Multi-scale based intelligent determination system and method for long-period maintenance target of highway

Abstract

The application discloses a multi-scale-based intelligent determination system and method for long-period maintenance targets of highways, and relates to the technical field of long-period maintenance of highways.The system comprises a highway monitoring module, a maintenance data management module and a maintenance data intelligent updating module.The highway monitoring module is used for collecting road condition indexes and monitoring traffic flow data in real time by using a monitoring device.The maintenance data management module is used for recording historical data in the process of maintenance of each highway.The maintenance data intelligent updating module is used for intelligently updating the maintenance target of the highway based on the size change of the road age and the change of the real-time traffic flow data during the long-period maintenance process of the highway.The application can strengthen the connection between each link, ensure the coherence and effectiveness of the maintenance work, improve the benefit of the maintenance investment, thereby realizing the high-quality development of the highway maintenance management, and comprehensively improving the scientific decision-making level of the highway maintenance, the efficiency of the use of funds, the market professional degree, the industry management level and the public travel experience.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of long-period maintenance of highways, in particular to a long-period maintenance target intelligent determination system and method for highways based on multi-scale. BACKGROUND

[0002] High-quality operation of highways depends on scientific and reasonable maintenance management. However, the development of highway maintenance management in China is relatively slow, and the maintenance mode is relatively single. The traditional highway maintenance mode has many problems. First, the division of maintenance responsibilities is not clear enough, resulting in unclear responsibilities. Second, the maintenance system is not perfect, lacking systematicness and standardization. Third, the layout of facilities is not reasonable, making it difficult to meet actual needs. Finally, the maintenance capacity is insufficient, making it difficult to cope with complex maintenance tasks. These problems seriously hinder the improvement of the level of highway maintenance management.

[0003] At present, the highway maintenance contracting mode in China mainly includes three types: daily maintenance general contracting, professional technical contracting and professional maintenance contracting. Although these modes promote the marketization process of highway maintenance to some extent, they still have many problems in actual operation. First, there is a lack of effective connection between detection, design, construction and other links, making it difficult for maintenance work to proceed smoothly. Second, there are many difficulties in management and maintenance, making it difficult to effectively supervise the maintenance process. Third, preventive maintenance measures for highways are difficult to implement, resulting in unsatisfactory maintenance effect. Finally, the benefits of maintenance investment are not maximized, and the resource utilization efficiency is low. SUMMARY

[0004] The purpose of the present application is to provide a long-period maintenance target intelligent determination system and method for highways based on multi-scale to solve the problems in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solution: a long-period maintenance target intelligent determination method for highways based on multi-scale, which comprises the following steps:

[0006] S1, a monitoring device is arranged on the side of the highway to collect real-time road condition indicators and monitor traffic flow data;

[0007] S2, a maintenance data management platform is constructed to record historical data in the process of highway maintenance, form a historical database, and be used for model calling;

[0008] S3, real-time changes in road age and real-time changes in traffic flow data are collected to intelligently update the highway maintenance target.

[0009] According to the above technical solution, the monitoring device comprises a sensor, a drone and an intelligent detection vehicle.

[0010] The road surface condition indicators include road surface flatness, road surface damage condition and rut depth; in the domestic highway maintenance field, the road surface flatness, the road surface damage condition and the rut depth are three particularly key evaluation angles. The road surface flatness is directly related to the comfort and safety of driving, and is one of important indicators for evaluating road surface quality; the damage condition can directly reflect the damage degree and durability of the road surface; and the rut depth is an important parameter for measuring the deformation degree of the road surface under heavy traffic.

[0011] The traffic flow data includes light traffic, medium traffic and heavy traffic.

[0012] According to the technical solution, the road age size includes: early road age, middle road age and late road age, wherein the early road age refers to 0-5 years of highway opening; the middle road age refers to more than 5 years and less than 10 years of highway opening; and the late road age refers to more than 10 years of highway opening.

[0013] According to the technical solution, the historical data in the highway maintenance process includes:

[0014] The highway road age size;

[0015] The acceptance road surface condition indicators in the highway maintenance process;

[0016] and the actual road surface condition indicators formed by the maintenance corresponding to the highway road age size;

[0017] and the traffic flow data corresponding to the maintenance process.

