A method and apparatus for maintaining a roadway within a tunnel

Abstract

The application provides a tunnel pavement maintenance method and device, and belongs to the technical field of road maintenance. The method obtains a plurality of real-time pavement monitoring images and corresponding acquisition information from an image acquisition device, inputs each pavement monitoring image into a damaged pavement identification model, and determines whether the monitoring pavement corresponding to the pavement monitoring image is a pavement to be maintained. If so, the corresponding tunnel location and pavement damage time are determined according to the acquisition information. Based on the historical driving data of the tunnel corresponding to the pavement to be maintained and the historical driving data of the replaceable path of the tunnel, a traffic flow weight set of the tunnel is determined. The reachable destination of the replaceable path is located on the downlink route and / or uplink route of the tunnel location or on the reachable extension line of the downlink route and / or the reachable extension line of the uplink route of the tunnel. Based on the pavement monitoring image, the pavement damage time and the traffic flow weight set, the pavement maintenance information of the tunnel is determined, and the pavement maintenance information is sent to a maintenance terminal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road maintenance, in particular to a road maintenance method and device in a tunnel. BACKGROUND

[0002] With the continuous development of the transportation industry, roads are everywhere, and in the past, roads were mostly laid on the ground, trying to bypass complex terrain such as hills and ravines. Today, with the development of road construction technology, in order to shorten the distance between two places, people have begun to dig tunnels through mountains.

[0003] The emergence of tunnels reduces travel time and brings convenience to people. Under the long-term high-intensity use of tunnels, the road surface of the tunnel will be damaged, and the driving environment of the tunnel is very complex. If the tunnel cannot be reasonably arranged for repair, it will not only affect the efficiency of the tunnel, but also may threaten the safety of the driver. SUMMARY

[0004] To solve the above problems, the embodiments of the present application provide a road maintenance method and device in a tunnel, which is used to provide reasonable maintenance information for repairing the tunnel, and to protect the efficiency of the tunnel and the safety of the driver.

[0005] In one aspect, the embodiments of the present application provide a road maintenance method in a tunnel, which comprises:

[0006] Obtain a plurality of real-time road monitoring images and corresponding acquisition information from an image acquisition device. Input each road monitoring image into a preset damaged road identification model to determine whether the monitoring road corresponding to the road monitoring image is a road to be maintained. In the case where the monitoring road corresponding to the road monitoring image is the road to be maintained, determine the tunnel position and road damage time corresponding to the road to be maintained according to the acquisition information of the road monitoring image. Based on the historical driving data of the tunnel corresponding to the road to be maintained and the historical driving data of the replaceable path of the tunnel, determine a traffic flow weight set of the tunnel. Wherein, the reachable destination of the replaceable path is on the downlink and / or uplink of the tunnel position or on the reachable extension line of the downlink and / or uplink of the tunnel. Based on the road monitoring image, the road damage time and the traffic flow weight set, determine the road maintenance information of the tunnel, and send the road maintenance information to a maintenance terminal.

[0007] In an implementation manner of the present application, the damaged pavement in each pavement monitoring image is determined by a damaged pavement identification model. A damaged label corresponding to each damaged image frame is identified. The damaged label includes at least one or more of the following: settlement, rut, cracking, shrinkage, and water seepage. Based on the damaged label, it is determined whether the damage degree of the monitoring pavement is greater than a preset threshold. In a case where it is determined that the damage degree of the monitoring pavement is greater than the preset threshold, the monitoring pavement is regarded as a pavement to be maintained.

[0008] In an implementation manner of the present application, a damage degree comparison image corresponding to each damaged label is determined. There are at least two damage degree comparison images corresponding to one damaged label. The preset threshold is within a damage degree value corresponding interval of the two damage degree comparison images. The damaged image frame corresponding to the damaged label is matched with each damage degree comparison image, so as to determine a damage degree value corresponding to the damaged label according to a matching result. The damage degree value is determined according to the matched damage degree comparison image. Whether the damage degree of the monitoring pavement is greater than the preset threshold is determined according to the damage weight corresponding to each damaged label and the corresponding damage degree value.

[0009] In an implementation manner of the present application, a first average traffic flow of the tunnel is determined according to historical driving data of the tunnel. A second average traffic flow corresponding to each alternative path is determined according to historical driving data of each alternative path. A first traffic flow weight corresponding to each second average traffic flow is determined based on the first average traffic flow and each second average traffic flow. Third average traffic flows of a plurality of preset time periods of the tunnel are determined based on the historical driving data of the tunnel, so as to determine a plurality of second traffic flow weights of the tunnel according to the third average traffic flows. A traffic flow weight set is determined according to the first traffic flow weight of the tunnel and each second traffic flow weight.

