first difference set, and the second difference set at each adjacent time between the first end time of the initial construction time period and the second end time of the first designated construction anchor point, so as to obtain a standardized difference set. The initial position mapping submodule is used to set the number of ant colony individuals according to the number of construction feature dimensions of the standardized difference set, map the initial position of each ant colony individual to the construction feature dimension node of the standardized difference set, and set the initial pheromone concentration between construction feature dimension nodes based on the feature correlation degree of asphalt mixture construction parameters. The construction feature dimension node is the node corresponding to a single construction feature dimension of the standardized difference set, and each node corresponds to a construction feature parameter. The path search submodule is used by each ant colony individual to conduct path search between nodes in the construction feature dimension with the core search guide of minimizing the prediction deviation of construction effectiveness. During the search process, the node transfer probability is dynamically corrected by combining the feature similarity of the fine construction set of the next time step corresponding to the adjacent time step. The update submodule is used to collect the difference sets corresponding to the search paths of all ant colony individuals to generate several candidate sets, and to perform local pheromone updates and global pheromone updates on the search paths of individual ant colonies to obtain the optimal difference set.

[0010] Preferably, the local pheromone update is performed immediately after an individual ant completes a single node transfer, reducing the pheromone concentration of the corresponding path node according to a preset basic evaporation coefficient; the global pheromone update is performed after all ant individuals complete a complete path search, releasing pheromone increments only for the search paths corresponding to the difference set within the candidate set, and the release increment value is negatively correlated with the objective function value of the construction effectiveness prediction deviation of the corresponding path; the objective function of the construction effectiveness prediction deviation is the reciprocal of the feature similarity between the predicted effectiveness map and the final effectiveness effect map of the corresponding stage, and the smaller the objective function value, the smaller the construction effectiveness prediction deviation.

[0011] Preferably, the improved sub-process includes the correction direction for the already constructed content at the previous specified construction anchor point and the improvement direction for the content to be constructed at the current specified construction anchor point. Both the correction direction and the improvement direction include adjustments to three dimensions: construction parameters, construction direction, and construction coverage threshold.

[0012] Preferably, the construction improvement module includes: The initial correction submodule is used to map the construction deviation features of the element sequence to each construction procedure of the standard sub-process from the first specified construction anchor point to the next specified construction anchor point. Combined with the construction effectiveness feature requirements of the standard as-built drawing of the next specified construction anchor point, the initial correction is performed on the construction parameters, construction direction, and construction coverage threshold of the corresponding procedures in the standard sub-process to generate the initial sub-process of the next specified construction anchor point. The construction deviation features are the construction parameter deviations and construction effectiveness deviation features contained in the difference effect set in the element sequence. The simulation iterative optimization submodule is used to input the initial subprocess into a pre-built asphalt mixture construction simulation model, incorporate real-time construction environment parameters to simulate the entire construction process, generate a simulation completion diagram, and calculate the feature similarity between the simulation completion diagram and the standard completion diagram of the next specified construction anchor point. If the feature similarity does not reach the preset threshold, the deviation between the simulation completion drawing and the standard completion drawing is analyzed for process traceability. For the deviation-related processes obtained from the traceability, the corresponding process correction parameters of the initial sub-process are adjusted in a targeted manner until the feature similarity between the simulation completion drawing and the standard completion drawing meets the preset threshold. The finally adjusted sub-process is determined as the improved sub-process for the next specified construction anchor point.

[0013] Compared with the prior art, the beneficial effects of this application are as follows: Through the collaborative operation of the process set acquisition module, fine-grained determination module, optimal search module, sequence construction module, and construction improvement module, the system achieves intelligent and refined management and control of the entire asphalt mixture construction process. Based on the construction site map, the system automatically matches the construction process of each connected area, monitors the construction direction and core parameters in real time, optimizes construction differences and accurately predicts construction results through ant colony optimization, quickly locates construction deviations, and generates targeted improvement sub-processes. This effectively reduces construction errors caused by reliance on human experience, improves construction quality and efficiency, ensures that the entire asphalt mixture construction process meets standard requirements, reduces rework rates and costs, and achieves closed-loop optimization of construction management and control. Furthermore, this system can be integrated with an industrial internet platform to achieve real-time cloud-based construction data upload, remote monitoring, and cross-terminal collaborative scheduling. Combined with the digital support of the industrial internet platform, it further realizes visualization and standardized closed-loop management and control of the entire construction process.

[0014] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of an intelligent control system for the entire construction process of asphalt mixtures, as described in an embodiment of the present invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, such as... Figure 1 As shown, it includes: The process set acquisition module is used to acquire the construction site map of the area to be constructed, and obtain the corresponding construction process set according to the standard construction requirements of each connected area in the construction site map; In this embodiment, the construction site map is an image data collection method using drones, vehicle-mounted cameras, fixed monitoring equipment, etc., which includes information such as the area of ​​asphalt pavement to be constructed, construction equipment, and construction environment. For example, in the maintenance and construction of a highway, a drone equipped with a high-definition zoom lens is used to take aerial photos of a 2-kilometer construction section at a flight speed of 5 meters per second, and collect a construction site map with a resolution of 4K. The map clearly shows the area of ​​asphalt surface layer to be milled, the parked milling machine equipment, the distribution of on-site construction personnel, and the surrounding road environment. At the same time, a vehicle-mounted camera is installed on the paver to collect the construction site map of the paving operation surface in real time.

[0019] A connected region is a spatially continuous asphalt pavement area with the same construction attributes in a construction site map. It is the basic unit for dividing construction units and matching corresponding construction processes. For example, in the asphalt repair construction of a municipal road, three adjacent potholes distributed in patches on the pavement are identified as a connected region by the system because they have no obvious spatial boundaries and all require the same construction process of hot-mix asphalt filling and compaction. On the other hand, an independent longitudinal crack zone on the pavement that is not spatially connected to the potholes is identified as another independent connected region because its construction process is cut-and-fill with adhesive, which is different from the pothole construction process.

[0020] In this embodiment, the standard construction requirements are the normative requirements for each process and parameter of asphalt mixture construction formulated by the state, industry, and project, including construction process steps, material specifications, construction parameter ranges, quality acceptance standards, etc. For example, for the paving process of SBS modified asphalt mixture, the standard construction requirements include: paving temperature not lower than 160℃, paving speed controlled at 2-6 meters / minute, and paving layer thickness deviation not exceeding ±5mm; for the asphalt mixture compaction process, the standard construction requirements include: initial compaction temperature not lower than 150℃, secondary compaction using a steel wheel roller of 20 tons or more, and final compaction temperature not lower than 90℃.

[0021] In this embodiment, the construction process set includes a complete set of construction processes for each connected area of ​​the area to be constructed. Specifically, it covers the standard construction process for several stages of the entire construction process, the final effect diagram of each stage, the designated construction anchor point for each stage, and the standard sub-process between adjacent designated construction anchor points. For example, the construction process for asphalt overlay on a certain road section includes three stages: The first stage is road milling, and the corresponding final effect diagram is a picture of a smooth road surface with a depth that meets the design requirements after milling. The designated construction anchor points are the starting end and the ending end of milling. The standard sub-process is: milling machine advances at a constant speed → depth detection → edge and corner treatment; The second stage is tack coat spraying, and the corresponding final effect diagram is a picture of a road surface evenly covered with tack coat, with no missed spraying or accumulation. The designated construction anchor points are the starting point and the ending point of spraying. The standard sub-process is: oil tank supply → uniform spraying from nozzles → flow monitoring; The third stage is asphalt paving, and the corresponding final effect diagram is a picture of a smooth and uniformly thick road surface. The designated construction anchor points are the starting point, the middle point, and the ending point of paving. The standard sub-process is: paver positioning → uniform paving → thickness adjustment.

[0022] The precision determination module is used to monitor the actual construction direction of each stage during the initial construction period when construction is carried out according to the construction process set, and compare it with the standard construction direction of the standard construction process of the corresponding stage. If the operation direction is inconsistent, it will remind to stop construction. If the operation direction is consistent, obtain the detailed construction set for each actual construction moment between the first end time of the initial construction period and the second end time of the first designated construction anchor point in the corresponding stage. The detailed construction set includes construction environment parameters, asphalt mixture construction parameters, construction coverage area and coverage effect map of the construction coverage area. In this embodiment, the designated construction anchor points are pre-set key locations within the construction process, used to divide construction stages, define construction nodes, compare construction directions, and generate sub-processes. These are typically the start and end points of a construction section, or key segmentation points. For example, for a 1-kilometer-long asphalt paving section, the designated construction anchor points are the 0-meter starting anchor point, the 500-meter midpoint anchor point, and the 1000-meter ending anchor point. In single-pothole repair construction, the designated construction anchor points are the pothole edge starting anchor point and the pothole edge ending anchor point.

[0023] In this embodiment, the standard construction direction corresponds to the construction process concentration, and the standard construction movement direction of each connected area in each construction stage, including the direction of advancement of construction equipment, the sequence of construction operations, etc. For example, for road milling construction, the standard construction direction is to advance unidirectionally from one end of the construction section to the other; for road paving construction, the standard construction direction is to advance unidirectionally from low to high along the longitudinal direction of the road, or to pave evenly along the transverse direction of the road.