[0018] According to the technical solution, in step S3, further includes:

[0019] Setting the road surface condition indicator threshold as the highway maintenance target, taking the historical data in the highway maintenance process as the training data, and constructing a learning model;

[0020] Selecting the highway road age size in the historical data in the highway maintenance process and the traffic flow data corresponding to the maintenance process as the independent variables of the training data, taking the average value of the traffic flow in a time period as the value of the traffic flow data corresponding to the maintenance process, and if the maintenance process is less than one time period, taking the average value of the traffic flow data in the maintenance process as the value;

[0021] The one time period refers to being determined according to the current highway maintenance project which needs long-period maintenance;

[0022] Establishing the independent variable sample data value X, wherein X=[x1, x2], x1 and x2 respectively represent the highway road age size and the traffic flow data corresponding to the maintenance process; establishing the dependent variable sample data value y0, y0 represents any one of the road surface condition indicators; and forming a data set of the training data wherein i is valued from 1 to N, N refers to the total number of training data;

[0023] Based on the data set of training data, the network structure is constructed:

[0024] Two neurons are set as input layer, corresponding to x1, x2 respectively; L layers of hidden layer are set, each layer is provided with j neurons, respectively denoted as {H1, H2, …, HL}; one neuron is set as output layer, corresponding to dependent variable sample data y0; j

[0025] For the bth layer in the hidden layer, the forward propagation formula is:

[0026] z b = W b a b-1 + k b

[0027] wherein W b is the weight matrix of the bth layer; k b is the bias vector of the bth layer; a b-1 is the activation value of the b-1th layer; b is a constant, greater than 0 and less than or equal to L;

[0028] Corresponding output layer:

[0029]

[0030] a b = ReLU(z b )

[0031]

[0032] wherein, represents the prediction value of the b+1th layer; W b+1 represents the weight matrix of the b+1th layer participating in calculation; k b+1 represents the bias vector of the b+1th layer participating in calculation; a b represents the activation value of the bth layer participating in calculation; ReLU represents the activation function; G represents the loss function; ω represents the learning rate;

[0033] Initialize the parameters of the weight matrix and the bias vector, set the training rounds, and constantly update the parameters. When the set training rounds are reached, stop training, and the formed model is denoted as learning model;

[0034] For new input X, the prediction value is output based on the learning model Further form the road surface condition index:

[0035] ​

[0036] Wherein, M refers to the road condition index threshold value; Δm is the difference value average of the previous period in the maintenance process, and the specific calculation includes:

[0037]

[0038] Wherein, t refers to the number of previous periods; s represents the serial number; Y s refers to the actual road condition index formed by the s period maintenance; p s refers to the acceptance road condition index.

[0039] In the above technical solution, we analyze based on the premise of long period and limit the road condition index. Due to the continuous change of road age in long period, the acceptance index cannot be fixed.

[0040] The multi-scale intelligent determination system of highway long-period maintenance target, the system comprises:

[0041] The highway monitoring module is used for collecting road condition index and monitoring traffic flow data in real time by using monitoring equipment;

[0042] The maintenance data management module is used for recording historical data in the process of highway maintenance;

[0043] The intelligent maintenance data updating module is used for intelligently updating the highway maintenance target based on the size change of road age and the change of real-time traffic flow data in the process of highway long-period maintenance.

[0044] According to the above technical solution,

[0045] The monitoring equipment includes sensors, unmanned aerial vehicles and intelligent detection vehicles;

[0046] The road condition index includes road roughness, road damage and rut depth;

[0047] The traffic flow data includes light traffic, medium traffic and heavy traffic.

[0048] According to the above technical solution, the size of road age includes: early road age, middle road age and late road age, wherein the early road age refers to 0 to 5 years of highway opening; the middle road age refers to more than 5 years to less than 10 years of highway opening; the late road age refers to more than 10 years of highway opening.

[0049] According to the above technical solution, the historical data in the process of highway maintenance includes:

[0050] The size of highway road age;

[0051] The acceptance road condition index in the process of highway maintenance;

[0052] And the actual road condition indicators formed by maintenance corresponding to the age of the highway;

[0053] And the corresponding traffic flow data during the maintenance process.

[0054] According to the above technical solution, the maintenance data intelligent update module includes a learning unit and an update unit;

[0055] The learning unit is used to set the road condition index threshold as the highway maintenance target, and use the historical data in the highway maintenance process as training data to build a learning model; the updating unit is used to intelligently update the highway maintenance target during the long-term highway maintenance process.

