Road maintenance management system and method based on AI deep learning technology

Abstract

The invention relates to the technical field of road maintenance, in particular to a road maintenance management system and method based on an AI deep learning technology. The road maintenance management system based on the AI deep learning technology comprises an information acquisition unit, an information processing unit and a maintenance management unit. The AI deep learning technology and the road maintenance management are deeply fused, the road image is analyzed through the intelligent identification model, the pavement disease information is accurately obtained, the pavement disease comprehensive evaluation value and the relative change rate at different detection moments are calculated, the pavement disease development trend can be accurately evaluated, and the road maintenance management efficiency is improved. And according to the size relation of the comprehensive evaluation values of the pavement diseases at the first detection moment and the second detection moment, different sorting modes are adopted to carry out maintenance sorting on each target area road section, so that the maintenance decision management is more scientific and reasonable.

Description

Technical Field

[0001] The present invention relates to the technical field of highway maintenance, and particularly to a highway maintenance management system and method based on AI deep learning technology. Background Technique

[0002] Highways, as key infrastructure for transportation, play a crucial role in economic development and social operation. With the continuous growth of traffic flow, the continuous increase of vehicle loads, and the long-term erosion of natural environmental factors, various diseases will inevitably occur on highway pavements, such as cracks and potholes. These diseases will not only affect the comfort and safety of driving, but may also lead to damage to the highway structure and shorten the service life of the highway. Therefore, timely and effective highway maintenance management is crucial for ensuring the good performance of highways, extending their service life, and ensuring traffic safety.

[0003] Traditional highway maintenance management mainly relies on manual inspections. Maintenance personnel patrol along the highway on foot or by vehicle, and identify pavement diseases by visual observation and simple tool detection. This method is extremely inefficient. For long-distance and large-scale highway networks, it takes a large amount of manpower, material resources, and time to conduct a comprehensive inspection once; moreover, the results of manual detection are greatly affected by factors such as the experience, fatigue level of maintenance personnel, and consistency of detection standards, with strong subjectivity, and it is easy to miss inspections and misjudgments, and it is impossible to accurately and comprehensively grasp the pavement disease conditions.

[0004] Based on this, the present invention provides a highway maintenance management system and method based on AI deep learning technology to solve the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a highway maintenance management system and method based on AI deep learning technology, which deeply integrates AI deep learning technology with highway maintenance management. By analyzing highway images through an intelligent recognition model, pavement disease information can be accurately obtained. By calculating the comprehensive evaluation value and relative change rate of pavement diseases at different detection times, the development trend of pavement diseases can be accurately evaluated. According to the magnitude relationship between the comprehensive evaluation values of pavement diseases at the first detection time and the second detection time, different sorting methods are used to sort the maintenance of each target area section, making the maintenance decision management more scientific and reasonable.

[0006] To achieve the above purpose, the present invention provides the following technical solutions: The first aspect of the present invention: provides a highway maintenance management system based on AI deep learning technology, including an information collection unit, an information processing unit, and a maintenance management unit, wherein: The information acquisition unit is used to obtain the highway image information of the original target area section, the highway image information of the target area section at the first detection moment, and the highway image information of the target area section at the second detection moment; The information processing unit uses an image recognition model to recognize the highway image information at the first detection moment, obtains the first highway recognition result, compares the first highway recognition result with the highway image information of the original target area section to obtain the first comparison result, and uses the image recognition model to recognize the highway image information at the second detection moment, obtains the second highway recognition result, compares the second highway recognition result with the first highway recognition result to obtain the second comparison result, and compares the first comparison result and the second comparison result to perform maintenance ranking on each target area section to obtain a maintenance ranking set, and performs maintenance on each target area section according to the maintenance ranking set. The information processing unit is connected to the information acquisition unit; The maintenance management unit is used to display the received maintenance information and send corresponding warning information according to the received maintenance information. The maintenance management unit is connected to the information processing unit.