[0024] In this embodiment, the actual construction direction is determined by analyzing images of the construction site captured by cameras during the construction process. It refers to the actual movement direction and operation direction of the construction equipment at each construction stage. For example, in the asphalt paving construction of a certain road section, the system analyzes images captured by the vehicle-mounted camera to determine that the actual construction direction of the paver is to move laterally from road section K0+000 to K0+500. If this direction is consistent with the standard construction direction (lateral movement from the shoulder to the center of the road), the construction direction is deemed compliant. If the actual construction direction is to move from road section K0+500 to K0+000, it is inconsistent with the standard direction, and the system triggers a stop construction reminder.

[0025] The optimal search module, based on the ant colony algorithm, performs optimal search on the fine difference set at each adjacent time point between the first end time of the initial construction period and the second end time of the first designated construction anchor point, the first difference set between the fine construction set at the next adjacent time point and the fine construction set at the second end time point, and the second difference set between the fine construction set at the next adjacent time point and the standard construction set at the first designated construction anchor point. The optimal difference set is then input into a pre-built predictive analysis model for the same stage to obtain the predicted performance map after the completion of the corresponding stage. The fine difference set is the set of parameter differences of the fine construction sets at two adjacent actual construction times. In this embodiment, the initial construction time period is a pre-set initial time range for each construction stage. It is the core time period for the system to monitor construction parameters and collect construction data, typically the time from when the construction equipment enters the area between the corresponding construction anchor points until the construction in that area is completed. For example, for asphalt paving construction on a 500-meter road section, the initial construction time period is from 9:00 AM to 10:30 AM, during which the paver enters the road section and completes the paving operation.

[0026] The first end time is the first critical time node after the start of the construction phase within the initial construction period. It is usually the designated time point after the construction equipment enters the corresponding construction area. For example, in the initial construction period of asphalt paving (9:00-10:30), the first end time is set at 9:10, which is the time node 10 minutes after the construction equipment enters the paving area.

[0027] The second end time is the end time node corresponding to the first designated construction anchor point within the initial construction period. That is, the time it takes for the construction equipment to complete the construction of the area between the initial construction start point and the first designated construction anchor point. For example, in a 500-meter paving section, the first designated construction anchor point is at 200 meters. If the construction equipment advances at a speed of 2 meters / minute and completes the paving of the 200-meter area at 9:20, then the second end time is set to 9:20. The system collects the construction parameters between 9:10 (first end time) and 9:20 (second end time) to form a detailed construction set for this period.

[0028] The detailed construction set is a complete set of parameters collected at each actual construction moment during the initial construction period, including construction environment parameters, asphalt mixture construction parameters, construction coverage, and coverage effect maps. These parameters include: construction environment parameters such as ambient temperature, ambient humidity, and wind speed; asphalt mixture construction parameters such as paving temperature, paving speed, paving thickness, compaction temperature, number of compaction passes, compaction speed, mixture gradation, asphalt-aggregate ratio, and mixture moisture content; and construction coverage parameters such as coverage area, coverage rate, and thickness deviation. All dimensions have been normalized to the 0-1 range to eliminate dimensional differences and are completely consistent with the feature vector dimensions of the standard construction set.

[0029] In this embodiment, the construction environment parameters are external environmental data that affect the construction quality of asphalt mixtures during the construction process, including ambient temperature, air humidity, wind speed, rainfall, and light intensity. For example, in a rainy construction scenario, if the rainfall in the construction environment parameters is 5 mm / hour, the system will detect that this parameter exceeds the standard construction requirements (construction is not allowed if the rainfall is greater than 2 mm / hour), trigger an alert, and suspend construction. In a sunny construction scenario, if the ambient temperature is 30℃ and the air humidity is 35%, the system will determine that the environmental parameters meet the construction requirements, and construction can continue.

[0030] Construction parameters for asphalt mixtures are the core operational parameters for each step of the asphalt mixture construction process. These include paving temperature, paving speed, compaction temperature, number of compaction passes, mixture gradation, paving thickness, and spraying flow rate. For example, in the asphalt paving process, the construction parameters for asphalt mixtures include: paving temperature 162℃, paving speed 4 m / min, paving thickness 40 mm, and SBS modified asphalt content of 4.5%. In the compaction process, the construction parameters include: initial compaction temperature 152℃, secondary compaction using a 22-ton steel wheel roller, 3 secondary compaction passes, and final compaction temperature 92℃.

[0031] In this embodiment, the construction coverage area is the road surface area to be constructed or already constructed covered by the construction equipment at the current construction time. For example, during the operation of a paver, the construction coverage area is the road surface area of ​​10 meters (length) × 15 meters (width) = 150 square meters covered by the current paver wheel; during the operation of a milling machine, the construction coverage area is the road surface area of ​​5 meters × 8 meters = 40 square meters covered by the milling head.

[0032] The coverage effect map is a real-time image captured to reflect the road surface construction effect within the construction coverage area at the current construction moment. It includes visual features such as road surface smoothness, mixture distribution, compaction, and coverage uniformity. For example, at 9:15 during asphalt paving construction, the coverage effect map is a real-time image captured of the paving operation surface. The image shows that the paved road surface has a uniform mixture distribution, no material shortage, no accumulation, and good smoothness. The effect map is divided into 16×16 sub-regions, and four features are extracted from each sub-region: smoothness, compaction, mixture distribution uniformity, and grayscale mean. A 16×16×4=1024-dimensional feature vector is constructed, and then a similarity comparison is performed to determine whether the construction effect at that time period meets the standard.

[0033] The fine difference set is the set of all parameter differences between the corresponding fine construction sets at two adjacent actual construction times. This includes differences in construction environment parameters, asphalt mixture construction parameters, and construction coverage differences. For example, taking two adjacent times, 9:10 and 9:15, the fine construction set parameters for 9:10 are: paving temperature 165℃, paving speed 3 m / min, and coverage 150 square meters. The fine construction set parameters for 9:15 are: paving temperature 163℃, paving speed 3.5 m / min, and coverage 180 square meters. Then the fine difference set is: paving temperature difference -2℃, paving speed difference +0.5 m / min, and coverage difference +30 square meters. It should be noted that the acquisition principles of the first and second difference sets are similar to those of the fine difference set, and will not be elaborated here.

[0034] In this embodiment, the ant colony algorithm is an intelligent optimization algorithm that simulates the foraging behavior of ants. By simulating the mechanism of ants releasing pheromones on the path and ants choosing paths based on pheromone concentration, it achieves the optimal solution to complex problems. In this invention, it is used to search for the optimal difference set among multiple sets of construction parameter differences.

[0035] The optimal difference set is obtained through optimization search using the ant colony algorithm. It is the set of construction parameter differences that minimizes the deviation in construction effectiveness and best meets the standard construction requirements. For example, if the system searches for three combinations of difference sets using the ant colony algorithm, where the construction effectiveness deviation of combination A is 5%, combination B is 8%, and combination C is 3%, then the difference set of combination C is the optimal difference set. The system inputs this difference set into the same-stage predictive analysis model to predict the effectiveness of subsequent construction.

[0036] In this embodiment, the same-stage prediction and analysis model is based on a residual convolutional neural network (ResNet18) architecture, combined with fully connected layers to map parameter differences to construction effect diagrams, as detailed below: Training dataset: 1000 sets of historical data on the construction of similar asphalt mixtures were collected, including a set of differences in construction parameters (input) and corresponding pavement effect maps after construction (output), which are divided into training set and test set in a 7:3 ratio; the parameter difference set includes 15 dimensions of difference features such as construction environment, mixture construction, and overlay; the pavement effect map is a grayscale image with a resolution of 512×512, and the labeled features include smoothness, compaction degree, and uniformity of mixture distribution. Training hyperparameters: Retain the basic ResNet18 network structure, remove the original classification layer, and add 3 fully connected layers with the number of neurons being 256, 128, and 512×512 respectively; the learning rate is set to 0.0005, the number of iterations is 300, the loss function is the mean squared error loss, the optimizer is the SGD optimizer, and the momentum is set to 0.9. Feature mapping rule: Input the standardized feature vector of the best difference set into the model, extract deep features through the convolutional layer, and then map the features into a 512×512 pixel matrix through the fully connected layer. After grayscale processing, the prediction result map is generated. Model validation: When the feature similarity between the predicted performance map generated by the test set and the actual performance map is ≥90%, the model training is complete and it can be put into practical application.