[0056] Compared with existing technologies, the present invention has the following beneficial effects: It adopts a long-term, multi-scale approach to highway maintenance, further clarifying management and maintenance responsibilities, improving management and maintenance systems, optimizing facility layout, and enhancing management and maintenance capabilities. Furthermore, it strengthens the connection between various links, ensures the consistency and effectiveness of maintenance work, and improves the benefits of maintenance investment, thereby achieving high-quality development of highway maintenance management and comprehensively improving the level of scientific decision-making, capital utilization efficiency, market professionalism, industry management level, and public travel experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 The figure is a flow chart of the intelligent method for determining long-term highway maintenance targets based on multiple scales according to the present invention. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] Example: Figure 1 As shown, the present invention provides a multi-scale-based intelligent method for determining long-term highway maintenance targets, the method comprising the following steps:

[0060] S1. Install monitoring equipment on the road side to collect road condition indicators in real time and monitor traffic flow data;

[0061] S2. Build a maintenance data management platform to record historical data from the maintenance process of each highway and form a large historical database for model invocation;

[0062] S3. Collect changes in road age and traffic flow data in real time, and intelligently update highway maintenance targets.

[0063] The monitoring device comprises a sensor, a drone, and a smart detection vehicle.

[0064] The road surface condition indicators comprise road flatness, road damage, and rut depth.

[0065] The traffic flow data comprises light traffic, medium traffic, and heavy traffic.

[0066] The road age comprises early road age, medium road age, and late road age, wherein the early road age refers to 0-5 years of road opening, the medium road age refers to more than 5 years and less than 10 years of road opening, and the late road age refers to more than 10 years of road opening.

[0067] The historical data in the highway maintenance process comprises:

[0068] The road age;

[0069] The acceptance road surface condition indicators in the highway maintenance process;

[0070] and the actual road surface condition indicators formed by the maintenance corresponding to the road age;

[0071] and the traffic flow data corresponding to the maintenance process.

[0072] In the present application, the system can automatically set the traffic level, and then automatically set the range of light traffic, medium traffic, and heavy traffic. In the embodiments of the present application, the medium traffic level is taken as an example, and the change trend of each sub-item indicator is analyzed for the same road age. According to the existing experimental data, the road flatness shows a downward trend in the early road age. With the increase of unit mileage maintenance investment, the attenuation rate of road flatness in each road age stage shows a downward trend. When the unit mileage maintenance investment is low, the road damage condition index shows a faster attenuation rate in the early road age. When the unit mileage maintenance investment is low, the road rut depth index fluctuates greatly in different road age stages. When the unit mileage maintenance investment is medium or high, it remains stable in each road age stage. Therefore, different acceptance indicators should be set under different road ages.

[0073] The road surface condition indicator threshold is set as the highway maintenance target, the historical data in the highway maintenance process is taken as the training data, and a learning model is constructed;

[0074] The road age and the traffic flow data corresponding to the maintenance process are taken as the independent variables of the training data, the average value of the traffic flow data in a time period is taken as the value of the traffic flow data corresponding to the maintenance process, and if the maintenance process is less than one time period, the average value of the traffic flow data in the maintenance process is taken as the value.

[0075] The one time period is determined according to the current need for long-period maintenance of the highway maintenance project; the specific determination method can include determining by using performance evaluation, period acceptance and the like, for example, if the user gives the maintenance contractor a project payment period of one year, then one year can be taken as the time period;

[0076] The independent variable sample data value X is established, wherein X=[x1, x2], x1 and x2 respectively represent the road age and the corresponding traffic flow data in the maintenance process; the dependent variable sample data value y0 is established, y0 represents any one of the pavement condition indexes; and the data set of the training data is formed Wherein, the value of i is 1 to N, and N represents the total number of training data; in the training process, the same road age is taken as the original training data;

[0077] Based on the data set of the training data, the network structure is constructed:

[0078] Two neurons are set as the input layer, corresponding to x1 and x2 respectively; L layers of hidden layers are set, each layer is provided with j neurons, which are respectively denoted as {H1, H2, …, HL}; one neuron is set as the output layer, corresponding to the dependent variable sample data y0; j

[0079] For the bth layer in the hidden layer, the forward propagation formula is:

[0080] z b =W b a b-1 +k b

[0081] Wherein, W b is the weight matrix of the bth layer; k b is the bias vector of the bth layer; a b-1 is the activation value of the b-1th layer; b is a constant, greater than 0 and less than or equal to L;

[0082] Corresponding output layer:

[0083]

[0084] a b =ReLU(z b )

[0085]

[0086] Wherein, represents the predicted value of the b+1th layer; W b+1 represents the weight matrix of the b+1th layer participating in the calculation; k​b+1 Represents the bias vector of the b+1th layer involved in the calculation; a b Represents the activation value of the bth layer involved in the calculation; ReLU represents the activation function; G represents the loss function; ω represents the learning rate;

[0087] The loss function is generally the mean square error function. The gradient of the loss function for each parameter is calculated through the back propagation algorithm, and the parameters are updated using the gradient descent method.