[0007] The present invention is further configured as follows: The information acquisition unit includes an image acquisition module, a time recording module, and a first communication module, where: The image acquisition module is used to obtain the highway image information of the target area section; The time recording module is used to mark the time of the obtained highway image information. The time recording module is connected to the image acquisition module; The first communication module is used to realize the information interaction between the information acquisition unit and the information processing unit. The first communication module is connected to the time recording module.

[0008] The present invention is further configured as follows: The information processing unit includes a second communication module, an image recognition module, a database module, a first comparison module, a second comparison module, and a maintenance ranking module, where: The second communication module is used to realize the information interaction between the information processing unit and the information acquisition unit and the maintenance management unit; The image recognition module uses an image recognition model to recognize the highway image information at the first detection moment and the second detection moment, and updates and trains the image recognition model. The image recognition module is connected to the second communication module; The database module is used to store the received information. The database module is connected to both the second communication module and the image recognition module; The first comparison module uses an image recognition model to recognize the highway image information at the first detection moment, obtains the first highway recognition result, compares the first highway recognition result with the highway image information of the original target area section, and obtains the first comparison result. The first comparison module is connected to the image recognition module. Among them, the process of obtaining the first comparison result is as follows: Calculate the comprehensive pavement distress evaluation values of the original target area section and the first detection moment respectively, , where, , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes and the area of potholes respectively; Compare the comprehensive pavement distress evaluation value at the first detection moment with the comprehensive pavement distress evaluation value of the original target area section, and calculate and obtain the first relative change rate , where, is the time interval between the first detection moment and the passing moment of the original target area section, and the first comparison result is obtained; The second comparison module uses an image recognition model to recognize the highway image information at the second detection moment, obtains the second highway recognition result, compares the second highway recognition result with the first highway recognition result, and obtains the second comparison result. The second comparison module is connected to the first comparison module. Among them, the process of obtaining the second comparison result is as follows: Calculate the comprehensive pavement distress evaluation value at the second detection moment; Compare the comprehensive pavement distress evaluation value at the second detection moment with the comprehensive pavement distress evaluation value at the first detection moment, and calculate and obtain the second relative change rate , where, is the time interval between the second detection moment and the first detection moment, and the second comparison result is obtained; The maintenance sorting module is used to compare the first comparison result and the second comparison result, sort the road sections in each target area for maintenance, obtain a maintenance sorting set, and perform maintenance on the road sections in each target area according to the maintenance sorting set. The maintenance sorting module is connected to the second communication module, the database module, the first comparison module, and the second comparison module. Among them, the process of maintenance sorting is as follows: If the comprehensive evaluation value of road surface diseases at the first detection time < the comprehensive evaluation value of road surface diseases at the second detection time , calculate the relative change rate , and sort according to the size of the relative change rate . If the comprehensive evaluation value of road surface diseases at the first detection time ≥ the comprehensive evaluation value of road surface diseases at the second detection time , then sort according to the size of the comprehensive evaluation value of road surface diseases.

[0009] A further setting of the present invention is that the maintenance management unit includes a third communication module, a maintenance display module, and a warning reminder module, where:[[]] The third communication module is used to realize information interaction between the maintenance management unit and the information processing unit; The maintenance display module is used to display the received information, and the maintenance display module is connected to the third communication module; The warning reminder module issues corresponding maintenance warning information according to the received maintenance sorting information, and the warning reminder module is connected to both the third communication module and the maintenance display module.

[0010] The second aspect of the present invention: It also provides a highway maintenance management method based on AI deep learning technology, including the following steps: Obtain the highway image information of the original target area road section, the highway image information of the target area road section at the first detection time, and the highway image information of the target area road section at the second detection time; Based on the image recognition model, identify the highway image information at the first detection time, obtain the first highway recognition result, compare the first highway recognition result with the highway image information of the original target area road section, and obtain the first comparison result; Based on the image recognition model, identify the highway image information at the second detection time, obtain the second highway recognition result, compare the second highway recognition result with the first highway recognition result, and obtain the second comparison result; Compare the first comparison result and the second comparison result, sort the road sections in each target area for maintenance, obtain a maintenance sorting set, and perform maintenance on the road sections in each target area according to the maintenance sorting set.