[0037] The sequence construction module is used to mark the element of the next time step of the corresponding adjacent time step as 0 if the similarity between the predicted effect map and the final effect map of the corresponding standard construction process is greater than a preset similarity threshold. Otherwise, extract the difference effect set based on the final effect map of the corresponding stage from the predicted effect map, use it as a new element for the next time moment, and generate an element sequence from the first end time of the initial construction period to the second end time of the first specified construction anchor point. The preset similarity threshold is a pre-set numerical threshold used to determine whether the predicted effect image matches the final effect image of the stage. It is usually set in the range of 0.8-0.95 (corresponding to the percentage of image feature similarity). In this embodiment, based on 1000 sets of historical data of asphalt mixture construction, the control variable method was used to conduct experiments to verify that when the similarity is ≥0.90, the construction quality meets the acceptance standard of the technical specification for highway asphalt pavement construction and the rework rate is ≤5%. Therefore, it is determined to be 0.9.

[0038] The element sequence contains an ordered set of elements corresponding to each actual construction time from the first end time of the initial construction period to the second end time of the first designated construction anchor point, such as [0, difference effect set, 0].

[0039] The difference effect set is a set of construction deviation features extracted from the predicted effect map when the similarity between the predicted effect map and the final effect map of the stage is less than a preset threshold. It includes deviations in construction parameters, construction coverage, and construction results. For example, at 9:15, the similarity between the predicted effect map and the final effect map of the stage is 0.88. The difference effect set extracted by the system includes: paving temperature deviation -3℃, paving speed deviation +0.6 m / min, construction coverage deviation +20 square meters, and road surface smoothness deviation +1 mm / m. This set clearly indicates the specific deviations between the current construction parameters and the standard requirements, providing a precise direction for subsequent construction improvements.

[0040] The construction improvement module is used to generate an improved sub-process for the next specified construction anchor point based on the element sequence, combined with the standard sub-process from the first specified construction anchor point to the next specified construction anchor point and the standard as-built drawing of the next specified construction anchor point, and to perform construction control on the corresponding connected area according to the improved sub-process until the construction is completed.

[0041] A standard sub-process is a collection of standardized construction steps between two adjacent designated construction anchor points. It serves as the basic template for the system to generate improved sub-processes. It includes the construction sequence, parameter requirements, and quality control points for each process. For example, in the paving construction between the first designated construction anchor point (200 meters) and the next designated construction anchor point (500 meters), the standard sub-process is: paver positioning → adjusting paving thickness → uniform paving → real-time temperature monitoring → initial compaction → secondary compaction → final compaction. Each process corresponds to specific parameter requirements (such as paving speed 3-4 meters / minute, initial compaction temperature ≥150℃). Based on this template and considering the deviations in the element sequence, the system generates improved sub-processes.

[0042] The standard as-built drawing is a pre-made drawing that shows the pavement condition after the next designated construction anchor point is completed, meeting the standard construction requirements. For example, for the designated construction anchor point at 500 meters, the standard as-built drawing is an ideal drawing showing that the pavement smoothness deviation is ≤3mm / meter, the compaction degree is ≥96%, the mixture is evenly distributed, and there are no cracks or ruts.

[0043] The improved sub-process is generated by the construction improvement module based on the element sequence, standard sub-process, and standard as-built drawing. It is a set of optimized construction steps for the area to be constructed at the next specified construction anchor point. It includes the correction direction for the already constructed content at the previous anchor point and the improvement direction for the content to be constructed at the current anchor point. It is used to guide on-site construction and reduce deviations. For example, based on the element sequence [0, difference effect set, 0], the improved sub-process generated by the system includes: the correction direction is: increase the paving temperature by 2℃ and decrease the paving speed by 0.5 m / min at 9:15, and re-pave the corresponding coverage deviation area; the improvement direction is: in subsequent paving construction, control the paving temperature at 165±2℃ and the paving speed at 3±0.2 m / min, monitor the flatness in real time, and ensure that the deviation is ≤3mm / m; on-site construction personnel operate according to this improved sub-process to ensure that subsequent construction meets the standard requirements.

[0044] In this embodiment, each stage corresponds to a standard construction process, and each stage includes at least two designated construction anchor points. Each designated construction anchor point corresponds to a standard as-built drawing. The construction from one anchor point to the next in adjacent anchor points is based on a pre-set standard sub-process, meaning a standard construction process contains several standard sub-processes. Each stage corresponds to a final result drawing, which is a pre-set ideal effect drawing. The final result drawing for each stage is pre-made and represents the ideal road surface condition that meets the standard construction requirements after each stage of the construction process is completed. It serves as the reference standard for the system to subsequently compare and predict construction effectiveness and determine whether the construction meets the standards. For example, in the asphalt mixture compaction stage, the final result drawing for the stage is a pre-set image showing that the road surface compaction degree reaches over 96%, the surface is free of wheel tracks, and the smoothness deviation is less than 3mm / meter. This image is created by collecting images of compacted road surfaces that meet the standards and includes visual features such as road texture, thickness, and smoothness. The system will subsequently compare the actual road surface image formed during construction with this image to determine whether the construction effectiveness meets the standards.

[0045] Preferably, the improved sub-process includes the correction direction for the already constructed content at the previous specified construction anchor point and the improvement direction for the content to be constructed at the current specified construction anchor point. Both the correction direction and the improvement direction include adjustments to three dimensions: construction parameters, construction direction, and construction coverage threshold.

[0046] In this embodiment, for different asphalt construction scenarios (such as highways, municipal roads, and rural roads), all the thresholds involved in this invention can be finely adjusted by ±0.05, with the upper limit for highways and the lower limit for rural roads.

[0047] The beneficial effects of the above technical solution are as follows: Through the coordinated operation of the process set acquisition module, the fine determination module, the optimal search module, the sequence construction module, and the construction improvement module, intelligent and refined management and control of the entire asphalt mixture construction process is realized; based on the construction site map, the system automatically matches the construction process of each connected area, monitors the construction direction and core parameters in real time, optimizes construction differences and accurately predicts construction results through the ant colony algorithm, quickly locates construction deviations and generates targeted improvement sub-processes, effectively reduces construction errors caused by reliance on human experience, improves construction quality and efficiency, ensures that the entire asphalt mixture construction process meets standard requirements, reduces construction rework rate and cost, and realizes closed-loop optimization of construction management and control.

[0048] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, wherein the process set acquisition module includes: The vector analysis submodule is used to perform grayscale processing and edge extraction on the acquired construction site map, and analyze the regional attribute vector of each connected region. The regional attribute vector contains several construction properties of the corresponding connected region and the confidence level of each construction property. In this embodiment, grayscale processing is an image processing operation that converts a color construction site image into an image containing only black, white, and gray brightness levels. For example, a construction site image containing a colored road surface, construction equipment, and surrounding greenery will be converted into an image composed only of black, white, and gray pixels after grayscale processing. The originally colorful scene is simplified to a contrast of brightness differences, and the light and dark boundaries between pothole areas and normal road surfaces will be more prominent. This is an existing technology.

[0049] The Canny edge detection algorithm is used to first perform Gaussian filtering on the grayscale image to eliminate noise interference, then calculate the gradient magnitude and direction of the image, refine the edge lines by non-maximum suppression, and finally filter out continuous effective edges by double threshold detection to obtain the edge feature results of the construction site image, which belongs to the existing technology.

[0050] A region attribute vector is a structured data vector that describes the construction-related characteristics of a single connected region. For example, for a pothole connected region on a municipal road, its region attribute vector can be represented as structured data containing (pothole repair, 0.92), (small area defects, 0.88), and (hot mix asphalt construction adaptation, 0.95).

[0051] Construction nature is a qualitative description of the construction-related characteristics of a connected area. It is used to characterize core information such as the type of defects, the adaptability of construction technology, and the scale of construction in the area. For example, construction nature can include categories such as pothole repair, crack filling, large-area overlay, hot-mix asphalt compatibility, and cold-mix asphalt compatibility.

[0052] Confidence level is a numerical value that quantifies the degree of matching between a certain construction property and the corresponding connected region, and its value ranges from 0 to 1.

[0053] The matching submodule is required to input the regional attribute vector into a pre-built vector analysis model to obtain the standard construction requirements of the corresponding connected region, and to match the construction process set of the corresponding connected region from the preset requirements-process database.

[0054] In this embodiment, the vector analysis model adopts a lightweight convolutional neural network + random forest fusion architecture. The model training and application details are as follows: Training dataset: 1000 sets of construction site images and corresponding annotation data for different asphalt construction scenarios (road maintenance, pothole repair, overlay construction, etc.) were collected, of which 800 sets were the training set and 200 sets were the test set; the sample feature dimensions include 28 dimensions such as grayscale features, texture features, and contour features of connected regions, and the annotation data are the standard construction requirements for the corresponding connected regions (such as hot-mix asphalt repair of potholes, longitudinal crack injection, etc.). Training hyperparameters: 3 convolutional layers with kernel sizes of 3×3, 5×5, and 3×3 respectively; 2×2 max pooling layers; 2 fully connected layers with 128 and 64 neurons respectively; 100 random forest decision trees with a depth of 10; a learning rate of 0.001; 200 iterations; cross-entropy loss; and the Adam optimizer. Training and validation methods: After normalizing the training set data, it is input into the model. The model parameters are optimized using the five-fold cross-validation method. The model training is complete when the accuracy of matching the construction requirements of the test set is ≥95%. Model application: After expanding the dimensions of the regional attribute vector, input it into the trained model. The model outputs the standard construction requirements with the highest matching degree. The matching degree threshold is set to 0.85. If it is lower than the threshold, a manual review reminder is triggered.