[0010] In an implementation manner of the present application, a traffic flow average value of the first average traffic flow and each second average traffic flow is determined. A first traffic flow weight corresponding to the first average traffic flow and each second average traffic flow is determined according to the traffic flow average value, the first average traffic flow and each second average traffic flow.

[0011] In an implementation manner of the present application, a road damage type of a road surface monitoring image is determined through a preset damage type identification model. According to the road damage type, a repair duration of the tunnel is determined. According to the road damage time, the vehicle flow weight set and the vehicle flow weight set of each replaceable path, a third vehicle flow weight sequence in a first preset time is determined. In each third vehicle flow weight sequence, the time period in which the third vehicle flow weight of the tunnel meets the minimum weight value is determined as the to-be-determined repair period. Whether the first vehicle flow weight of the replaceable path is greater than the first vehicle flow weight of the tunnel is determined in the to-be-determined repair period. In the case where the first vehicle flow weight of the replaceable path is greater than the first vehicle flow weight of the tunnel, the to-be-determined repair period is taken as the repair period. Otherwise, the minimum weight value is removed, and the time period corresponding to the minimum weight value in the third vehicle flow weight sequence after removal is determined as the to-be-determined repair period, until the repair period is determined. In the case where the repair period is less than the repair duration, the repair period in Mth preset time is determined, until the sum of each repair period is greater than the repair duration, and each repair period is determined as the road maintenance information. Wherein, M is a natural number.

[0012] In an implementation manner of the present application, a plurality of road damage images and their corresponding road damage types in a block chain platform are obtained. Each road damage image is taken as a training sample, and its corresponding road damage type is taken as a training label of the training sample, and input into a preset neural network identification model. In the case where the loss function of the neural network identification model is less than a preset value, the trained neural network identification model is determined as a damage type identification model.

[0013] In an implementation manner of the present application, a plurality of sample data images are obtained, and each sample data image and a corresponding damage degree comparison image are input into a preset classifier for training the classifier. The classifier divides the sample data into a plurality of image combinations consistent with the number of damage degree comparison images. The damage degree value in the image combination matches the damage degree value of the corresponding damage degree comparison image. Through the classifier, the image combination corresponding to the damage image frame is matched to determine the damage degree value corresponding to the damage image frame according to the matched image combination.

[0014] In an implementation manner of the present application, in the case where the damage degree of the monitored road surface is less than or equal to a preset threshold, the damage image frame in the monitored road surface is sent to the maintenance terminal.

[0015] On the other hand, the present application provides a road maintenance device in a tunnel, which comprises:

[0016] at least one processor; and a memory connected to the at least one processor in communication. Wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0017] Obtain a plurality of real-time road surface monitoring images and corresponding acquisition information from an image acquisition device. Each road surface monitoring image is input into a preset damaged road surface identification model to determine whether the monitoring road surface corresponding to the road surface monitoring image is a road surface to be maintained. In the case where the monitoring road surface corresponding to the road surface monitoring image is determined to be the road surface to be maintained, the tunnel position corresponding to the road surface to be maintained and the road damage time are determined according to the acquisition information of the road surface monitoring image. Based on the historical driving data of the tunnel corresponding to the road surface to be maintained and the historical driving data of the replaceable path of the tunnel, a traffic flow weight set of the tunnel is determined. The reachable destination of the replaceable path is on the downlink and / or uplink of the tunnel position or on the reachable extension of the downlink and / or uplink of the tunnel. Based on the road surface monitoring image, the road damage time, and the traffic flow weight set, road maintenance information of the tunnel is determined, and the road maintenance information is sent to a maintenance terminal.