[0011] The further setting of the present invention is that the image recognition model is a trained convolutional neural network, and its training process is as follows: Obtain historical highway image information, use the historical highway image information as the input of the convolutional neural network, train the convolutional neural network, and output the pavement disease information of the highway; Among them, the pavement disease information includes the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes.

[0012] The further setting of the present invention is that the process of obtaining the first comparison result is as follows: Calculate the comprehensive evaluation values of the pavement diseases of the original target area section and the first detection time respectively, , where, , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes respectively; Compare the comprehensive evaluation value of the pavement disease at the first detection time with the comprehensive evaluation value of the pavement disease of the original target area section, and calculate and obtain the first relative change rate , where, is the time interval between the first detection time and the passing time of the original target area section, and the first comparison result is obtained.

[0013] The further setting of the present invention is that the process of obtaining the second comparison result is as follows: Calculate the comprehensive evaluation value of the pavement disease at the second detection time; Compare the comprehensive evaluation value of the pavement disease at the second detection time with the comprehensive evaluation value of the pavement disease at the first detection time, and calculate and obtain the second relative change rate , where, is the time interval between the second detection time and the first detection time, and the second comparison result is obtained.

[0014] The further setting of the present invention is that the process of maintenance ranking is as follows: If the comprehensive evaluation value of the pavement disease at the first detection time <Comprehensive evaluation value of road diseases at the second detection time , calculate the relative change difference rate , and sort according to the magnitude of the relative change difference rate ; If the comprehensive evaluation value of road diseases at the first detection time ≥ the comprehensive evaluation value of road diseases at the second detection time , then sort according to the magnitude of the comprehensive evaluation value of road diseases.

[0015] A further setting of the present invention is: It further includes using the highway image information of the target area section at the first detection time and the highway image information of the target area section at the second detection time as historical highway image information to update the convolutional neural network.

[0016] Compared with the prior art, the beneficial effects of the present invention are: The present invention deeply integrates AI deep learning technology with highway maintenance management. By analyzing highway images through an intelligent recognition model, it accurately obtains road disease information. By calculating the comprehensive evaluation value and relative change rate of road diseases at different detection times, it can accurately evaluate the development trend of road diseases. According to the magnitude relationship between the comprehensive evaluation values of road diseases at the first detection time and the second detection time, different sorting methods are used to sort each target area section for maintenance sorting. That is, if the comprehensive evaluation value of road diseases at the first detection time is less than that at the second detection time, calculate the relative change difference rate and sort according to the magnitude of this difference rate, giving priority to sections with a fast disease development speed. If the comprehensive evaluation value of road diseases at the first detection time is greater than or equal to that at the second detection time, then directly sort according to the magnitude of the comprehensive evaluation value of road diseases, that is, first focus on dealing with sections with more serious diseases. Through such sorting, a maintenance sorting set is generated, providing a scientific and reasonable execution order for maintenance management work. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a system diagram of the highway maintenance management system based on AI deep learning technology of the present invention.

[0018] Figure 2 is a system diagram of the information collection unit in the highway maintenance management system based on AI deep learning technology of the present invention.

[0019] Figure 3 is a system diagram of the information processing unit in the highway maintenance management system based on AI deep learning technology of the present invention.

[0020] Figure 4 is a system diagram of the maintenance management unit in the highway maintenance management system based on AI deep learning technology of the present invention.

[0021] Explanation of the reference numerals in the drawings: 100. Information acquisition unit; 110. Image acquisition module; 120. Time recording module; 130. First communication module; 200. Information processing unit; 210. Second communication module; 220. Image recognition module; 230. Database module; 240. First comparison module; 250. Second comparison module; 260. Maintenance sorting module; 300. Maintenance management unit; 310. Third communication module; 320. Maintenance display module; 330. Early warning reminder module. Detailed implementation manners