[0055] In this embodiment, the requirement-process database is a pre-built database that stores the mapping relationship between standard construction requirements and construction process sets. It can quickly retrieve and match the corresponding construction process set based on the standard construction requirements, realizing the automated connection from construction requirements to specific construction steps. The database stores the mapping relationship of small area pothole hot-mix asphalt repair → grooving and cleaning → applying tack coat → filling hot-mix asphalt → compaction and molding, longitudinal crack injection → joint treatment → crack cleaning → injection → curing, etc. When the standard construction requirements for small area pothole hot-mix asphalt repair are input, the corresponding complete construction process set is quickly retrieved. This database uses a relational database such as MySQL or a non-relational database such as MongoDB, with the standard construction requirements as the primary key and the corresponding construction process set as the stored value. The corresponding construction process set can be quickly obtained based on the standard construction requirements through SQL query or document retrieval.

[0056] The beneficial effects of the above technical solution are as follows: the vector analysis submodule completes the grayscale processing and edge extraction of the construction site map, accurately identifies each connected region and generates regional attribute vectors containing construction properties and confidence levels. Then, the requirement matching submodule uses the vector analysis model to complete the automated mapping of regional features to standard construction requirements. Finally, the corresponding construction process set is retrieved from the requirement-process database, realizing efficient and accurate docking of construction site regional features to standardized construction processes. This effectively reduces the subjective error of manual judgment, improves the efficiency and standardization of construction process matching, lays a reliable process foundation for the intelligent management and control of the entire asphalt mixture construction process, and ensures the simultaneous improvement of construction quality and construction efficiency.

[0057] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, wherein the vector analysis submodule includes: The grayscale and edge processing unit is used to perform grayscale processing and edge extraction on the construction site map to obtain a first closed edge region and a first unclosed edge region. The complete coordinates and feature data of the first closed edge region are retained. The remaining image region after removing the first closed edge region is magnified and contrast enhanced once. The image after magnification and enhancement is further processed with grayscale and edge extraction to obtain a second closed edge region and a second unclosed edge region. In this embodiment, the first closed edge region is the area with a completely closed outline after the initial grayscale and edge extraction of the construction site map. It usually represents the core construction area (such as a complete road surface to be constructed or a large area of ​​regular defects). Its coordinates and features are preserved so that it can be spliced ​​with the unclosed edge to form a closed region later. For example, the complete asphalt pavement area of ​​an urban main road, after edge extraction, forms a closed outline with the beginning and end connected, which is the first closed edge region, representing the main road surface range to be constructed.

[0058] The first unclosed edge area is the area where the outline has breaks and cannot be closed after the initial grayscale and edge extraction. It usually represents edge-type defects (such as longitudinal cracks and pothole breaks). For example, the longitudinal cracks on the edge of the main road surface are not connected at both ends of the outline due to the shooting angle or the defect itself, forming the first unclosed edge area.

[0059] The magnification and contrast enhancement process involves using bilinear interpolation to magnify the remaining image (mainly the unclosed edge region) after removing the first closed edge, and then enhancing the image contrast through gamma correction or histogram equalization algorithms to increase the grayscale difference between the edge and the background.

[0060] Similar to the initial edge extraction process, Canny edge detection and contour closure judgment are performed on the magnified image to select closed contours as the second closed edge region, which is usually a small area of ​​closed defects (such as small pits or local damage) within the unclosed edge region.

[0061] The feature extraction and annotation unit is used to extract the edge grayscale features of the second unclosed edge region and annotate them on the edge line of the matched region; The initial feature acquisition unit is used to obtain the initial common features of each edge point on the edge line of the region based on the first nearest distance point set between the region edge line and the boundary line of the first closed edge region, and the second nearest distance point set between the region edge line and the boundary line of the second closed edge region. For each edge point of the region edge line, the minimum distance to all boundary pixels of the first closed edge region and the second closed edge region is calculated using the Euclidean distance algorithm, and the coordinates of the corresponding nearest point are recorded to form the first nearest distance point set and the second nearest distance point set.

[0062] The feature pair generation unit is used to perform normal probability analysis on all the initial common features of the edge line of the region to obtain the common gray-scale features of the corresponding edge line of the region, and to generate feature pairs by combining the edge gray-scale features of the corresponding edge line of the region, wherein the edge gray-scale features are the breakpoint extension trend features and contour smoothness features of the corresponding edge line of the region. Normal probability analysis statistically models the distribution of initial common features to identify frequently occurring feature patterns. Specifically, it fits a normal distribution to the gray-level components of the initial common features, calculates the mean and variance, and selects gray-level features with probability densities greater than a preset threshold as common gray-level features. For example, after analyzing the initial common features of all points at the crack edge, it was found that 85% of the gray-level values ​​were concentrated between 120 and 140. Normal analysis then extracted this range as the common gray-level features.

[0063] In this embodiment, least squares line fitting is performed at the breakpoint of the unclosed edge to obtain the fitted line y=kx+b; the angle between the fitted line and the horizontal axis is calculated, with the angle ranging from 0° to 180°, which is the breakpoint extension trend feature; if the edge line at the breakpoint is a curve, 10 pixels before and after the breakpoint are taken for curve fitting, the slope of the tangent line at the breakpoint is calculated, and then converted into an angle, which is used as the breakpoint extension trend feature; all contour points in the edge region are extracted, and the curvature ki between adjacent contour points is calculated; the variance of all curvatures is calculated and normalized (normalization range 0~1) to obtain the contour smoothness coefficient. The closer the coefficient is to 1, the smoother the contour; the calculation formula is: contour smoothness coefficient = 1 - corresponding variance / maximum value of curvature variance in historical data, which is used as the contour smoothness feature.

[0064] A feature pair is a feature group composed of common gray-level features and edge gray-level features (breakpoint extension trend, contour smoothness). For example, a feature pair can be represented as (130, 45°, 0.8), which represent common gray-level value, breakpoint extension angle, and contour smoothness coefficient, respectively.

[0065] A global search unit is used to perform a global search from the first nearest distance point set and the second nearest distance point set according to the features, to obtain a first point that matches the initial common features of the first point of the region edge line and a second point that matches the initial common features of the last point of the region edge line. The first point and the second point are located on the same continuous boundary line of the original closed edge region, and the matching degree of the initial common features is calculated based on the feature cosine similarity. Points with the matching degree of the initial common features greater than a preset matching threshold are regarded as matching points, and points that match the first point and the last point are selected from the matching points. Connect the first point to the first point, connect the last point to the second point, and incorporate the continuous boundary line segments of the original closed edge region between the first point and the second point into the outline to form a new closed edge region containing the edge lines of the corresponding region, wherein each closed edge region is a construction connected region; In this embodiment, the first point is the starting endpoint of the edge line of the second unclosed edge region, which is a break point that needs to be spliced ​​and closed, and the last point is the ending endpoint of the edge line of the second unclosed edge region, which is another break point that needs to be spliced ​​and closed.

[0066] The first point is the point with the highest initial common feature matching degree with the first point, located on the continuous boundary line of the original closed edge, and is used to connect with the first point to achieve closure; the second point is the point with the highest initial common feature matching degree with the last point, and is located on the same continuous boundary line of the original closed edge as the first point, and is used to connect with the last point to achieve closure.

[0067] The original closed edge region refers to the first closed edge region or the second closed edge region, which is the target closed region that needs to be spliced ​​and closed by the unclosed edge.

[0068] Use formula Calculate the feature cosine similarity between the vectors corresponding to two features, where A and B represent the normalized vector A of the initial common features of the first or last point and the normalized vector B corresponding to the feature of a candidate point on the edge of the region, respectively. The vectors corresponding to the initial common features and the feature of the candidate point have 8 dimensions, namely: edge point gray value, mean gray value of the first nearest line, mean gray value of the second nearest line, mean gray value of the first circle, mean gray value of the second circle, angle of the breakpoint extension trend, contour smoothness coefficient, and normalized value of the edge point coordinates. The values ​​of each dimension need to be normalized to between 0 and 1 for easy calculation.

[0069] In this embodiment, the preset matching threshold is usually set between 0.7 and 0.9 based on historical construction data. In this invention, the value is 0.8. Experimental verification shows that when the matching degree is ≥0.80, the splicing accuracy of feature points is ≥98%, which can effectively achieve accurate splicing of unclosed edges and original closed edges.

[0070] Matching points are points whose initial shared feature similarity is greater than a preset matching threshold, and are candidate splicing closure points.

[0071] The newly closed edge region connects pixels in the order of first point → first point → second point → last point → original edge line to form a closed contour.