[0018] Through the above scheme, the road surface monitoring image and acquisition information of the tunnel are used to determine whether the tunnel road needs to be maintained, and then the historical driving data of the tunnel and the related information of the position are used to determine the traffic flow weight set of the tunnel and its replaceable path. Further, the image recognition technology is used to determine the road repair duration and determine the repair period of the tunnel road, so as to avoid unreasonable repair of the tunnel and affect the tunnel passing efficiency. At the same time, timely repair can guarantee the safety of the driver, and also improve the experience of the driver driving in the tunnel. The present application considers a plurality of replaceable paths of the tunnel, which can relieve the traffic pressure of the road maintenance tunnel and further improve the maintenance and repair efficiency of the tunnel. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of the present application and illustrate the illustrative embodiments of the present application and its description serve to explain the present application, and do not constitute improper limitations on the present application. In the drawings:

[0020] Figure 1 It is a flowchart of a road maintenance method in a tunnel in an embodiment of the present application;

[0021] Figure 2 It is a schematic diagram of a road maintenance method in a tunnel in an embodiment of the present application;

[0022] Figure 3 It is another schematic diagram of a road maintenance method in a tunnel in an embodiment of the present application;

[0023] Figure 4 It is still another schematic diagram of a road maintenance method in a tunnel in an embodiment of the present application;

[0024] Figure 5 FIG. 6 is another flowchart of a tunnel pavement maintenance method according to an embodiment of the present application;

[0025] Figure 6 FIG. 7 is a structural diagram of a tunnel pavement maintenance device according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in detail with reference to the embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0027] The embodiments of the present application provide a tunnel pavement maintenance method and device to provide reasonable maintenance information for repairing the tunnel, to ensure the tunnel passing efficiency and the life safety of the driver.

[0028] The various embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0029] The embodiments of the present application provide a tunnel pavement maintenance method, as shown in FIG. 1, which can include steps S101-S105: Figure 1

[0030] S101, the server acquires a plurality of real-time pavement monitoring images and corresponding acquisition information from the image acquisition device.

[0031] The image acquisition device can be a camera arranged in the tunnel, or a handheld image acquisition device of the tunnel inspection personnel. The real-time pavement monitoring image refers to the image sent by the image acquisition device to the server, and the acquisition information is the acquisition time and acquisition location of the real-time pavement monitoring image.

[0032] It should be noted that the server is an example of the execution subject of the tunnel pavement maintenance method, and the execution subject is not limited to the server. The present application does not make specific limitations thereto.

[0033] S102, the server inputs each pavement monitoring image into a preset damaged pavement recognition model to determine whether the monitoring pavement corresponding to the pavement monitoring image is a pavement to be maintained.

[0034] The damaged pavement recognition model can be a neural network model, such as AlexNet, ZFNET, VGG, etc.

[0035] ​In this embodiment, the server inputs each road surface monitoring image into a preset damaged road surface identification model to determine whether the monitored road surface corresponding to the road surface monitoring image is a road surface requiring maintenance, specifically including:

[0036] First, the server uses a damaged road surface recognition model to determine the damaged image frame corresponding to the damaged road surface in each road surface monitoring image.

[0037] The damaged road surface recognition model can be trained using several road surface damage image samples. After multiple training sessions, the model can identify the damaged image frame corresponding to the damaged road surface in the road monitoring image. In this embodiment, since the number of image acquisition devices is not limited, multiple image acquisition devices may acquire road surface images from different directions for the same road surface. Therefore, there are multiple road monitoring images, which are images containing several image frames.

[0038] Next, the server uses the recognition operation of the damaged road surface recognition model to determine the damage label corresponding to each damaged image frame.

[0039] The damage label must include at least one or more of the following: subsidence, ruts, cracks, shrinkage cracks, and water seepage.

[0040] Damage labels are determined for the road damage image samples used to train the damaged road surface recognition model. Different damage labels correspond to multiple road damage image samples to ensure the recognition accuracy of the damaged road surface recognition model.

[0041] Then, based on each damage tag, the server determines whether the degree of damage to the monitored road surface exceeds a preset threshold.

[0042] Specifically, the server determines whether the degree of damage to the monitored road surface exceeds a preset threshold based on each damage tag, including the following steps:

[0043] The first step is for the server to determine the degree of damage for each damaged tag by comparing it with the image.

[0044] In this scenario, a damaged label must exist in at least two images comparing the degree of damage. The preset threshold is located within the range corresponding to the damage degree values ​​of the two images.

[0045] like Figure 2 As shown, the damage level comparison images are 201, 202, 203, and 204, and the damaged image frame corresponding to the damage label is 205.

[0046] The second step is for the server to match the damaged image frame corresponding to the damaged tag with images of each degree of damage, so as to determine the degree of damage corresponding to the damaged tag based on the matching results.

[0047] The damage degree value is determined according to the matched damage degree comparison image.