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention. Embodiment

[0023] As Figures 1-4 shown, this embodiment provides a highway maintenance management system based on AI deep learning technology, including an information acquisition unit 100, an information processing unit 200, and a maintenance management unit 300, where: the information acquisition unit 100 is used to obtain the highway image information of the original target area section, the highway image information of the target area section at the first detection time, and the highway image information of the target area section at the second detection time; the information processing unit 200 uses an image recognition model to recognize the highway image information at the first detection time, obtains the first highway recognition result, compares the first highway recognition result with the highway image information of the original target area section, obtains the first comparison result, and uses the image recognition model to recognize the highway image information at the second detection time, obtains the second highway recognition result, compares the second highway recognition result with the first highway recognition result, obtains the second comparison result, and compares the first comparison result and the second comparison result to perform maintenance sorting on each target area section, obtains a maintenance sorting set, and performs maintenance on each target area section according to the maintenance sorting set. The information processing unit 200 is connected to the information acquisition unit 100; the maintenance management unit 300 is used to display the received maintenance information and send corresponding warning information according to the received maintenance information. The maintenance management unit 300 is connected to the information processing unit 200.

[0024] In this embodiment, it should be noted that the information acquisition unit 100 obtains the highway image information of the original target area section, the highway image information of the target area section at the first detection time, and the highway image information of the target area section at the second detection time, marks the time of each highway image, and then uploads it to the information processing unit 200. The information processing unit 200 trains a convolutional neural network using the stored historical image data to obtain the trained convolutional neural network. The convolutional neural network identifies the input highway image information at the first detection time and outputs the first highway recognition result, that is, the road surface disease information, which includes the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes. The first highway recognition result is compared with the highway image information of the original target area section to obtain the first comparison result. Then, the convolutional neural network identifies the input highway image information at the second detection time, outputs the second highway recognition result, compares the second highway recognition result with the first highway recognition result to obtain the second comparison result, and compares the first comparison result and the second comparison result to sort the maintenance of each target area section to obtain a maintenance sorting set. The maintenance of each target area section is carried out according to the maintenance sorting set, and the sorted maintenance information is fed back to the maintenance management unit 300. The maintenance management unit 300 displays the maintenance information for the management personnel to view. At the same time, when receiving the maintenance information, a corresponding warning information will be issued to provide to the management personnel.

[0025] In the present invention, the information acquisition unit 100 includes an image acquisition module 110, a time recording module 120, and a first communication module 130, wherein: the image acquisition module 110 is used to obtain the highway image information of the target area section; the time recording module 120 is used to mark the time of the obtained highway image information, and the time recording module 120 is connected to the image acquisition module 110; the first communication module 130 is used to realize the information interaction between the information acquisition unit 100 and the information processing unit 200, and the first communication module 130 is connected to the time recording module 120.

[0026] In this embodiment, it should be noted that the image acquisition module 110 obtains the highway image information of the original target area section, the highway image information of the target area section at the first detection time, and the highway image information of the target area section at the second detection time, uploads the above information to the time recording module 120, and the time recording module 120 marks the time and uploads the marked highway image information to the information processing unit 200.