[0072] The alignment processing unit is used to convert the regional construction drawing of each connected region into R-channel, G-channel, and B-channel respectively, extract the feature set of each channel, and perform alignment processing on the feature set of each channel to obtain several intersection features and several non-intersection features. Among them, the overlapping features of two or more channels are regarded as intersection features. The confidence level determination unit is used to match the construction properties corresponding to each feature from a preset attribute-property lookup table; When the feature is an intersection feature, the confidence level of the corresponding construction nature is obtained based on the construction nature, feature shape and number of overlapping channels of the intersection feature; When the features are non-intersecting, the corresponding confidence level is matched from the historical database based on the construction nature of the non-intersecting features.

[0073] In this embodiment, the split function is used to split the regional construction drawing into RGB channels. Edge, texture and other feature sets are extracted for each channel. Based on the original image coordinates, the feature sets of each channel are mapped to the same coordinate system to achieve pixel-level alignment. Overlapping features are marked as intersection features, and features that exist only in a single channel are marked as non-intersection features. The R channel, G channel and B channel are the three color component channels of the color image, representing red, green and blue information respectively. After splitting, different dimensions of visual features can be extracted. For example, the R channel can reflect the temperature-related features of the asphalt mixture, the G channel reflects texture details, and the B channel reflects brightness information.

[0074] The channel feature set is a collection of visual features extracted from a single color channel. It integrates the edge features (number of edge points, contour length, closure), texture features (contrast, correlation, energy, uniformity of each channel), and grayscale gradient features (gradient mean, gradient variance, and gradient maximum of each channel) of each channel to form a 10-dimensional channel feature set.

[0075] Alignment processing involves matching and aligning feature sets from different channels at pixel coordinates. For example, it aligns the edge features of the R, G, and B channels to the same pixel coordinates. It determines which features coexist in multiple channels. Based on the pixel coordinates of the original image, it maps the coordinates of each channel feature set to the same coordinate system to achieve pixel-level alignment.

[0076] Intersecting features are features that overlap in two or more color channels. For example, a crack edge exists in three channels (R, G, and B) or two channels, which is an intersecting feature. Non-intersecting features are features that exist only in a single color channel. For example, a temperature feature exists only in the R channel, which is a non-intersecting feature.

[0077] In this embodiment, for intersecting features, the confidence level is calculated as follows: Construction nature matching coefficient × Morphological integrity coefficient × Number of overlapping channels / Total number of channels. For non-intersecting features, the historical database is queried using the feature as the keyword, and the average historical confidence level is taken as the current confidence level. The construction nature matching coefficient ranges from 0.7 to 1.0 and is determined based on the degree of matching between the visual feature and the construction nature. For example, the matching coefficient between crack edge features and crack repair construction nature is 1.0, and the matching coefficient between pit contour features and small area defects construction nature is 0.9. The morphological integrity coefficient ranges from 0.6 to 1.0 and is calculated based on the contour integrity of the feature. The formula is: Morphological integrity coefficient = Actual feature contour length / Standard feature contour length. When the deviation between the actual contour length and the standard contour length is ≤5%, the coefficient is 1.0.

[0078] The attribute-property mapping table is a pre-built mapping table that stores the mapping relationship between visual feature attributes and construction properties. It is used to convert image features into qualitative labels in the construction field, as shown in Table 1: Table 1 Attribute-Property Comparison Table

[0079] It should be noted that all visual feature attributes are determined through quantitative indicators extracted from image processing, avoiding subjective human judgment.

[0080] Construction nature is a qualitative label that describes the construction characteristics of a connected area, such as crack repair, pothole repair, temperature monitoring, etc. The construction nature label corresponding to the feature can be found in the attribute-nature lookup table.

[0081] The historical database stores historical construction features and their corresponding confidence levels. By searching the historical database using the type of non-intersecting feature as the keyword, the confidence levels of the last 50 sets of similar features are filtered out. After removing the maximum and minimum values, the average value is taken as the confidence level of the current non-intersecting feature. For example, R channel temperature feature → 0.88, G channel texture feature → 0.92.

[0082] The beneficial effects of the above technical solution are as follows: by analyzing the construction site map in stages through grayscale and edge processing units, closed and unclosed edge regions are accurately identified and small-sized regions are enhanced. Then, through feature extraction, initial common feature generation, feature pair construction, and global search, the unclosed edges and the original closed edges are accurately spliced ​​and closed to form a complete construction connectivity region. Subsequently, through multi-channel feature alignment and differential confidence calculation, reliable construction properties and confidence levels are generated for each region, providing accurate and quantitative basic data for subsequent construction process matching. This effectively solves the problem of incomplete region identification caused by edge breakpoints in the construction site map and greatly improves the accuracy and reliability of construction connectivity region identification.

[0083] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, wherein the initial feature acquisition unit includes: The sub-unit is used to acquire the first grayscale representation of each first nearest distance line in the first nearest distance point set, and the first auxiliary grayscale representation of a first circle constructed with the first nearest distance as the diameter and the point on the boundary line of the first nearest distance line located on the first closed edge region as the center. Simultaneously, the second grayscale representation of each second nearest distance line in the second nearest distance point set is obtained, as well as the second auxiliary grayscale representation of a second circle constructed with the second nearest distance as the diameter and the point on the boundary line of the second nearest distance line located in the second closed edge region as the center; The relationship determination subunit is used to determine the positional relationship between the first circle and the second circle corresponding to each edge point on the edge line of the region. When the positional relationship is an intersection relationship, the intersection representation of the first grayscale representation and the second grayscale representation, as well as the grayscale representation of the intersection region, are extracted as the initial common features of the corresponding edge points; When the positional relationship is non-intersecting, the intersection of the first grayscale representation and the second grayscale representation, as well as the intersection of the first auxiliary grayscale representation and the second auxiliary grayscale representation, are extracted as the initial common features of the corresponding edge points.

[0084] In this embodiment, the first nearest distance line is a straight line connecting a certain edge point on the edge line of the region to the nearest point on the boundary line of the first closed edge region. The first grayscale representation is the set of grayscale features of all pixels on the first nearest distance line, which includes the grayscale mean and grayscale variance of the first nearest distance line, and the first nearest distance is the length of the corresponding first nearest distance line. The first circle is a circular region constructed with the point on the boundary line of the first closed edge region located on the first nearest distance line as the center and the first nearest distance as the diameter. The first auxiliary grayscale representation is the set of grayscale features of all pixels in the first circle, that is, it includes the grayscale mean and grayscale variance of the circle. It should be noted that the acquisition principles of the second nearest distance, the second nearest distance line, the second grayscale representation, and the first auxiliary grayscale representation are similar to those of the first nearest distance, the first nearest distance line, the first grayscale representation, and the first auxiliary grayscale representation, and will not be repeated here.

[0085] In this embodiment, the positional relationship refers to the geometric positional relationship between the first circle and the second circle, which is divided into intersecting relationship and non-intersecting relationship. For example, if the center distance between the first circle and the second circle is 9 pixels, and the radii of the two circles are 5 pixels and 4 pixels respectively, the center distance is equal to the sum of the two radii. At this time, the two circles are externally tangent (belonging to a type of non-intersecting relationship). If the center distance is 7 pixels, which is less than the sum of the two radii and greater than the difference of the radii, it is an intersecting relationship.

[0086] In this embodiment, the intersection representation refers to the gray-level feature portion that exists in common between two gray-level representations (such as the first gray-level representation and the second gray-level representation, or the first auxiliary gray-level representation and the second auxiliary gray-level representation), used to characterize the common gray-level characteristics of edge points and two types of closed edges. For example, if the gray-level mean of the first gray-level representation is 125 and the gray-level mean of the second gray-level representation is 120, the intersection representation is the feature set of gray-level values ​​between 120 and 125, representing the common gray-level features of two types of closest lines.

[0087] The grayscale representation of the intersecting region refers to the set of grayscale features of all pixels within the overlapping area of ​​the first and second circles. It is extracted only when the two circles intersect and is used to supplement the local common grayscale features of edge points and two types of closed edges. For example, if the mean grayscale value of the pixels within the overlapping area of ​​the two circles is 123 and the grayscale variance is 6, the combination of this mean and variance is the grayscale representation of the intersecting region.

[0088] Intersecting relationship refers to the positional relationship between the first circle and the second circle where there is an overlapping area, including cases where there is a common pixel area such as intersection and internal tangency. Non-intersecting relationship refers to the positional relationship between the first circle and the second circle where there is no overlapping area, including cases where there is external separation, external tangency, internal tangency, and internal inclusion.

[0089] The initial shared features are extracted by the relationship judgment subunit based on the positional relationship between the two circles. They are feature vectors composed of the intersection representation and the corresponding supplementary grayscale features, representing the shared features of the edge point with the first closed edge and the second closed edge. For example, when the two circles intersect, the initial shared features are (first-second grayscale intersection representation, intersection region grayscale representation); when the two circles do not intersect, the initial shared features are (first-second grayscale intersection representation, first-second auxiliary grayscale intersection representation).