[0088] Initialize the parameters of the weight matrix and bias vector, set the training rounds, continuously update the parameters, stop training when the set training rounds are reached, and the resulting model is recorded as the learning model;

[0089] For new input X, output the predicted value based on the learned model Further forming road condition indicators:

[0090]

[0091] Where M is the pavement condition index threshold; Δm is the average difference between the previous cycles during the maintenance process. The specific calculation includes:

[0092]

[0093] Where t refers to the number of cycles in the previous sequence; s represents the sequence number; Y s Refers to the actual road condition index formed by the maintenance of the sth cycle; p s Refers to the road condition indicators for acceptance.

[0094] In this embodiment, a multi-scale-based intelligent system for determining long-term highway maintenance targets is also provided. The system includes:

[0095] Highway monitoring module, used to collect road condition indicators in real time using monitoring equipment, and monitor traffic flow data;

[0096] Maintenance data management module, used to record historical data of each highway maintenance process;

[0097] The maintenance data intelligent update module is used to intelligently update highway maintenance targets during the long-term maintenance process based on changes in road age and real-time traffic flow data.

[0098] The monitoring equipment includes sensors, drones, and smart inspection vehicles;

[0099] The road condition indicators include road surface smoothness, road surface damage and rutting depth;

[0100] The traffic flow data includes light traffic, medium traffic and heavy traffic.

[0101] The road age size comprises: road age early stage, road age middle stage and road age late stage, wherein the road age early stage refers to 0-5 years of highway opening, the road age middle stage refers to more than 5 years and less than 10 years of highway opening, and the road age late stage refers to more than 10 years of highway opening.

[0102] The historical data in the highway maintenance process comprises:

[0103] The highway road age size;

[0104] The acceptance pavement condition index in the highway maintenance process;

[0105] and the actual pavement condition index formed by maintenance corresponding to the highway road age size;

[0106] and the corresponding traffic flow data in the maintenance process.

[0107] The maintenance data intelligent updating module comprises a learning unit and an updating unit;

[0108] The learning unit is used for setting a pavement condition index threshold as a highway maintenance target, using historical data in the highway maintenance process as training data, and constructing a learning model; and the updating unit is used for intelligently updating the highway maintenance target in the long-period maintenance process of the highway.

[0109] It is apparent for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, but can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered as exemplary and non-restrictive in any respect, and the scope of the present application is defined by the appended claims rather than the above description, and all changes falling within the meaning and range of the equivalent elements of the claims are intended to be embraced in the present application. Any reference signs in the claims should not be considered as limiting the claims involved.

