Intelligent management and control method and system for cement concrete pavement cracks

Abstract

The application discloses an intelligent management and control method and system for cement concrete pavement cracks, and belongs to the field of road engineering, wherein the method comprises the following steps: setting sensitive switching parameters; controlling the collection of the motion of a traveling vehicle, pre-collecting road surface data, and constructing an initial reference data set; initializing a decision unit, inputting the initial reference data set into the decision unit, and outputting decision data; selecting infrared sensors and / or imaging sensors to collect image data and construct an image data set; identifying defects from the image data set, recording abnormal coordinates and a disease feature set according to the identification result; calling a road surface archive, matching maintenance strategies, and intelligently managing and controlling the road surface according to the matching result. The application solves the technical problem that the existing technology cannot intelligently and accurately detect and maintain road surface cracks, and achieves the technical effect of intelligently managing and controlling road surface cracks through sensor collection, intelligent analysis and accurate matching.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of road engineering, in particular to an intelligent management and control method and system for cement concrete pavement cracks. BACKGROUND

[0002] With the development of social economy, people's requirements for road quality are getting higher and higher. How to detect and evaluate the road condition in real time and accurately, and realize the intelligent maintenance and management of the road, is an important problem in urban construction. For cement concrete roads, cracks are one of the main diseases. At present, the detection of pavement cracks mainly relies on manual inspection, which is subjective and inefficient. In addition, the detection of pavement cracks also has the problems of low detection accuracy, inability to distinguish and identify different types of cracks, and inability to match intelligent maintenance strategies accordingly, which limits the efficiency and intelligent level of cement concrete pavement crack management. SUMMARY

[0003] The present application provides an intelligent management and control method and system for cement concrete pavement cracks, aiming to solve the technical problem that the existing technology cannot detect and maintain pavement cracks intelligently and accurately.

[0004] In view of the above problems, the present application provides an intelligent management and control method and system for cement concrete pavement cracks.

[0005] The first aspect of the present application provides an intelligent management and control method for cement concrete pavement cracks, which comprises: setting a sensitive switching parameter, the sensitive switching parameter being a switching response parameter collected from the pavement, and the sensitive switching parameter being set by collecting the basic data of an infrared sensor and an imaging sensor; collecting the motion of a collection vehicle at a predetermined speed, and pre-collecting pavement data through a pre-collection unit to construct an initial reference data set, and a decision unit, a pre-collection unit, an infrared sensor and an imaging sensor are arranged on the collection vehicle and are communicatively connected with each other; initializing the decision unit through the sensitive switching parameter, inputting the initial reference data set into the decision unit, and outputting decision data; selecting the infrared sensor and / or the imaging sensor to collect image data according to the decision data, and constructing an image data set; identifying defects in the image data set through the decision unit, recording abnormal coordinates and a disease feature set according to the identification result; calling a pavement archive, matching a maintenance strategy through the pavement archive, the abnormal coordinates and the disease feature set, and intelligently managing and controlling the pavement according to the matching result.

[0006] In another aspect of the present application, an intelligent management and control system for cracks in cement concrete pavement is provided, which comprises: a sensitive switching parameter module for setting a sensitive switching parameter, the sensitive switching parameter being a switching response parameter collected from the pavement, and the sensitive switching parameter being set by collecting basic data of an infrared sensor and an imaging sensor; an initial reference data module for controlling the movement of a collection vehicle at a predetermined speed, and pre-collecting pavement data by a pre-collection unit to construct an initial reference data set, and the decision unit, the pre-collection unit, the infrared sensor and the imaging sensor being arranged on the collection vehicle and being communicatively connected with each other; a decision data output module for initializing the decision unit by the sensitive switching parameter, and inputting the initial reference data set into the decision unit to output decision data; an image data collection module for collecting image data by the decision data to select the infrared sensor and / or the imaging sensor to construct an image data set; a data defect identification module for identifying defects in the image data set by the decision unit, and recording abnormal coordinates and a disease feature set according to the identification result; and a pavement intelligent management and control module for calling a pavement file, matching a maintenance strategy by the pavement file and the abnormal coordinates and the disease feature set, and intelligently managing and controlling the pavement according to the matching result.