[0090] The beneficial effects of the above technical solution are as follows: by representing the extraction sub-unit to extract the grayscale features of the shortest distance line from the edge point to the two types of closed edges and the corresponding circle, and then by the relationship judgment sub-unit to extract the intersection features according to the difference in the positional relationship of the two circles, the initial common features that accurately represent the common attributes of the edge point and the two types of closed edges are generated. This not only preserves the grayscale transition information between the edge point and the closed edge, but also enhances the distinguishability of the features through the geometric positional relationship, effectively improving the accuracy of subsequent feature matching and connected region splicing, and providing a reliable feature basis for the complete identification of connected regions in asphalt mixture construction.

[0091] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, wherein the optimal search module includes: The normalization submodule is used to normalize the construction feature dimension of the fine difference set, the first difference set, and the second difference set at each adjacent time between the first end time of the initial construction time period and the second end time of the first designated construction anchor point, so as to obtain a standardized difference set. The initial position mapping submodule is used to set the number of ant colony individuals according to the number of construction feature dimensions of the standardized difference set, map the initial position of each ant colony individual to the construction feature dimension node of the standardized difference set, and set the initial pheromone concentration between construction feature dimension nodes based on the feature correlation degree of asphalt mixture construction parameters. The construction feature dimension node is the node corresponding to a single construction feature dimension of the standardized difference set, and each node corresponds to a construction feature parameter. The path search submodule is used by each ant colony individual to conduct path search between nodes in the construction feature dimension with the core search guide of minimizing the prediction deviation of construction effectiveness. During the search process, the node transfer probability is dynamically corrected by combining the feature similarity of the fine construction set of the next time step corresponding to the adjacent time step. The update submodule is used to collect the difference sets corresponding to the search paths of all ant colony individuals to generate several candidate sets, and to perform local pheromone updates and global pheromone updates on the search paths of individual ant colonies to obtain the optimal difference set.

[0092] Preferably, the local pheromone update is performed immediately after an individual ant completes a single node transfer, reducing the pheromone concentration of the corresponding path node according to a preset basic evaporation coefficient; the global pheromone update is performed after all ant individuals complete a complete path search, releasing pheromone increments only for the search paths corresponding to the difference set within the candidate set, and the release increment value is negatively correlated with the objective function value of the construction effectiveness prediction deviation of the corresponding path; the objective function of the construction effectiveness prediction deviation is the reciprocal of the feature similarity between the predicted effectiveness map and the final effectiveness effect map of the corresponding stage, and the smaller the objective function value, the smaller the construction effectiveness prediction deviation.

[0093] The construction feature dimensions are the fine difference set, the first difference set, and the second difference set. Each construction parameter corresponds to a dimension, such as the paving temperature dimension, paving speed dimension, construction coverage dimension, and mixture gradation dimension. For example, a difference set may contain four construction feature dimensions: paving temperature, paving speed, construction coverage, and mixture gradation. Each dimension corresponds to the difference value of a construction parameter.

[0094] Normalization is a process that converts the differences in construction feature dimensions with different dimensions and numerical ranges into numerical values ​​on a uniform scale (e.g., the range of 0 to 1). Its core purpose is to eliminate the interference of dimensional differences on subsequent ant colony search, ensuring that each feature dimension has equal computational weight in the search. For example, differences in paving temperature and construction coverage have extremely large numerical ranges. After normalization, they are all converted into values ​​between 0 and 1. Min-max normalization or Z-score standardization methods are used to linearly transform the difference values ​​of each construction feature dimension, mapping them to a uniform range. This is a current technology.

[0095] The standardized difference set is a difference set in which all differences in construction feature dimensions are converted into a uniform scale after normalization.

[0096] In this embodiment, the initial pheromone concentration = base value + correlation degree × weight is set to determine the initial pheromone concentration between nodes. The base value is a fixed concentration constant (e.g., 0.1) to avoid the initial pheromone being zero.

[0097] An ant colony individual is a search unit in the ant colony algorithm that simulates the foraging behavior of ants. Each individual represents a candidate solution of a set of differences. It moves between nodes of construction feature dimensions and gradually finds the difference set path that minimizes the prediction deviation of construction effect. For example, four ant colony individuals start from the nodes of paving temperature, paving speed, construction coverage, and mixture gradation, respectively, and move between the nodes of each dimension to form different difference set combination paths.

[0098] Construction feature dimension nodes are abstract nodes corresponding to a single construction feature dimension in the standardization difference set. Each node corresponds to a construction feature parameter and is divided into three categories according to the type of construction parameter: construction environment, asphalt mixture construction, and construction coverage.

[0099] Characteristic correlation is the degree of physical correlation between different asphalt mixture construction parameters. It is based on historical construction data and calculates the Pearson correlation coefficient between different construction parameters. The correlation coefficient is used as the characteristic correlation. The higher the correlation, the greater the initial pheromone concentration between nodes. The correlation between paving temperature and paving speed is 0.8 (high correlation), and the correlation between construction coverage and mixture gradation is 0.3 (low correlation). Therefore, the former has a higher initial pheromone concentration between nodes.

[0100] In this embodiment, minimizing the prediction deviation of construction effectiveness is the core objective of ant colony search. That is, finding the combination of differences that maximizes the feature similarity (minimum deviation) between the predicted effectiveness map and the final effect map of the stage is the core guide for the movement of individual ants. The objective function of construction effectiveness prediction deviation (the inverse of similarity) is used as a heuristic function. The smaller the objective function value, the larger the heuristic function value, guiding individual ants to move in that direction. For example, if the combination of differences corresponding to a certain path makes the similarity between the predicted effectiveness map and the standard effect map 0.95 (small deviation), while another path has a similarity of 0.88 (large deviation), then the former is better, and the ants will gradually gather towards that path.

[0101] In this embodiment, the standard transition probability formula of the ant colony algorithm is used: ,in For pheromone concentration, Let be the heuristic function, s be the feature cosine similarity between the predicted effect map and the final effect map of the stage, and α, β, and γ be weight parameters used to adjust the influence of each factor, with values ​​of 0.2, 0.3, and 2 respectively; sum1 is the sum of all adjacent construction feature dimension nodes of the node where the current ant colony individual is located. The sum of.

[0102] The feature similarity of the fine construction set is the degree of feature similarity between the fine construction set at the next adjacent time and the standard construction set. The feature cosine similarity between the fine construction set and the standard construction set is calculated, and the similarity value is used as a correction factor and substituted into the node transition probability formula for dynamic correction.

[0103] Collect the difference sets corresponding to the paths of all ant colony individuals, sort them according to the objective function value of the construction effectiveness prediction deviation, and select the top K groups (such as the top 3 groups) as the candidate set.

[0104] New pheromone concentration ,in Based on the volatility coefficient, and 0 < <1, The path pheromone concentration of an individual ant colony before completing a single node transfer; Based on algorithm debugging experience, the settings are configured. The value is typically between 0.1 and 0.3 to ensure continuous pheromone replenishment while avoiding excessive evaporation. In this invention, a value of 0.1 is used. Multiple experimental comparisons have shown that... When the value is 0.1, the convergence speed and global search capability of the ant colony algorithm are balanced, and the global optimal solution can be found within 50 iterations.

[0105] Global pheromone update is a pheromone update operation performed after all ant colonies have completed a full path search. It only releases pheromone increments for search paths corresponding to the differences within the candidate set. The released increment value is negatively correlated with the objective function value of the construction effectiveness prediction deviation for the corresponding path, thus strengthening the pheromone concentration of the optimal path and guiding the ant colony to converge towards the global optimum. For example, if the objective function value of the construction effectiveness prediction deviation for a certain path in the candidate set is 0.05 (small deviation), the released pheromone increment is 0.2; if the objective function value of another path is 0.1 (large deviation), the released increment is 0.1. The former has a larger increment, further strengthening the pheromone advantage of the optimal path. Where Q is a preset constant with a value of 10, and f is the objective function value. For pheromone increment, , This represents the pheromone concentration after a global update. The path pheromone concentration after the ant colony completes a full path search.

[0106] The objective function for predicting construction effectiveness deviation is the reciprocal of the feature similarity between the predicted effectiveness map and the corresponding final effectiveness map. That is, it calculates the feature cosine similarity s between the predicted effectiveness map and the final effectiveness map of the stage. The objective function is f=1 / s. The smaller f is, the better the combination of difference sets.

[0107] The optimal difference set is the combination of difference sets corresponding to the path with the highest pheromone concentration after multiple rounds of ant colony search and pheromone update. For example, if after multiple rounds of search, a certain path has the highest pheromone concentration, and the corresponding difference set combination makes the similarity between the predicted effect map and the standard effect map 0.95, then it is the optimal difference set.

[0108] The beneficial effects of the above technical solution are as follows: the normalization submodule eliminates the differences in dimensions and numerical ranges of different construction features, generating a standardized difference set; the initial position mapping submodule initializes the ant colony by combining the correlation of construction parameters; the path search submodule dynamically corrects the node transfer probability with the goal of minimizing the deviation of construction results; and finally, the update submodule guides the ant colony to converge toward the optimal difference set path through local pheromone volatilization and global pheromone increment release. This not only ensures the global optimality of the difference set search but also conforms to the actual construction characteristics, effectively improving the accuracy of construction results prediction. This lays a reliable foundation for the subsequent generation of precise construction improvement sub-processes and helps to achieve intelligent and refined management of the entire asphalt mixture construction process.