Claims

1. A multi-scale based intelligent determination method for long-period maintenance targets of highways, characterized in that: The method comprises the following steps: S1, setting a monitoring device on the side of the highway to collect road condition indicators in real time and monitor traffic flow data; S2, constructing a maintenance data management platform to record historical data in the highway maintenance process to form a historical database for model calling; S3, collecting the size change of road age and the change of real-time traffic flow data to intelligently update the highway maintenance target; In step S3, further comprising: setting the road condition indicator threshold as the highway maintenance target, and using the historical data in the highway maintenance process as training data to construct a learning model; selecting the size of the highway age and the corresponding traffic flow data in the maintenance process as the independent variables of the training data, and taking the average value of the traffic flow in a time period as the value, if the maintenance process is less than one time period, taking the average value of the traffic flow data in the maintenance process as the value; The one time period refers to the determination according to the current highway maintenance project that needs long-period maintenance; A sample data value X of an independent variable is established, wherein X = [x1, x2], x1 and x2 respectively represent the road age and the corresponding traffic flow data in the maintenance process; a sample data value y0 of a dependent variable is established, y0 represents any one of the pavement condition indexes; and a data set of the training data is formed wherein i is valued from 1 to N, and N represents the total number of the training data Based on the data set of the training data, the network structure is constructed: Two neurons are set as the input layer, corresponding to x1 and x2 respectively; L layers of hidden layers are set, each layer having j neurons, denoted as {H1, H2, …, H j}; one neuron is set as the output layer, corresponding to the dependent variable sample data y0; The forward propagation formula for the bth layer in the hidden layer is: z b = W b a b-1 + k b where W b is the weight matrix of the bth layer; k b is the bias vector of the bth layer; a b-1 is the activation value of the b-1th layer; b is a constant, greater than 0 and less than or equal to L; The corresponding output layer: a b = ReLU(z b ) wherein, represents the prediction value of the b+1th layer; W b+1 represents the weight matrix of the b+1th layer involved in the calculation; k b+1 represents the bias vector of the b+1th layer involved in the calculation; a b represents the activation value of the bth layer involved in the calculation; ReLU represents the activation function; G represents the loss function; ω represents the learning rate; Initialize the weight matrix and bias vector parameters, set the training rounds, and constantly update the parameters. Stop training when the set training rounds are reached, and the formed model is recorded as a learning model; For a new input X, output a predicted value based on the learning model Further form a road condition indicator: Wherein, M refers to the road condition indicator threshold; Δm is the average value of the difference in the previous period in the maintenance process, and the specific calculation includes: wherein t refers to the number of previous cycles; s represents the sequence number; Y s Ys refers to the actual pavement condition index formed by the maintenance in the s cycle; p s Yp refers to the acceptance pavement condition index.

2. The intelligent determination method for long-period maintenance target of highway based on multi-scale according to claim 1, wherein: The monitoring device comprises a sensor, a drone, and an intelligent detection vehicle; The road condition indicators include road flatness, road damage, and rut depth; The traffic flow data includes light traffic, medium traffic, and heavy traffic.

3. The intelligent determination method for long-period maintenance target of highway based on multi-scale according to claim 1, wherein: The size of the road age includes early road age, middle road age, and late road age, wherein the early road age refers to 0-5 years of highway opening, the middle road age refers to more than 5 years but less than 10 years of highway opening, and the late road age refers to more than 10 years of highway opening.

4. The multi-scale based intelligent determination method of long-period maintenance targets for highways according to claim 1, characterized in that: The historical data in the highway maintenance process includes: The size of the highway age; The acceptance road condition indicators in the highway maintenance process; The actual road condition indicators formed under the maintenance of the corresponding highway age; and The corresponding traffic flow data in the maintenance process.

5. A multi-scale based intelligent determination system for long-period maintenance targets of highways, using the multi-scale based intelligent determination method for long-period maintenance targets of highways according to any one of claims 1-4, characterized in that: The system comprises: A highway monitoring module for collecting road condition indicators in real time and monitoring traffic flow data using a monitoring device; A maintenance data management module for recording historical data in the highway maintenance process; A maintenance data intelligent updating module for intelligently updating the highway maintenance target based on the size change of road age and the change of real-time traffic flow data in the long-period maintenance process of the highway.

6. The intelligent determination system for long-period maintenance target of highway based on multi-scale according to claim 5, wherein: The monitoring device comprises a sensor, a drone, and an intelligent detection vehicle; The road surface condition indexes include road flatness, road damage and rut depth. The traffic flow data includes light traffic, medium traffic and heavy traffic. 7.The multi-scale based intelligent long-period highway maintenance target determination system according to claim 5, characterized in that: The road age includes early road age, middle road age and late road age, wherein the early road age refers to 0-5 years of highway opening, the middle road age refers to more than 5 years but less than 10 years of highway opening, and the late road age refers to more than 10 years of highway opening.

8. The multi-scale based intelligent highway long-cycle maintenance target determination system of claim 5, wherein: The historical data in the highway maintenance process includes: Highway road age; Acceptance road surface condition indexes in the highway maintenance process; Actual road surface condition indexes formed under the corresponding highway road age in the maintenance; And corresponding traffic flow data in the maintenance process.

9. The multi-scale based intelligent highway long-cycle maintenance target determination system of claim 5, wherein: The intelligent maintenance data updating module includes a learning unit and an updating unit; The learning unit is configured to set road surface condition index thresholds as highway maintenance targets, use historical data in the highway maintenance process as training data, and construct a learning model; and the updating unit is configured to intelligently update the highway maintenance targets in the long-period highway maintenance process.

Images