[0007] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] By setting the sensitive switching parameter by collecting the basic data of the infrared sensor and the imaging sensor, the accurate switching control of the sensors is realized. By controlling the movement of the collection vehicle at a predetermined speed, and pre-collecting the pavement data by the pre-collection unit arranged on the collection vehicle, the initial reference data set is constructed for subsequent decision analysis. By initializing the decision unit by the set sensitive switching parameter, and inputting the constructed initial reference data set into the decision unit, the decision data is output by the decision unit to guide the subsequent image collection. According to the indication of the decision data, the infrared sensor and / or the imaging sensor is selected for image collection to construct the image data set, which provides data support for crack identification. The image data set is analyzed by the decision unit to accurately identify the pavement crack position coordinates and disease features. The pavement file database is called to perform matching analysis on the file and the identification result to formulate a corresponding maintenance strategy, and the technical solution for intelligent management and control of cracks in cement concrete pavement is completed, which solves the technical problem that the pavement cracks cannot be intelligently and accurately detected and maintained in the prior art, and achieves the technical effect of intelligent management and control of pavement cracks by sensor collection, intelligent analysis and accurate matching.

[0009] The above description is only a summary of the technical solutions of the present application. In order to enable the technical means of the present application to be more clearly understood, and to be implemented in accordance with the content of the description, and in order to enable the above and other purposes, characteristics and advantages of the present application to be more apparent and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 A flowchart of an intelligent management and control method for cement concrete pavement cracks is provided for the embodiments of the present application;

[0011] Figure 2 A flowchart of determining abnormal coordinates and disease feature sets in an intelligent management and control method for cement concrete pavement cracks is provided for the embodiments of the present application;

[0012] Figure 3 A structural schematic diagram of an intelligent management and control system for cement concrete pavement cracks is provided for the embodiments of the present application.

[0013] Explanation of reference signs: sensitive switching parameter module 11, initial reference data module 12, decision data output module 13, image data acquisition module 14, data defect identification module 15, pavement intelligent management and control module 16. DETAILED DESCRIPTION

[0014] The general idea of the technical solutions provided by the present application is as follows:

[0015] The embodiments of the present application provide an intelligent management and control method and system for cement concrete pavement cracks, realizing intelligent detection and treatment of the whole process of cement concrete pavement cracks. First, precise control of the acquisition sensor is realized by setting sensitive switching parameters, improving the detection efficiency of pavement cracks. Then, various sensor acquisition devices are set on the acquisition vehicle to construct an initial reference data set and an image data set, providing data support for subsequent intelligent analysis. Next, a decision unit is used to realize automatic identification of the position and characteristics of pavement cracks. Finally, through matching analysis with the pavement archives, a corresponding maintenance strategy is formulated to complete the treatment of pavement cracks, effectively solving the technical problem that the pavement cracks cannot be intelligently managed and controlled in the prior art, achieving intelligent and refined pavement maintenance and management.

[0016] After introducing the basic principles of the present application, the various non-limiting embodiments of the present application will be specifically introduced in combination with the drawings of the specification.

[0017] Embodiment one

[0018] As shown in Figure 1 The embodiments of the present application provide an intelligent management and control method for cement concrete pavement cracks, which comprises:

[0019] The sensitive switching parameter is a switching response parameter for collecting the road surface, and is set by collecting basic data of the infrared sensor and the imaging sensor;

[0020] In the embodiments of the present application, the sensitive switching parameter refers to a switching response parameter for image collection of the road surface, which is used to determine the switching use of the infrared sensor or the imaging sensor, and is determined by sensor basic data, so that the infrared sensor and the imaging sensor actively switch according to the actual road surface conditions to obtain better road surface image collection effect. The infrared sensor is an infrared night vision device, which can form an infrared image of the road surface in an environment without visible light; and the imaging sensor is an image collection device, which can obtain image information in the visible light range of the road surface.

[0021] First, the imaging sensor and the infrared sensor are respectively used to collect images of the road surface to obtain image basic data of the two sensors under different light conditions. Second, the collected image data is analyzed to determine the intervals in which the imaging sensor and the infrared sensor can respectively obtain the best image quality in different light intensity ranges, for example, the imaging sensor works well under strong light and the infrared sensor works well under weak light. Then, the best applicable intervals of the two sensors are comprehensively set to set a corresponding relationship between the light intensity range and the sensor type as the sensitive switching parameter. In subsequent collection, the best sensor or sensor combination currently used is determined according to the light intensity by using the sensitive switching parameter.