[0048] Each damage degree comparison image corresponds to a unique damage degree value, for example, the damage degree value of 201 is a, the damage degree value of 202 is b, the damage degree value of 203 is c, and the damage degree value of 204 is d. The server matches the damage image frame of the damage label with each damage degree comparison image to obtain the damage degree value of the damage label.

[0049] In some embodiments of the present application, the server matches the damage image frame corresponding to the damage label with each damage degree comparison image to determine the damage degree value corresponding to the damage label according to the matching result, specifically including:

[0050] The server obtains a plurality of sample data images, and inputs each sample data image and the corresponding damage degree comparison image into a preset classifier to train the classifier.

[0051] The classifier divides the sample data into at least a plurality of image combinations consistent with the number of damage degree comparison images. The damage degree value in the image combination matches the damage degree value of the corresponding damage degree comparison image.

[0052] The classifier can use a support vector machine classification algorithm, input the sample data image and each damage degree comparison image, and output the classification of the sample data image and each damage degree comparison image, so as to classify the sample data image into image combinations. The number of classified image combinations is at least equal to the number of damage degree comparison images. For example, the input damage degree comparison image with different damage degree values is 4, the classifier divides the sample data image into at least 4 image combinations to distinguish the sample data image, so as to realize feature classification of images with different damage degrees.

[0053] The support vector machine classification algorithm can realize the classification of a nonlinear model, establish a binary classifier between each two classes, and accurately classify the sample data image. Different sample data images match damage degree comparison images with different damage degree values. Until the accuracy of the classifier training is higher than a preset accuracy, for example, the accuracy of the classified image combination reaches 99%, the training of the classifier is determined to be completed.

[0054] Then, the server matches the image combination corresponding to the damage image frame through the classifier to determine the damage degree value corresponding to the damage image frame according to the matched image combination.

[0055] The server obtains the damage degree value of the damage image frame according to the image combination of the damage image frame divided by the classifier after the classification.

[0056] In a third step, the server determines whether the damage degree of the monitored road surface is greater than a preset threshold according to the damage weight corresponding to each damage label and the corresponding damage degree value.

[0057] In the embodiments of the present application, the damage weight is obtained according to the damage label. The damage weight can be pre-set by the user. Different damage weights represent the severity of different damage labels. For example, the damage weight of a sink is 0.2, the damage weight of a rut is 0.15, and so on. The server calculates the product of the damage weight and the damage degree value, and compares the product value with the preset threshold, so as to determine whether the damage degree of the monitored road surface is greater than the preset threshold.

[0058] For example, the damage weight is A, the damage degree value is B, and the server can determine the damage degree of the monitored road surface by comparing the size relationship between A*B and the preset threshold C.

[0059] Finally, the server determines that the monitored road surface is the road surface to be maintained in the case where the damage degree of the monitored road surface is greater than the preset threshold.

[0060] In the embodiments of the present application, the preset threshold is set according to the actual use of the user. The preset threshold represents that the damage degree reaches or approaches the degree that needs to be maintained and repaired. The specific value of the preset threshold is not limited in the present application.

[0061] In an embodiment of the present application, in the case where the damage degree of the monitored road surface is less than or equal to the preset threshold, the server sends the damage image frame in the monitored road surface to the maintenance terminal.

[0062] In the actual use of the road surface maintenance method in the tunnel, the maintenance terminal can real-time view the actual situation of the monitored road surface. In the case where there is a damage image frame in the monitored road surface, even if immediate maintenance is not needed, the corresponding expert of the maintenance terminal can make a judgment, so as to avoid that the road surface with a small damage degree is not maintained in time, the damage degree of the road surface is aggravated, and the road surface is seriously damaged. At the same time, to a certain extent, the subsequent damage can be avoided from being aggravated, the communication efficiency of the tunnel is affected, and the life safety of the driver is threatened.

[0063] In S103, in the case where the server determines that the monitored road surface corresponding to the road surface monitoring image is the road surface to be maintained, the server determines the tunnel position and the road surface damage time corresponding to the road surface to be maintained according to the collection information of the road surface monitoring image.

[0064] In other words, after the server determines that the monitored road surface is the road surface to be maintained, the server can determine the tunnel position where the road surface to be maintained is located and the first moment when the road surface to be maintained is damaged, i.e., the road surface damage time, according to the collection information.

[0065] S104, the server determines a set of traffic weight of the tunnel based on historical driving data of the tunnel corresponding to the road to be maintained and historical driving data of the alternative path of the tunnel.