[0109] This invention provides an intelligent management and control system for the entire construction process of asphalt mixtures, wherein the construction improvement module includes: The initial correction submodule is used to map the construction deviation features of the element sequence to each construction procedure of the standard sub-process from the first specified construction anchor point to the next specified construction anchor point. Combined with the construction effectiveness feature requirements of the standard as-built drawing of the next specified construction anchor point, the initial correction is performed on the construction parameters, construction direction, and construction coverage threshold of the corresponding procedures in the standard sub-process to generate the initial sub-process of the next specified construction anchor point. The construction deviation features are the construction parameter deviations and construction effectiveness deviation features contained in the difference effect set in the element sequence. The simulation iterative optimization submodule is used to input the initial subprocess into a pre-built asphalt mixture construction simulation model, incorporate real-time construction environment parameters to simulate the entire construction process, generate a simulation completion diagram, and calculate the feature similarity between the simulation completion diagram and the standard completion diagram of the next specified construction anchor point. If the feature similarity does not reach the preset threshold, the deviation between the simulation completion drawing and the standard completion drawing is analyzed for process traceability. For the deviation-related processes obtained from the traceability, the corresponding process correction parameters of the initial sub-process are adjusted in a targeted manner until the feature similarity between the simulation completion drawing and the standard completion drawing meets the preset threshold. The finally adjusted sub-process is determined as the improved sub-process for the next specified construction anchor point.

[0110] In this embodiment, the construction deviation characteristics are the construction parameter deviations and construction performance deviations contained in the difference effect set in the element sequence. For example, the construction parameter deviations include paving temperature -3℃ and paving speed +0.5m / min; the construction performance deviations include smoothness +1mm / m and compaction degree -2%, which directly reflect the current construction problems in terms of parameters and quality.

[0111] Construction performance deviation characteristics refer to the quality deviation characteristics between the predicted construction performance (predicted performance diagram) and the standard construction performance (final performance effect diagram of the stage). For example, if the standard flatness requirement is ≤3mm / meter and the predicted performance diagram shows a flatness of 4mm / meter, then the construction performance deviation is +1mm / meter; if the standard compaction degree is ≥96% and the prediction is 94%, then the deviation is -2%.

[0112] The construction direction is the direction in which construction equipment moves during construction. It is a clearly defined operational requirement in the standard sub-process and includes longitudinal advancement direction and transverse paving direction.

[0113] The construction coverage threshold is the standard range of the construction road surface area covered by construction equipment per unit time. For example, the construction coverage threshold for the paving process is 150-200 square meters per hour. Any area exceeding or falling below this range needs to be corrected.

[0114] The initial sub-process is generated by the initial correction sub-module. It is a construction sub-process after preliminary correction for the next specified construction anchor point. For example, the corrected paving temperature is 163-167℃, the paving speed is 2.7-3.3m / min, and the construction coverage threshold is 150-200 square meters / hour. These are combined to form the initial sub-process.

[0115] In this embodiment, the asphalt mixture construction simulation model adopts a hybrid architecture of finite element analysis and gradient lifting regression (GBR) to achieve digital simulation of the entire construction process, as detailed below: Model input and output: The input consists of construction sub-process parameters (construction parameters, construction direction, coverage threshold) and real-time construction environment parameters, totaling 22 features; the output consists of construction simulation data (smoothness, compaction, paving thickness, etc.) and a simulation completion image with a resolution of 512×512. Finite element analysis details: Using ABAQUS finite element analysis software, the construction area was divided into 10cm×10cm grid units. The constitutive model of the asphalt mixture was defined as a viscoelastic model. Boundary conditions such as the load and speed of the construction equipment were set to simulate the mechanical processes of paving, compaction and other procedures. Regression model training: 800 sets of finite element simulation data and 200 sets of on-site construction data were collected as training sets. The number of gradient boosting regression trees was 200, the learning rate was 0.01, the loss function was absolute error loss, and the training was completed when the model fit R² ≥ 0.92. Real-time parameter fusion algorithm: The weighted fusion method is adopted, and the influence weights of real-time construction environment parameters are set according to empirical values ​​(ambient temperature 0.3, air humidity 0.2, wind speed 0.1) and integrated into the boundary conditions of finite element analysis to realize dynamic simulation of the construction process.

[0116] The full-process simulation of construction is a simulation model that simulates the complete construction process from the first designated construction anchor point to the next designated construction anchor point based on the initial sub-process and real-time construction environment parameters. It includes the execution of each process, parameter changes, and environmental impacts, and finally generates a simulated as-built diagram. For example, simulating paving construction from 200 meters to 500 meters, from the paver being in place to the final compaction, the changes in parameters of each process and the environmental impacts are calculated step by step to obtain the simulated road construction effect.

[0117] The simulation completion image is a visual image output by the simulation model that represents the road surface condition after the initial sub-process construction is completed. For example, the simulation completion image shows that the road surface smoothness is 4mm / meter, the compaction degree is 95%, and the mixture is evenly distributed, which intuitively presents the construction quality of the initial sub-process.

[0118] Feature similarity is the degree of visual feature similarity between the simulated completed image and the standard completed image at the next specified construction anchor point. The value range is 0-1. That is, the cosine similarity algorithm is used to calculate the similarity of features such as edge, texture, and brightness between the two images to obtain the feature similarity.

[0119] Based on project quality requirements and historical construction data, the preset threshold is usually set between 0.85 and 0.95. In this invention, it is consistent with the construction quality acceptance standard. When the similarity is ≥0.90, the deviation between the simulation result and the actual construction result is ≤3%, which can be used as the basis for optimizing the construction sub-process. Therefore, the preset threshold is set to 0.9.

[0120] Process traceability analysis is the process of analyzing the specific construction process that causes the deviation when there is a discrepancy between the simulated as-built drawing and the standard as-built drawing. For example, if the flatness deviation of the simulated as-built drawing is +1mm / meter, the traceability analysis will find that the deviation mainly comes from the paving process where the paving speed is too fast, resulting in a worse road surface flatness. In this case, the paving process is the deviation-related process.

[0121] Correction parameters are the construction parameters, construction direction, and construction coverage thresholds that need to be adjusted for the deviation-related processes in the initial sub-process. For example, the paving speed of the paving process is a correction parameter that needs to be adjusted from 3.5m / min to 3m / min.

[0122] Oriented adjustment is an operation that precisely adjusts correction parameters according to the direction and magnitude of deviation in deviation-related processes. That is, it linearly adjusts correction parameters according to a preset ratio based on the magnitude and direction of the deviation. For example, if the deviation is flatness +1mm / meter, caused by excessive paving speed, the directional adjustment will reduce the paving speed from 3.5m / min to 3m / min to reduce the negative impact of speed on flatness.

[0123] The improved sub-process is the final construction sub-process determined after simulation and iterative optimization for the next specified construction anchor point. For example, the adjusted paving temperature is 165℃, the paving speed is 3m / min, and the construction coverage area is 180 square meters / hour. These are combined to form the improved sub-process.

[0124] The beneficial effects of the above technical solution are as follows: the initial correction submodule accurately maps the construction deviation characteristics to the standard subprocess, generating an initial subprocess for preliminary correction. Then, the simulation iterative optimization submodule integrates real-time construction environment parameters to simulate the entire process. With the standard as-built drawing as the target, deviation source tracing and directional parameter adjustment are carried out. Iterative optimization is performed until the construction results meet the standards, and finally, a reliable improved subprocess is output. This realizes closed-loop management of construction deviations from identification to correction, ensuring that the entire asphalt mixture construction process meets the standard requirements.

[0125] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent control system for the entire construction process of asphalt mixtures, characterized in that, include: The process set acquisition module is used to acquire the construction site map of the area to be constructed, and obtain the corresponding construction process set according to the standard construction requirements of each connected area in the construction site map; The precision determination module is used to monitor the actual construction direction of each stage during the initial construction period when construction is carried out according to the construction process set, and compare it with the standard construction direction of the standard construction process of the corresponding stage. If the operation direction is inconsistent, it will remind to stop construction. If the operation direction is consistent, obtain the detailed construction set for each actual construction moment between the first end time of the initial construction period and the second end time of the first designated construction anchor point in the corresponding stage. The detailed construction set includes construction environment parameters, asphalt mixture construction parameters, construction coverage area and coverage effect map of the construction coverage area. The optimal search module, based on the ant colony algorithm, performs optimal search on the fine difference set at each adjacent time point between the first end time of the initial construction period and the second end time of the first designated construction anchor point, the first difference set between the fine construction set at the next adjacent time point and the fine construction set at the second end time point, and the second difference set between the fine construction set at the next adjacent time point and the standard construction set at the first designated construction anchor point. The optimal difference set is then input into a pre-built predictive analysis model for the same stage to obtain the predicted performance map after the completion of the corresponding stage. The fine difference set is the set of parameter differences of the fine construction sets at two adjacent actual construction times. The sequence construction module is used to mark the element of the next time step of the corresponding adjacent time step as 0 if the similarity between the predicted effect map and the final effect map of the corresponding standard construction process is greater than a preset similarity threshold. Otherwise, extract the difference effect set based on the final effect map of the corresponding stage from the predicted effect map, use it as a new element for the next time moment, and generate an element sequence from the first end time of the initial construction period to the second end time of the first specified construction anchor point. The construction improvement module is used to generate an improved sub-process for the next specified construction anchor point based on the element sequence, combined with the standard sub-process from the first specified construction anchor point to the next specified construction anchor point and the standard as-built drawing of the next specified construction anchor point, and to perform construction control on the corresponding connected area according to the improved sub-process until the construction is completed.