[0022] By setting the sensitive switching parameter, the best collaborative working mode of the infrared sensor and the image sensor can be dynamically determined according to the data collection capability of the two sensors, so that accurate and efficient collection of road surface image information is realized.

[0023] The motion of the collection vehicle is controlled by a predetermined speed, and the road surface data is pre-collected by a pre-collection unit to construct an initial reference data set, and the decision unit, the pre-collection unit, the infrared sensor and the imaging sensor are arranged on the collection vehicle and are communicatively connected with each other;

[0024] In the embodiments of the present application, the collection vehicle is a vehicle provided with a road surface data collection device, which can travel along the road surface at a predetermined speed. The pre-collection unit is a road surface data pre-collection device arranged on the vehicle, such as a light sensor, which is used to collect road light intensity data in real time during the travel of the vehicle. The decision unit, the pre-collection unit, the infrared sensor and the imaging sensor are all installed on the collection vehicle and are connected for data transmission to work collaboratively.

[0025] In the road surface data collection process, first, the predetermined speed set according to the actual situation of the road surface is controlled to control the stable driving of the collection vehicle. At the same time, the pre-collection unit starts to work, and the road surface light data is collected in advance to obtain the initial road surface data and construct the initial reference data set. The pre-collection data set contains light intensity information at different positions of the road surface. Since each unit is set on the same vehicle and connected, real-time data transmission and sharing can be realized. During the collection process, the units are connected by communication, such as 5G network, Bluetooth, wired transmission, etc., to realize information exchange, instruction transmission and other collaborative work, so as to facilitate the subsequent accurate collection of road surface images.

[0026] The decision unit is initialized by the sensitive switching parameter, and the initial reference data set is input into the decision unit to output decision data.

[0027] Further, the step specifically includes:

[0028] When the initial reference data set is input into the decision unit, the light evaluation sub-network is used to evaluate the initial reference data set.

[0029] Generate light intensity evaluation data, and use the light intensity evaluation data as the first decision data.

[0030] Generate light uniformity evaluation data, and use the light uniformity evaluation data as the second decision data.

[0031] The first decision data and the second decision data are sent to the decision unit to output the decision data.

[0032] In a feasible implementation, the decision unit is first initialized by using the predefined sensitive switching parameter, the parameter is loaded into the unit, and the decision unit is configured. Secondly, the initial reference data set obtained by pre-collection is input into the decision unit to provide road light environment information. When the initial reference data set obtained by pre-collection is input into the decision unit, the decision unit will automatically trigger the light evaluation sub-network to start. The sub-network is a network module set in the decision unit for light analysis, which is a convolutional neural network. The light data in the input initial reference data set can be feature extracted and analyzed, and the light can be evaluated.

[0033] Then, the sub-network extracts the light intensity value of each road surface collection point in the initial reference data set, determines the light intensity distribution of the overall road surface, and calculates the statistical parameters of the light intensity at different positions of the road surface. Then, according to these statistical parameters, the overall light intensity level of the entire road surface corresponding to the initial reference data set, i.e., whether the light intensity is generally sufficient, is comprehensively evaluated to form evaluation result data representing the light intensity level. The light intensity evaluation data will be used as the first decision data of the decision unit to determine which image sensor or sensor combination to use for subsequent fine image collection.

[0034] At the same time, the lighting evaluation sub-network detects and evaluates the light uniformity of the initial reference data set. Specifically, the sub-network analyzes the distribution range and variation trend of the light intensity values of each collection point on the road surface. If the light intensity varies little on the entire road surface and is uniformly distributed, the light uniformity is evaluated as good. Conversely, if the light varies greatly, the uniformity is evaluated as poor. Thus, the evaluation result representing the light uniformity level of the road surface of the initial data set, i.e., the light uniformity evaluation data, is generated. The evaluation data is used as the second decision data of the decision unit, which is used together with the light intensity evaluation data to determine which image sensor or combination to use for fine collection.