[0027] In the present invention, the information processing unit 200 includes a second communication module 210, an image recognition module 220, a database module 230, a first comparison module 240, a second comparison module 250, and a maintenance sorting module 260, where: The second communication module 210 is used to implement information interaction between the information processing unit 200 and the information collection unit 100 and the maintenance management unit 300; The image recognition module 220 uses an image recognition model to recognize the highway image information at the first detection time and the second detection time, and to update and train the image recognition model. The image recognition module 220 is connected to the second communication module 210; The database module 230 is used to store the received information. The database module 230 is connected to both the second communication module 210 and the image recognition module 220; The first comparison module 240 uses an image recognition model to recognize the highway image information at the first detection time, obtains a first highway recognition result, compares the first highway recognition result with the highway image information of the original target area section, and obtains a first comparison result. The first comparison module 240 is connected to the image recognition module 220. Among them, the process of obtaining the first comparison result is as follows: Calculate the comprehensive pavement distress evaluation values of the original target area section and the first detection time respectively, , where, , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes respectively; Compare the comprehensive pavement distress evaluation value at the first detection time with the comprehensive pavement distress evaluation value of the original target area section , and calculate to obtain the first relative change rate , where, is the time interval between the first detection time and the passing time of the original target area section, and the first comparison result is obtained; The second comparison module 250 uses an image recognition model to recognize the highway image information at the second detection time, obtains a second highway recognition result, compares the second highway recognition result with the first highway recognition result, and obtains a second comparison result. The second comparison module 250 is connected to the first comparison module 240. Among them, the process of obtaining the second comparison result is as follows: Calculate the comprehensive pavement distress evaluation value at the second detection time; The comprehensive evaluation value of road surface diseases at the second detection time and the comprehensive evaluation value of road surface diseases at the first detection time are compared, and the second relative change rate is calculated. In the formula, is the time interval between the second detection time and the first detection time, and the second comparison result is obtained; The maintenance sorting module 260 is used to compare the first comparison result and the second comparison result, sort the road sections in each target area for maintenance, obtain a maintenance sorting set, and maintain the road sections in each target area according to the maintenance sorting set. The maintenance sorting module 260 is connected to the second communication module 210, the database module 230, the first comparison module 240, and the second comparison module 250. Among them, the process of maintenance sorting is as follows: If the comprehensive evaluation value of road surface diseases at the first detection time < the comprehensive evaluation value of road surface diseases at the second detection time , calculate the relative change difference rate , and sort according to the size of the relative change difference rate ; If the comprehensive evaluation value of road surface diseases at the first detection time ≥ the comprehensive evaluation value of road surface diseases at the second detection time , then sort according to the size of the comprehensive evaluation value of road surface diseases.

[0028] In this embodiment, it should be noted that the second communication module 210 receives the information of the information acquisition unit 100. The image recognition module 220 uses a large amount of historical highway image information as input data to train the convolutional neural network, and then outputs the road surface disease information of the highway, including the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes and the area of potholes. Then, the first comparison module 240 first calculates the comprehensive evaluation values of the road surface diseases in the original target area section and at the first detection time respectively. During the calculation process, according to the weight coefficients corresponding to each disease index, each disease index is weighted and summed to obtain the comprehensive evaluation value. Then, by calculating the relative change rate between the two, the change degree of the road surface diseases from the original time to the first detection time is quantified, and the first comparison result is obtained, which intuitively reflects the development of the diseases in this section during this stage. The second comparison module 250 also calculates the comprehensive evaluation value of the road surface diseases at the second detection time, and compares it with this value at the first detection time to calculate the second relative change rate, so as to obtain the second comparison result, further clarifying whether the road surface diseases are intensifying, alleviating or maintaining a stable state in the subsequent time period, and more precisely grasping the disease change trend from the time dimension. The set maintenance sorting module 260 performs maintenance sorting based on the first comparison result and the second comparison result obtained above, that is, if the comprehensive evaluation value of the road surface diseases at the first detection time is less than that at the second detection time, the relative change difference rate is calculated and sorted according to the size of this difference rate, and priority is given to paying attention to the sections with a fast disease development speed; if the comprehensive evaluation value of the road surface diseases at the first detection time is greater than or equal to that at the second detection time, it is directly sorted according to the size of the comprehensive evaluation value of the road surface diseases, that is, the sections with more serious diseases are processed first. Through such sorting, a maintenance sorting set is generated, providing a scientific and reasonable execution order for the maintenance work.

[0029] In the present invention, the maintenance management unit 300 includes a third communication module 310, a maintenance display module 320 and a warning reminder module 330, where: the third communication module 310 is used to realize the information interaction between the maintenance management unit 300 and the information processing unit 200; the maintenance display module 320 is used to display the received information, and the maintenance display module 320 is connected to the third communication module 310; the warning reminder module 330 issues corresponding maintenance warning information according to the received maintenance sorting information, and the warning reminder module 330 is connected to both the third communication module 310 and the maintenance display module 320.