[0066] Wherein, the reachable destination of the alternative path is on the downlink route and / or uplink route of the tunnel location or on the reachable extension of the downlink route and / or the uplink route of the tunnel.

[0067] The relationship between the alternative path and the tunnel is as shown in Figure 3 (bidirectional lane), Figure 4 (unidirectional lane). When the tunnel is a bidirectional lane, there is another route with the same starting point and destination as the tunnel in the driving direction of the current tunnel, and the other route does not contain the current tunnel, and the other route can also contain tunnels, highways, passages, etc. Or in the driving direction of the current tunnel, the extension of a certain route is the reachable destination of the current tunnel.

[0068] In the embodiment of the application, the server determines a set of traffic weight of the tunnel based on historical driving data of the tunnel corresponding to the road to be maintained and historical driving data of the alternative path of the tunnel, which specifically includes:

[0069] First, the server determines the first average traffic of the tunnel according to the historical driving data of the tunnel.

[0070] The historical driving data can be the vehicle driving data of the tunnel in the past month or year, for example, in the past month, the average traffic of the tunnel is N, and N is taken as the first average traffic.

[0071] Secondly, the server determines the corresponding second average traffic of each alternative path according to the historical driving data of each alternative path.

[0072] The second average traffic is the average traffic of the alternative path in the same time period as the first average traffic.

[0073] Thirdly, the server determines the corresponding first traffic weight based on the first average traffic and each second average traffic.

[0074] Specifically, the server determines the corresponding first traffic weight based on the first average traffic and each second average traffic, which includes the following steps:

[0075] First, the server determines the traffic average of the first average traffic and each second average traffic.

[0076] For example, the first average traffic is N, and each second evaluation traffic includes:

[0077] M1, M2, M3,..., Mn, and the traffic average is:

[0078] V = (N + M1 + M2 +... + Mn) / (n + 1). Wherein, n is a natural number.

[0079] Second, according to the average traffic flow, the first average traffic flow and each second average traffic flow, the first average traffic flow and each second average traffic flow are determined.

[0080] In the embodiment of the application, according to the average traffic flow, the first average traffic flow and each second average traffic flow, the formula for determining the first traffic flow weight is as follows:

[0081]

[0082] Wherein, β is the first traffic flow weight, X is the first average traffic flow or the second average traffic flow.

[0083] Subsequently, the server determines the third average traffic flow of the tunnel in a plurality of preset time periods based on the historical driving data of the tunnel, so as to determine a plurality of second traffic flow weights of the tunnel according to the third average traffic flow.

[0084] The server can determine the third average traffic flow of the tunnel in each time period such as weekdays and holidays according to the historical driving data of the tunnel, for example, from 1 o'clock to 2 o'clock on a certain day, the third average traffic flow is q1; from 2 o'clock to 3 o'clock, the third average traffic flow is q1; from 3 o'clock to 4 o'clock, the third average traffic flow is q2; from 4 o'clock to 5 o'clock, the third average traffic flow is q2... The preset time period can be divided according to the third average traffic flow, for example, the third average traffic flow from 1 o'clock to 2 o'clock and from 2 o'clock to 3 o'clock is the same, and the time is adjacent, which can be a preset time period; according to the above scheme, it can be known that 3 o'clock to 4 o'clock and 4 o'clock to 5 o'clock are also a preset time period. According to the proportion of each third average traffic flow in the total traffic flow of each preset time period on a certain day, the second traffic flow weight is determined, for example, the third average traffic flow is Y1, and the total traffic flow is Y0, then the weight is Y1 / Y0.

[0085] Subsequently, the server determines the traffic flow weight set according to the first traffic flow weight and each second traffic flow weight of the tunnel.

[0086] The traffic flow weight set is a weight set containing the first traffic flow weight and each second traffic flow weight.

[0087] S105, the server determines the road maintenance information of the tunnel based on the road surface monitoring image, the road damage time and the traffic flow weight set, and sends the road maintenance information to the maintenance terminal.

[0088] The maintenance terminal can be a handheld terminal of a maintenance personnel, such as a mobile phone, an iPad, or the like, or a notebook computer, and the present application does not make a specific limitation thereon.

[0089] In the embodiment of the present application, the server determines the tunnel pavement maintenance information based on the pavement monitoring image, the pavement damage time, and the traffic weight, such as Figure 5 as shown, specifically comprising the following steps:

[0090] S501, the server determines the pavement damage type of the pavement monitoring image through the preset damage type recognition model.