[0035] Subsequently, the first decision data and the second decision data are sent to the decision unit together. The decision unit uses an ensemble learning algorithm to comprehensively consider the two factors of light intensity and uniformity to make a decision calculation. The first decision data is the light intensity evaluation result generated by the lighting evaluation sub-network, reflecting the overall light intensity level of the initial reference data set. The second decision data is the light uniformity evaluation result generated by the sub-network, reflecting the uniformity of the light on the road surface. Finally, the decision unit outputs a set of decision data, indicating which image sensor or combination should be selected for use under the current road lighting conditions.

[0036] The decision data is used to select the infrared sensor and / or the imaging sensor for image data collection to construct an image data set.

[0037] Further, the step specifically includes:

[0038] If the decision data is that the imaging sensor works independently, the light intensity median value of the first decision data is calculated.

[0039] The light intensity median value calculation result is used as an auxiliary control parameter and sent to the imaging sensor.

[0040] The auxiliary control parameter is used as a light measurement reference to control the imaging sensor to complete image data collection.

[0041] Further, the step further includes:

[0042] If the decision data is for mixed working of the imaging sensor, data segmentation instruction is generated;

[0043] The first decision data is subjected to two-stage data segmentation according to the data segmentation instruction;

[0044] The average value of the high-brightness level in the two-stage segmentation result is taken as the imaging photometry reference, and the average value of the low-brightness level in the two-stage segmentation result is taken as the infrared control reference, to control the infrared sensor and the imaging sensor to complete image data acquisition.

[0045] In a preferred embodiment, based on the decision data, it is determined to select the infrared sensor or the imaging sensor to perform image acquisition. In the process of image acquisition, in order to obtain more complete and detailed image data, the sensor will scan and acquire the road surface at a certain frequency. That is, during the travel of the vehicle, the sensor will perform image acquisition every certain distance interval, and there will be a certain overlapping area between adjacent two acquisitions to ensure that a complete image can be spliced. Through continuous and frequent acquisition, the sensor obtains multiple sets of image data with a certain repetition. Finally, these image data containing the repeated area are spliced using an image processing algorithm to construct a high-quality image data set with continuous content and rich details.

[0046] Firstly, it is judged whether the decision data is for image acquisition by only enabling the imaging sensor, i.e., whether the decision data only contains an instruction for selecting the imaging sensor to work independently. If yes, the first decision data is extracted, and the median value of the light intensity statistical values of multiple acquisition points contained therein is calculated, i.e., the light intensity values of all acquisition points are counted, and the median value of these light intensity values, i.e., the representative light intensity value between the maximum value and the minimum value, is found out to provide a basis for subsequent parameter optimization of the imaging sensor. Then, the calculated light intensity median value is taken as an auxiliary control parameter and sent to the imaging sensor. After receiving the auxiliary control parameter, the imaging sensor takes it as a photometry reference and correspondingly configures parameters such as exposure time and aperture size of the imaging sensor. Under the configuration of the parameters, the imaging sensor is controlled to acquire image information of the road surface to obtain optimal image data under a given lighting condition and complete image data acquisition.

[0047] If the decision data indicates that the infrared sensor and the imaging sensor should be enabled to work simultaneously, corresponding data segmentation instructions are generated for two-level segmentation of the light intensity statistical values in the generated first decision data (light intensity evaluation data), i.e., into two data sets of high-brightness level and low-brightness level, such as using a K-means clustering analysis algorithm to cluster the light intensity evaluation data into two data clusters. Then, the average value of the light intensity of the high-brightness cluster is calculated as the light metering reference parameter of the imaging sensor, and the average value of the light intensity of the low-brightness cluster is calculated as the control reference parameter of the infrared sensor. Subsequently, the corresponding parameters are sent to the two sensors to configure the parameters, and the imaging sensor performs light metering control with the high-brightness average value, and the infrared sensor uses the low-brightness average value as the reference for image processing. Finally, the two sensors are controlled simultaneously to perform image acquisition according to the respective optimized parameters, and high-quality image data combined from both are obtained.

[0048] Defect recognition is performed on the image data set by the decision unit, and abnormal coordinates and disease feature sets are recorded according to the recognition results.

[0049] Further, as shown in Figure 2 , the step specifically includes:

[0050] If the image data set is a double-layer data, a segmentation contour line is generated according to the two-level segmentation results to determine a standard image area and an infrared image area.