[0030] In this embodiment, it should be noted that the sorted maintenance information is received by the third communication module 310 and transmitted to the maintenance display module 320, and the received information is displayed by the maintenance display module 320 for the management personnel to view the maintenance information. At the same time, the warning reminder module 330 issues corresponding maintenance warning information based on the received maintenance sorting information to remind the maintenance personnel to take corresponding maintenance measures for the key sections in time to ensure the safety and normal use of the highway.

[0031] In addition, this embodiment also provides a highway maintenance management method based on AI deep learning technology, including the following steps: Obtain the highway image information of the original target area section, the highway image information of the target area section at the first detection moment, and the highway image information of the target area section at the second detection moment.

[0032] Based on the image recognition model, recognize the highway image information at the first detection moment, obtain the first highway recognition result, and compare the first highway recognition result with the highway image information of the original target area section to obtain the first comparison result.

[0033] Among them, the image recognition model is a trained convolutional neural network, and its training process is as follows: Obtain historical highway image information, use the historical highway image information as the input of the convolutional neural network, train the convolutional neural network, and output the pavement disease information of the highway; Among them, the pavement disease information includes the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes.

[0034] The process of obtaining the first comparison result is as follows: Calculate the comprehensive evaluation values of the pavement diseases of the original target area section and the first detection moment respectively. , where , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes respectively; Compare the comprehensive evaluation value of the pavement disease at the first detection moment with the comprehensive evaluation value of the pavement disease of the original target area section, and calculate and obtain the first relative change rate , where is the time interval between the first detection moment and the passing time of the original target area section, and the first comparison result is obtained.

[0035] Based on the image recognition model, the highway image information at the second detection moment is recognized to obtain the second highway recognition result. The second highway recognition result is compared with the first highway recognition result to obtain the second comparison result.

[0036] In this embodiment, it should be noted that, as an example, assume that the weight coefficients of each disease index are as follows: the weight of the longitudinal crack number is 0.15, the weight of the transverse crack number is 0.15, the weight of the longitudinal crack area is 0.2, the weight of the transverse crack area is 0.2, the weight of the pothole number is 0.15, the weight of the pothole area is 0.15, the longitudinal crack number is 3, the transverse crack number is 2, the longitudinal crack area is 0.5 square meters, the transverse crack area is 0.3 square meters, the pothole number is 1, and the pothole area is 0.2 square meters. The comprehensive evaluation value of the road surface disease corresponding to the first detection moment is calculated as 0.4, and the time interval between the first detection moment and the passing time of the original target area section is 10 days, then the first relative change rate is calculated as 0.01.

[0037] Among them, the process of obtaining the second comparison result is as follows: Calculate the comprehensive evaluation value of the road surface disease at the second detection moment ; Compare the comprehensive evaluation value of the road surface disease at the second detection moment with the comprehensive evaluation value of the road surface disease at the first detection moment to calculate and obtain the second relative change rate , where is the time interval between the second detection moment and the first detection moment, and the second comparison result is obtained.

[0038] In this embodiment, it should be noted that, as an example, the comprehensive evaluation value of the road surface disease at the second detection moment is calculated as 0.5, and the time interval from the first detection moment is 9 days, then the second relative change rate is calculated as 0.011, and the second comparison result is obtained, showing that the road surface disease is still developing continuously in the following days, and the development speed is slightly faster than that in the previous stage, intuitively presenting the change trend of the disease on this section of the road over time.

[0039] Compare the first comparison result and the second comparison result, sort the maintenance of each target area section to obtain a maintenance sorting set, and perform maintenance on each target area section according to the maintenance sorting set.

[0040] Among them, the process of maintenance sorting is as follows: If the comprehensive evaluation value of the road surface disease at the first detection moment <Comprehensive evaluation value of road surface diseases at the second detection time , calculate the relative change difference rate , and sort according to the magnitude of the relative change difference rate ; If the comprehensive evaluation value of road surface diseases at the first detection time ≥ the comprehensive evaluation value of road surface diseases at the second detection time , then sort according to the magnitude of the comprehensive evaluation value of road surface diseases.

[0041] In addition, it also includes using the road image information of the target area section at the first detection time and the road image information of the target area section at the second detection time as historical road image information to update the convolutional neural network.