[0091] In the embodiment of the present application, the tunnel pavement maintenance method is applied to a pre-built blockchain platform, and the blockchain platform includes a plurality of sub-nodes, and the sub-nodes are used to upload pavement damage images and corresponding pavement damage types. Before determining the pavement damage type of the pavement monitoring image through the preset damage type recognition model, the method further comprises:

[0092] Firstly, the server obtains a plurality of pavement damage images and corresponding pavement damage types in the blockchain platform.

[0093] Each sub-node in the blockchain platform can upload a pavement damage image and a pavement damage type of the pavement damage image, and the sub-node can be a user's mobile phone. For example, a user can take a photo and send the photo and the corresponding pavement damage type to the blockchain platform when the user finds pavement damage during driving.

[0094] Then, the server inputs each pavement damage image as a training sample and the corresponding pavement damage type as a training label of the training sample into the preset neural network recognition model.

[0095] The neural network recognition model can be the same as the neural network model used in the above-mentioned damage pavement recognition model, or can be different, and the present application does not make a specific limitation thereon.

[0096] In the case that the loss function of the neural network recognition model is less than the preset value, the server determines that the trained neural network recognition model is the damage type recognition model.

[0097] In the case that the loss function of the neural network recognition model is less than the preset value, i.e., the accuracy of the neural network recognition model meets the requirements, the server uses the trained neural network recognition model as the damage type recognition model.

[0098] S502, the server determines the repair duration of the tunnel according to the pavement damage type.

[0099] In the embodiments of the present application, the road damage type can have a damage type table, which contains the corresponding relationship between the road damage type and the repair duration thereof, and the damage type table can be generated by the user in advance.

[0100] In S503, the server determines a third traffic weight sequence in a first preset time according to the road damage time, the traffic weight set, and the traffic weight set of the alternative path.

[0101] The third traffic weight sequence in the first preset time refers to a time period after the road damage time, such as the time period containing three sub-periods S1, S2, and S3. The first traffic weight of the tunnel is multiplied by the second traffic weight to obtain the third traffic weight, and the first weight sequence is generated in time sequence. The first traffic weight of each alternative path is multiplied by the second traffic weight of the alternative path (the method for obtaining the second traffic weight of the alternative path can refer to the method for obtaining the second traffic weight of the tunnel) to obtain the third traffic weight of the alternative path, and the second weight sequence is generated in time sequence.

[0102] According to the time sequence, the server obtains the weights in the first weight sequence of the tunnel and each second weight sequence of the alternative path corresponding to the same time period S1, and arranges the weights in order of weight size to obtain the third traffic weight sequence.

[0103] In S504, the server determines that the time period in which the third traffic weight of the tunnel in the third traffic weight sequence satisfies the minimum weight value is the to-be-determined repair period.

[0104] That is, in the same time period, the minimum value of the third traffic weight sequence is the third traffic weight of the tunnel, and then the time period is the to-be-determined repair period.

[0105] In S505, the server determines whether the first traffic weight of the alternative path is greater than the first traffic weight of the tunnel in the to-be-determined repair period.

[0106] The server will compare the first traffic weight of the alternative path with the first traffic weight of the tunnel after determining the to-be-determined repair period.

[0107] In S506, the server takes the to-be-determined repair period as the repair period when the first traffic weight of the alternative path is greater than the first traffic weight of the tunnel.

[0108] That is, in the period when the driving vehicles of the alternative path are more than the driving vehicles of the tunnel, as the repair period, the comparison effect can be better through the comparison of the weights, and the weight can refer to the importance of the path, if the importance of the tunnel in the pending repair period is higher than the alternative path, maintaining the tunnel will affect the driving of normal vehicles.

[0109] S507, otherwise, the server eliminates the minimum weight, and determines that the time period corresponding to the minimum weight in the third traffic weight sequence after elimination is the pending repair period, until the repair period is determined.

[0110] That is, the server will eliminate the sub-time period of the minimum weight in the case that the first traffic weight of the alternative path is less than the first traffic weight of the tunnel, and continue to execute the step of S504.

[0111] S508, in the case that the repair period is less than the repair duration, the server determines the repair period in the Mth preset time, until the sum of each repair period is greater than the repair duration, and determines each repair period as the road maintenance information.