[0051] The standard image area and the infrared image area are respectively executed for binary comparison with the road surface calibration gray value as the center gray value, and are determined as abnormal areas.

[0052] The convolution kernel of the image is traversed for the abnormal area, and the convolution kernel is an abnormal convolution kernel constructed by big data.

[0053] Abnormal coordinates and disease feature sets are determined according to the traversal results.

[0054] In a preferred embodiment, the decision unit is used to recognize defects in the collected image data set. The decision unit scans and analyzes each position of the image data set to identify various road diseases such as cracks and potholes. For each identified abnormal disease, the corresponding coordinate position in the image, i.e., the abnormal coordinate, is accurately recorded. At the same time, the feature information of the disease, such as shape, size, and density, is extracted to form a feature set of the disease. The process is repeated to record the coordinates and feature sets corresponding to each abnormality, complete the comprehensive defect recognition and feature extraction of the image data set, and form the abnormal coordinate and feature set data to provide a basis for subsequent processing and analysis.

[0055] If the decision unit selects to use both the infrared sensor and the imaging sensor, the final obtained image dataset will contain two layers of data, i.e. infrared data and standard image data. In this case, a segmentation contour line is generated using the obtained two-level segmentation results. This contour line represents the boundary between the infrared image data area and the standard imaging data area. Then, based on this segmentation contour line, the area covered by the infrared data in the image dataset, i.e. the infrared image area, is determined and labeled; at the same time, the standard image area other than the infrared area is labeled. Thus, the clear labeling of the two types of areas in the image data is obtained, which provides the basis for differentiated processing of the two types of areas.

[0056] The calibration grayscale value of the standard image of the road surface is set in advance, representing the typical grayscale characteristics of the normal road surface. In addition, since the infrared image and the standard image are different, the infrared image also sets an independent contrast grayscale value. Subsequently, using the calibration grayscale value of the standard image as the central contrast grayscale for the abnormal detection of the standard image area, the difference between the grayscale value of each point in the standard image area and the central contrast grayscale is compared, and if the difference exceeds the preset threshold, the point is marked as a candidate abnormal point. For the infrared image area, a similar comparison is performed using the corresponding infrared central contrast grayscale to obtain the abnormal candidate points of the infrared image. The abnormal candidate point sets of the standard image and the infrared image form the abnormal area. The characteristics of the two types of image data are compared respectively to improve the accuracy of abnormal identification.

[0057] Then, the determined abnormal area is traversed using the constructed abnormal convolution kernel to accurately identify the abnormality, wherein the abnormal convolution kernel is an image feature extraction tool constructed using the convolutional neural network algorithm and a large number of abnormal image samples, which can capture the feature information of various types of abnormalities. For the obtained abnormal area, the abnormal convolution kernel is used to traverse it comprehensively, and the convolution response value is calculated. If the convolution result responds strongly, it is confirmed as a real abnormality; if the response is weak, it is excluded as a false alarm. Through the accurate identification of the abnormal convolution kernel, a large number of false alarms can be filtered out, and only the real abnormal area is retained, thereby improving the detection accuracy.

[0058] After traversing the convolution kernel on the candidate abnormal area, the convolution response analysis result of the abnormal area is obtained. For those areas with strong traversal response, they are confirmed as real abnormal areas, and the coordinate positions of these abnormal areas in the original image dataset are recorded as abnormal coordinates. At the same time, for each confirmed abnormal area, a feature extraction algorithm is used to obtain the specific feature information of the abnormal area, such as disease shape, size, distribution, etc., to form a feature set describing the abnormality. Integrating all the abnormal coordinate information and the corresponding disease feature set, accurate judgment, positioning and feature extraction of the abnormality are realized, which lays a foundation for accurate matching of subsequent maintenance strategies.

[0059] Call the road surface archive, maintain the maintenance strategy matching through the road surface archive and the abnormal coordinates, the disease feature set, and intelligently control the road surface according to the matching result.

[0060] Further, the embodiments of the application also include:

[0061] Parse the road surface archive to obtain construction record information and environmental record information.

[0062] Take the abnormal coordinates as a mapping comparison node, and perform correlation analysis on the construction record information, the environmental record information, and the disease feature set to construct a correlation identifier.