[0042] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0043] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art in the relevant technical field can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A highway maintenance management system based on AI deep learning technology, characterized in that, It includes an information collection unit (100), an information processing unit (200) and a maintenance management unit (300), where: The information collection unit (100) is used to obtain the road image information of the original target area section, the road image information of the target area section at the first detection moment, and the road image information of the target area section at the second detection moment; The information processing unit (200) uses an image recognition model to recognize the road image information at the first detection moment, obtains the first road recognition result, compares the first road recognition result with the road image information of the original target area section to obtain the first comparison result, and uses the image recognition model to recognize the road image information at the second detection moment, obtains the second road recognition result, compares the second road recognition result with the first road recognition result to obtain the second comparison result, and compares the first comparison result and the second comparison result to perform maintenance ranking on each target area section to obtain a maintenance ranking set, and performs maintenance on each target area section according to the maintenance ranking set. The information processing unit (200) is connected to the information collection unit (100); The maintenance management unit (300) is used to display the received maintenance information and send corresponding warning information according to the received maintenance information. The maintenance management unit (300) is connected to the information processing unit (200).

2. The highway maintenance management system based on AI deep learning technology according to claim 1, characterized in that, The information collection unit (100) includes an image collection module (110), a time recording module (120) and a first communication module (130), where: The image collection module (110) is used to obtain the road image information of the target area section; The time recording module (120) is used to mark the time of the obtained road image information. The time recording module (120) is connected to the image collection module (110); The first communication module (130) is used to realize the information interaction between the information collection unit (100) and the information processing unit (200). The first communication module (130) is connected to the time recording module (120).

3. The highway maintenance management system based on AI deep learning technology according to claim 1, characterized in that, The information processing unit (200) includes a second communication module (210), an image recognition module (220), a database module (230), a first comparison module (240), a second comparison module (250) and a maintenance ranking module (260), where: The second communication module (210) is used to realize the information interaction between the information processing unit (200) and the information collection unit (100) and the maintenance management unit (300); The image recognition module (220) uses an image recognition model to recognize the road image information at the first detection moment and the second detection moment, and updates and trains the image recognition model. The image recognition module (220) is connected to the second communication module (210); The database module (230) is used to store the received information. The database module (230) is connected to both the second communication module (210) and the image recognition module (220); The first comparison module (240) uses an image recognition model to recognize the highway image information at the first detection moment, obtains the first highway recognition result, compares the first highway recognition result with the highway image information of the original target area section, and obtains the first comparison result. The first comparison module (240) is connected to the image recognition module (220). Among them, the process of obtaining the first comparison result is as follows: Calculate the comprehensive evaluation values of pavement diseases for the original target area road section and the first detection time respectively, , where, , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes and the area of potholes respectively; Compare the comprehensive evaluation value of road surface diseases at the first detection moment with the comprehensive evaluation value of road surface diseases in the original target area section to calculate and obtain the first relative change rate , where is the time interval between the first detection moment and the passing time of the original target area section, and the first comparison result is obtained; The second comparison module (250) uses an image recognition model to recognize the highway image information at the second detection moment, obtains the second highway recognition result, compares the second highway recognition result with the first highway recognition result, and obtains the second comparison result. The second comparison module (250) is connected to the first comparison module (240). Among them, the process of obtaining the second comparison result is as follows: Calculate the comprehensive evaluation value of road surface diseases at the second detection moment ; The comprehensive evaluation value of road surface diseases at the second detection time and the comprehensive evaluation value of road surface diseases at the first detection time are compared to calculate and obtain the second relative change rate , where is the time interval between the second detection time and the first detection time, and the second comparison result is obtained; The maintenance ranking module (260) is used to compare the first comparison result and the second comparison result, rank the maintenance of each target area section, obtain a maintenance ranking set, and perform maintenance on each target area section according to the maintenance ranking set. The maintenance ranking module (260) is connected to the second communication module (210), the database module (230), the first comparison module (240), and the second comparison module (250). Among them, the process of maintenance ranking is as follows: If the comprehensive evaluation value of road surface diseases at the first detection time <is less than the comprehensive evaluation value of road surface diseases at the second detection time , calculate the relative change difference rate , and sort according to the magnitude of the relative change difference rate ; If the comprehensive evaluation value of road surface diseases at the first detection time ≥ the comprehensive evaluation value of road surface diseases at the second detection time , then sort according to the size of the comprehensive evaluation value of road surface diseases.