2. The intelligent control system for the entire construction process of asphalt mixtures according to claim 1, characterized in that, The process set acquisition module includes: The vector analysis submodule is used to perform grayscale processing and edge extraction on the acquired construction site map, and analyze the regional attribute vector of each connected region. The regional attribute vector contains several construction properties of the corresponding connected region and the confidence level of each construction property. The matching submodule is required to input the regional attribute vector into a pre-built vector analysis model to obtain the standard construction requirements of the corresponding connected region, and to match the construction process set of the corresponding connected region from the preset requirements-process database.

3. The intelligent control system for the entire construction process of asphalt mixtures according to claim 2, characterized in that, The vector analysis submodule includes: The grayscale and edge processing unit is used to perform grayscale processing and edge extraction on the construction site map to obtain a first closed edge region and a first unclosed edge region. The complete coordinates and feature data of the first closed edge region are retained. The remaining image region after removing the first closed edge region is magnified and contrast enhanced once. The image after magnification and enhancement is further processed with grayscale and edge extraction to obtain a second closed edge region and a second unclosed edge region. The feature extraction and annotation unit is used to extract the edge grayscale features of the second unclosed edge region and annotate them on the edge line of the matched region; The initial feature acquisition unit is used to obtain the initial common features of each edge point on the edge line of the region based on the first nearest distance point set between the region edge line and the boundary line of the first closed edge region, and the second nearest distance point set between the region edge line and the boundary line of the second closed edge region. The feature pair generation unit is used to perform normal probability analysis on all the initial common features of the edge line of the region to obtain the common gray-scale features of the corresponding edge line of the region, and to generate feature pairs by combining the edge gray-scale features of the corresponding edge line of the region, wherein the edge gray-scale features are the breakpoint extension trend features and contour smoothness features of the corresponding edge line of the region. A global search unit is used to perform a global search from the first nearest distance point set and the second nearest distance point set according to the features, to obtain a first point that matches the initial common features of the first point of the region edge line and a second point that matches the initial common features of the last point of the region edge line. The first point and the second point are located on the same continuous boundary line of the original closed edge region. The matching degree of the initial common features is calculated based on the feature cosine similarity. Points with a matching degree of the initial common features greater than a preset matching threshold are regarded as matching points. From the matching points, the point with the largest matching degree with the first point is regarded as the first point and the point with the largest matching degree with the last point is regarded as the second point. Connect the first point to the first point, connect the last point to the second point, and incorporate the continuous boundary line segments of the original closed edge region between the first point and the second point into the outline to form a new closed edge region containing the edge lines of the corresponding region, wherein each closed edge region is a construction connected region; The alignment processing unit is used to convert the regional construction drawing of each connected region into R-channel, G-channel, and B-channel respectively, extract the feature set of each channel, and perform alignment processing on the feature set of each channel to obtain several intersection features and several non-intersection features. Among them, the overlapping features of two or more channels are regarded as intersection features. The confidence level determination unit is used to match the construction properties corresponding to each feature from a preset attribute-property lookup table; When the feature is an intersection feature, the confidence level of the corresponding construction nature is obtained based on the construction nature, feature shape and number of overlapping channels of the intersection feature; When the features are non-intersecting, the corresponding confidence level is matched from the historical database based on the construction nature of the non-intersecting features.

4. The intelligent control system for the entire construction process of asphalt mixtures according to claim 3, characterized in that, The initial feature acquisition unit includes: The sub-unit is used to acquire the first grayscale representation of each first nearest distance line in the first nearest distance point set, and the first auxiliary grayscale representation of a first circle constructed with the first nearest distance as the diameter and the point on the boundary line of the first nearest distance line located on the first closed edge region as the center. Simultaneously, the second grayscale representation of each second nearest distance line in the second nearest distance point set is obtained, as well as the second auxiliary grayscale representation of a second circle constructed with the second nearest distance as the diameter and the point on the boundary line of the second nearest distance line located in the second closed edge region as the center; The relationship determination subunit is used to determine the positional relationship between the first circle and the second circle corresponding to each edge point on the edge line of the region. When the positional relationship is an intersection relationship, the intersection representation of the first grayscale representation and the second grayscale representation, as well as the grayscale representation of the intersection region, are extracted as the initial common features of the corresponding edge points; When the positional relationship is non-intersecting, the intersection of the first grayscale representation and the second grayscale representation, as well as the intersection of the first auxiliary grayscale representation and the second auxiliary grayscale representation, are extracted as the initial common features of the corresponding edge points.

5. The intelligent control system for the entire construction process of asphalt mixtures according to claim 1, characterized in that, The optimal search module includes: The normalization submodule is used to normalize the construction feature dimension of the fine difference set, the first difference set, and the second difference set at each adjacent time between the first end time of the initial construction time period and the second end time of the first designated construction anchor point, so as to obtain a standardized difference set. The initial position mapping submodule is used to set the number of ant colony individuals according to the number of construction feature dimensions of the standardized difference set, map the initial position of each ant colony individual to the construction feature dimension node of the standardized difference set, and set the initial pheromone concentration between construction feature dimension nodes based on the feature correlation degree of asphalt mixture construction parameters. The construction feature dimension node is the node corresponding to a single construction feature dimension of the standardized difference set, and each node corresponds to a construction feature parameter. The path search submodule is used by each ant colony individual to conduct path search between nodes in the construction feature dimension with the core search guide of minimizing the prediction deviation of construction effectiveness. During the search process, the node transfer probability is dynamically corrected by combining the feature similarity of the fine construction set of the next time step corresponding to the adjacent time step. The update submodule is used to collect the difference sets corresponding to the search paths of all ant colony individuals to generate several candidate sets, and to perform local pheromone updates and global pheromone updates on the search paths of individual ant colonies to obtain the optimal difference set.

6. The intelligent control system for the entire construction process of asphalt mixtures according to claim 5, characterized in that, The local pheromone update is executed immediately after an individual ant completes a single node transfer, reducing the pheromone concentration of the corresponding path node according to a preset base evaporation coefficient; the global pheromone update is executed after all ant individuals complete a complete path search, releasing pheromone increments only for the search paths corresponding to the difference set within the candidate set, with the release increment value negatively correlated with the objective function value of the construction effectiveness prediction deviation for the corresponding path; the objective function of the construction effectiveness prediction deviation is the reciprocal of the feature similarity between the predicted effectiveness map and the final effectiveness effect map of the corresponding stage, and the smaller the objective function value, the smaller the construction effectiveness prediction deviation.

7. The intelligent control system for the entire construction process of asphalt mixtures according to claim 1, characterized in that, The improved sub-process includes the correction direction for the already constructed content at the previous specified construction anchor point and the improvement direction for the content to be constructed at the current specified construction anchor point. Both the correction direction and the improvement direction include adjustments to three dimensions: construction parameters, construction direction, and construction coverage threshold.

8. The intelligent control system for the entire construction process of asphalt mixtures according to claim 1, characterized in that, The construction improvement module includes: The initial correction submodule is used to map the construction deviation features of the element sequence to each construction procedure of the standard sub-process from the first specified construction anchor point to the next specified construction anchor point. Combined with the construction effectiveness feature requirements of the standard as-built drawing of the next specified construction anchor point, the initial correction is performed on the construction parameters, construction direction, and construction coverage threshold of the corresponding procedures in the standard sub-process to generate the initial sub-process of the next specified construction anchor point. The construction deviation features are the construction parameter deviations and construction effectiveness deviation features contained in the difference effect set in the element sequence. The simulation iterative optimization submodule is used to input the initial subprocess into a pre-built asphalt mixture construction simulation model, incorporate real-time construction environment parameters to simulate the entire construction process, generate a simulation completion diagram, and calculate the feature similarity between the simulation completion diagram and the standard completion diagram of the next specified construction anchor point. If the feature similarity does not reach the preset threshold, the deviation between the simulation completion drawing and the standard completion drawing is analyzed for process traceability. For the deviation-related processes obtained from the traceability, the corresponding process correction parameters of the initial sub-process are adjusted in a targeted manner until the feature similarity between the simulation completion drawing and the standard completion drawing meets the preset threshold. The finally adjusted sub-process is determined as the improved sub-process for the next specified construction anchor point.

Images