[0063] Take the disease feature set as the basic data and the correlation identifier as the constraint data to perform maintenance strategy matching.

[0064] In a preferred embodiment, the road surface archive exists in the form of a structured digital database and is stored in a cloud server. The system accesses the road surface archive database through a high-speed network connection, sends an access request with a unique identification code of the current road surface to be analyzed as a query parameter. After receiving the request, the server searches for the corresponding road surface archive in the database according to the road surface identification code, and returns the found road surface archive data to the management and control system through the network connection.

[0065] After receiving the returned road surface archive data, natural language processing technology is used to parse the archive text data, and sentences related to the words such as "construction" and "repair" are identified in the text data. Similarly, sentences related to the words such as "climate", "temperature", and "rainfall" are identified, thereby forming a structured construction record and environmental record data set to provide a basis for subsequent decision analysis. The construction record information is the construction time of the road surface, the road repair content performed, and other information; the environmental record information is the past climate, temperature change, rainfall, and other information.

[0066] Subsequently, the disease distribution map formed by the abnormal coordinates is marked on the map, and the construction activity distribution in the construction record is also marked on the same map. Then, through an image registration algorithm, the alignment mapping of the two distribution maps is realized, the spatial coincidence of the two maps is compared, the correlation between the disease distribution and the historical construction activity is determined, and the correlation identifier of the construction activity to the road surface disease is performed. At the same time, the time distribution of the environmental record is compared with the disease distribution, the correlation analysis method is used to calculate the correlation coefficients of different factors and the current disease distribution, and according to the correlation coefficients, the correlation between the environmental factors and the disease distribution is determined, and the correlation identifier of the environmental factors to the road surface disease is performed.

[0067] Then, a strategy knowledge base is constructed, in which a large number of encoding vectors of maintenance strategies are pre-stored. Data in the disease feature set are encoded into a structured digital vector as a basic vector input for strategy matching; meanwhile, each factor in the association identifier is encoded into a constraint condition vector. Next, according to the basic vector input, a candidate strategy set closest to the basic vector in the strategy vector space is searched, and the candidate set is filtered according to the constraint conditions one by one to filter out the strategies that do not meet the conditions. Then, the matching degree of each candidate strategy with the basic vector is compared, and the maintenance strategy that is most similar in semantics to the basic vector and meets the constraint conditions is selected as the best matching result.

[0068] Then, the best matching result is converted into corresponding maintenance operation instructions, such as which procedures need to be performed, which equipment and materials are used, etc. The maintenance operation instructions are transmitted to the management system. The management system arranges actual maintenance operations according to the instructions, such as allocating equipment, personnel, and purchasing materials. The operation personnel arrive at the designated section with the required equipment and materials to perform accurate maintenance. After the maintenance operation is completed, the operation data is uploaded to the management system to realize closed-loop management and control.

[0069] Further, the embodiments of the application also include:

[0070] According to the disease feature set, an outlier analysis is performed;

[0071] If there is an outlier disease feature that meets the preset requirements, a repeated collection instruction is generated according to the abnormal coordinates corresponding to the disease feature;

[0072] The rich collection is re-executed through the repeated collection instruction.

[0073] In a preferred embodiment, the extracted disease feature set is subjected to outlier detection to determine whether there is a feature value that meets the preset abnormal condition, such as an outlier with an out-of-range size or an atypical shape. If such an abnormal feature exists, it indicates that the corresponding position needs to be collected again to obtain more rich information. At this time, a repeated collection instruction for the position is generated according to the coordinates corresponding to the abnormal feature. The instruction will be transmitted to the corresponding sensor to repeatedly scan the coordinate position in the instruction. Ultimately, more rich and multi-angle image data are obtained for further decision analysis.