4. The highway maintenance management system based on AI deep learning technology according to claim 1, characterized in that The maintenance management unit (300) includes a third communication module (310), a maintenance display module (320), and a warning reminder module (330), where: The third communication module (310) is used to implement information interaction between the maintenance management unit (300) and the information processing unit (200); The maintenance display module (320) is used to implement the received information. The maintenance display module (320) is connected to the third communication module (310); The warning reminder module (330) issues corresponding maintenance warning information according to the received maintenance ranking information. The warning reminder module (330) is connected to both the third communication module (310) and the maintenance display module (320).

5. A highway maintenance management method based on AI deep learning technology, characterized in that, Including the following steps: Obtain the highway image information of the original target area section, the highway image information of the target area section at the first detection moment, and the highway image information of the target area section at the second detection moment; Based on the image recognition model, recognize the highway image information at the first detection moment, obtain the first highway recognition result, compare the first highway recognition result with the highway image information of the original target area section, and obtain the first comparison result; Based on the image recognition model, recognize the highway image information at the second detection moment, obtain the second highway recognition result, compare the second highway recognition result with the first highway recognition result, and obtain the second comparison result; Compare the first comparison result and the second comparison result, rank the maintenance of each target area section, obtain a maintenance ranking set, and perform maintenance on each target area section according to the maintenance ranking set.

6. The highway maintenance management method based on AI deep learning technology according to claim 5, wherein, The image recognition model is a trained convolutional neural network, and its training process is as follows: Obtain historical highway image information, use the historical highway image information as the input of a convolutional neural network, train the convolutional neural network, and output the pavement disease information of the highway; Among them, the pavement disease information includes the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes, and the area of potholes.

7. The highway maintenance management method based on AI deep learning technology according to claim 5, characterized in that The process of obtaining the first comparison result is as follows: Calculate the comprehensive evaluation values of pavement diseases for the original target area section and the first detection moment respectively, , where, , , , , and are all weight coefficients, , , , , and are the number of longitudinal cracks, the number of transverse cracks, the area of longitudinal cracks, the area of transverse cracks, the number of potholes and the area of potholes respectively; The comprehensive evaluation value of road surface diseases at the first detection time and the comprehensive evaluation value of road surface diseases in the original target area section are compared, and the first relative change rate is calculated. In the formula, is the time interval between the first detection time and the passing time of the original target area section, and the first comparison result is obtained.

8. The highway maintenance management method based on AI deep learning technology according to claim 5, characterized in that The process of obtaining the second comparison result is as follows: Calculate the comprehensive evaluation value of road surface diseases at the second detection moment ; The comprehensive evaluation value of road surface diseases at the second detection time and the comprehensive evaluation value of road surface diseases at the first detection time are compared, and the second relative change rate is calculated. In the formula, is the time interval between the second detection time and the first detection time, and the second comparison result is obtained.

9. The highway maintenance management method based on AI deep learning technology according to claim 5, wherein The process of maintenance ranking is as follows: If the comprehensive evaluation value of road surface diseases at the first detection time <is less than the comprehensive evaluation value of road surface diseases at the second detection time , calculate the relative change difference rate , and sort according to the magnitude of the relative change difference rate ; If the comprehensive evaluation value of road surface diseases at the first detection time ≥ the comprehensive evaluation value of road surface diseases at the second detection time , then sort according to the size of the comprehensive evaluation value of road surface diseases.

10. The highway maintenance management method based on AI deep learning technology according to claim 5, characterized in that, It also includes using the highway image information of the target area section at the first detection time and the highway image information of the target area section at the second detection time as historical highway image information to update the convolutional neural network.

Images