[0112] Wherein, M is a natural number. That is, the repair period can be 2-3 points of the first day, 3-4 points of the second day, and the repair duration is 10 hours, so that the server needs to obtain the sum of the multiple repair periods greater than the repair duration, so as to complete the maintenance of the tunnel, and each repair period is used as the road maintenance information.

[0113] Through the above scheme, the road surface monitoring image and the collection information of the tunnel are used to determine whether the tunnel road needs to be maintained, and then the historical driving data and the related information of the position of the tunnel are used to determine the traffic weight set of the tunnel and the alternative path. Further, the image recognition technology is used to determine the road repair duration, and the repair period of repairing the tunnel road is determined, so as to avoid unreasonable repair of the tunnel and affect the tunnel passing efficiency. At the same time, timely repair can guarantee the life safety of the driver, and improve the experience of the driver driving the tunnel.

[0114] Figure 6 A road maintenance device in a tunnel is provided for the embodiment of the present application, and the device comprises:

[0115] At least one processor; and a memory connected in communication with the at least one processor. Wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0116] Obtain a plurality of real-time road surface monitoring images and corresponding acquisition information from an image acquisition device. Input each road surface monitoring image into a preset damaged road surface identification model to determine whether the monitoring road surface corresponding to the road surface monitoring image is a to-be-maintained road surface. If yes, determine the tunnel position corresponding to the to-be-maintained road surface and the road damage time according to the acquisition information of the road surface monitoring image. Based on historical driving data of the tunnel corresponding to the to-be-maintained road surface and historical driving data of an alternative path of the tunnel, determine a traffic flow weight set of the tunnel. The reachable destination of the alternative path is on a downlink and / or uplink of the tunnel position or on a reachable extension line of the downlink and / or uplink of the tunnel. Based on the road surface monitoring image, the road damage time, and the traffic flow weight set, determine road surface maintenance information of the tunnel, and send the road surface maintenance information to a maintenance terminal.

[0117] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.

[0118] The device and the method provided by the embodiments of the present application are one-to-one correspondence, and therefore, the device also has the similar beneficial technical effects as the method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the device will not be described here.

[0119] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles, or devices that include a series of elements not only include those elements, but also include other elements not explicitly listed, or other elements inherent to such processes, methods, articles, or devices. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, article, or device that includes the element.

[0120] The above only describes the embodiments of the present application and is not intended to limit the present application. The present application can have various changes and modifications for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. A method for road surface maintenance in tunnels, characterized in that, The method includes: Acquire several real-time road monitoring images and corresponding acquisition information from the image acquisition device; Each of the road surface monitoring images is input into a preset damaged road surface identification model to determine whether the monitored road surface corresponding to the road surface monitoring image is a road surface that needs maintenance. If so, based on the information collected from the road surface monitoring images, determine the location of the tunnel corresponding to the road surface to be maintained and the time of road surface damage; Based on the historical driving data of the tunnel corresponding to the road surface to be maintained and the historical driving data of the alternative routes of the tunnel, the traffic flow weight set of the tunnel is determined; wherein, the reachable destination of the alternative route is located on the downhill route and / or uphill route of the tunnel location or on the reachable extension of the downhill route and / or the reachable extension of the uphill route of the tunnel. Based on the road surface monitoring images, the road surface damage time, and the traffic flow weight set, the road surface maintenance information of the tunnel is determined, and the road surface maintenance information is sent to the maintenance terminal. Specifically, determining the traffic flow weight set for the tunnel based on historical driving data of the tunnel corresponding to the road surface to be maintained and historical driving data of the alternative routes to the tunnel includes: Based on the historical traffic data of the tunnel, the first average traffic flow of the tunnel is determined; Based on the historical driving data of each of the alternative routes, determine the corresponding second average traffic flow for each of the alternative routes; Based on the first average traffic flow and each of the second average traffic flows, the corresponding first traffic flow weight is determined; Based on the historical driving data of the tunnel, a third average traffic flow for several preset time periods of the tunnel is determined, and based on the third average traffic flow, several second traffic flow weights of the tunnel are determined. The traffic flow weight set is determined based on the first traffic flow weight and each of the second traffic flow weights of the tunnel.

2. The method according to claim 1, characterized in that, The step of inputting each of the road monitoring images into a preset damaged road surface identification model to determine whether the monitored road surface corresponding to the road monitoring image is a road surface requiring maintenance specifically includes: The damaged road surface identification model is used to determine the damaged image frame corresponding to the damaged road surface in each of the road surface monitoring images; Identify the damage label corresponding to each of the damaged image frames; wherein the damage label includes at least one or more of the following: subsidence, ruts, cracks, shrinkage cracks, and water seepage; Based on each of the damage tags, determine whether the degree of damage to the monitored road surface is greater than a preset threshold; If so, the monitored road surface shall be regarded as the road surface to be maintained.

3. The method according to claim 2, characterized in that, The step of determining whether the degree of damage to the monitored road surface exceeds a preset threshold based on each of the damage tags specifically includes: Determine the corresponding damage degree comparison image for each damaged label; wherein, a damaged label has at least two damage degree comparison images; the preset threshold is within the range corresponding to the damage degree values ​​of the two damage degree comparison images; The damaged image frame corresponding to the damaged label is matched with each of the damage degree comparison images to determine the damage degree value corresponding to the damaged label based on the matching results; wherein the damage degree value is determined based on the matched damage degree comparison images; Based on the damage weight corresponding to each damage label and the corresponding damage degree value, it is determined whether the damage degree of the monitored road surface is greater than the preset threshold.

4. The method according to claim 1, characterized in that, The step of determining the corresponding first traffic flow weight based on the first average traffic flow and each of the second average traffic flows specifically includes: Determine the average traffic flow of the first average traffic flow and the average traffic flow of each of the second average traffic flows; Based on the average traffic flow, the first average traffic flow, and each of the second average traffic flows, a first traffic flow weight corresponding to the first average traffic flow and each of the second average traffic flows is determined.

5. The method according to claim 4, characterized in that, The determination of tunnel road maintenance information based on the road monitoring images, the road damage time, and the traffic flow weight specifically includes: The type of road damage in the road monitoring image is determined by using a preset damage type identification model. The repair time for the tunnel is determined based on the type of road surface damage. Based on the road damage time, the traffic flow weight set, and the traffic flow weight set of each of the alternative paths, a number of third traffic flow weight sequences are determined within a first preset time period. The time period in which the third traffic flow weight of the tunnel satisfies the minimum weight value in each of the third traffic flow weight sequences is the undetermined repair period. Determine whether the first traffic flow weight of the alternative path is greater than the first traffic flow weight of the tunnel during the pending repair period. If the first traffic flow weight of the alternative path is greater than the first traffic flow weight of the tunnel, the undetermined repair period shall be used as the repair period. Otherwise, the minimum weight value is removed, and the time period corresponding to the minimum weight value in the third traffic flow weight sequence after removal is determined as the undetermined repair period, until the repair period is determined; If the repair time period is less than the repair duration, M repair time periods within the Mth preset time period are determined until the sum of the repair time periods is greater than the repair duration, and each repair time period is determined as the road maintenance information; wherein, M is a natural number.

6. The method according to claim 5, characterized in that, The method is applied to a pre-built blockchain platform, which includes several sub-nodes, and the sub-nodes are used to upload images of road damage and their corresponding road damage types. Before determining the type of road damage in the road monitoring image using a preset damage type identification model, the method further includes: Obtain several road damage images and their corresponding road damage types from the blockchain platform; Each of the aforementioned road damage images is used as a training sample, and the corresponding road damage type is used as the training label for the training sample. These are then input into a preset neural network recognition model. If the loss function of the neural network recognition model is less than a preset value, the trained neural network recognition model is determined to be the damage type recognition model.

7. The method according to claim 3, characterized in that, The step of matching the damaged image frame corresponding to the damaged label with each of the damage degree comparison images, and determining the damage degree value corresponding to the damaged label based on the matching result, specifically includes: Acquire several sample data images, and input each sample data image and the corresponding damage degree comparison image into a preset classifier to train the classifier; wherein, the classifier divides the sample data into at least multiple image combinations with the same number of damage degree comparison images; the damage degree value in the image combination matches the damage degree value of the corresponding damage degree comparison image; The classifier matches the image combinations corresponding to the damaged image frame to determine the degree of damage corresponding to the damaged image frame based on the matched image combinations.

8. The method according to claim 2, characterized in that, The method further includes: If the degree of damage to the monitored road surface is determined to be less than or equal to a preset threshold, the damaged image frame of the monitored road surface is sent to the maintenance terminal.

9. A road surface maintenance device for tunnels, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed, enables the at least one processor to perform a road surface maintenance method in a tunnel as described in any one of claims 1-8.

Images