[0074] In summary, the intelligent management and control method for cracks in cement concrete pavement provided by the embodiments of the application has the following technical effects:

[0075] The sensitive switching parameter is a switching response parameter collected for the road surface, and the sensitive switching parameter is set by collecting basic data of the infrared sensor and the imaging sensor, so that the sensor is accurately controlled, and the detection efficiency is improved. The initial reference data set is constructed by controlling the motion of the collection vehicle at a predetermined speed and pre-collecting road surface data by the pre-collection unit, and the decision unit, the pre-collection unit, the infrared sensor and the imaging sensor are arranged on the collection vehicle and are in communication connection with each other to provide basic data for subsequent analysis. The decision unit is initialized by the sensitive switching parameter, and the initial reference data set is input into the decision unit to output decision data, thereby providing support for controlling the sensor to collect images. The infrared sensor and / or the imaging sensor are selected by the decision data to collect image data and construct an image data set, thereby providing information for intelligent identification of defects of the road surface. The decision unit identifies defects of the image data set, records abnormal coordinates and a disease feature set according to the identification result, and provides a basis for controlling the road surface. The road surface archive is called, and the maintenance strategy is matched by the road surface archive, the abnormal coordinates and the disease feature set, the intelligent control of the road surface is performed according to the matching result, the whole-process intelligent control from collection detection to governance decision is realized, the detection and maintenance of the road surface crack are more intelligent and accurate, and the quality of the cement concrete road surface and the driving experience are improved.

[0076] Embodiment two

[0077] Based on the same inventive concept as the intelligent control method of the cement concrete road surface crack in the foregoing embodiment, as shown in Figure 3 The embodiment of the present application provides an intelligent control system of a cement concrete road surface crack, and the system comprises:

[0078] A sensitive switching parameter module 11 is configured to set a sensitive switching parameter, wherein the sensitive switching parameter is a switching response parameter collected for a road surface, and the sensitive switching parameter is set by collecting basic data of an infrared sensor and an imaging sensor.

[0079] An initial reference data module 12 is configured to control the motion of a collection vehicle at a predetermined speed, pre-collect road surface data by a pre-collection unit, and construct an initial reference data set, wherein a decision unit, the pre-collection unit, the infrared sensor and the imaging sensor are arranged on the collection vehicle and are in communication connection with each other.

[0080] A decision data output module 13 is configured to initialize the decision unit by the sensitive switching parameter, input the initial reference data set into the decision unit, and output decision data.

[0081] An image data collection module 14 is configured to select the infrared sensor and / or the imaging sensor by the decision data to collect image data and construct an image data set.

[0082] a data defect identification module 15 configured to identify defects in the image dataset by the decision unit, and record abnormal coordinates and a disease feature set according to the identification result;

[0083] a road surface intelligent management and control module 16 configured to call a road surface archive, match a maintenance strategy by the road surface archive and the abnormal coordinates and the disease feature set, and perform intelligent management and control of the road surface according to the matching result.

[0084] Further, the decision data output module 13 includes the following execution steps:

[0085] When the initial reference dataset is input into the decision unit, light evaluation is performed on the initial reference dataset by a light evaluation sub-network;

[0086] Light intensity evaluation data is generated, and the light intensity evaluation data is taken as first decision data;

[0087] Light uniformity evaluation data is generated, and the light uniformity evaluation data is taken as second decision data;

[0088] The first decision data and the second decision data are sent to the decision unit, and the decision data is output.

[0089] Further, the decision data output module 13 further includes the following execution steps:

[0090] It is judged whether the decision data is independent work of the imaging sensor;

[0091] If the decision data is independent work of the imaging sensor, then light intensity median calculation is performed on the first decision data;

[0092] The light intensity median calculation result is taken as an auxiliary control parameter, and is sent to the imaging sensor;

[0093] The auxiliary control parameter is taken as a light measurement reference to control the imaging sensor to complete image data acquisition.

[0094] Further, the decision data output module 13 further includes the following execution steps:

[0095] If the decision data is mixed work with the imaging sensor, then a data segmentation instruction is generated;

[0096] The first decision data is two-level segmented according to the data segmentation instruction;

[0097] The average value of the high-brightness level in the two-level segmentation result is taken as an imaging photometry reference, and the average value of the low-brightness level in the two-level segmentation result is taken as an infrared control reference, to control the infrared sensor and the imaging sensor to complete image data acquisition.

[0098] Further, the data defect identification module 15 includes the following execution steps:

[0099] If the image data set is a double-layer data, a segmentation contour line is generated according to the two-level segmentation result, and a standard image area and an infrared image area are determined;

[0100] The calibration gray value of the road surface is taken as a center comparison gray value, and binary comparison of the standard image area and the infrared image area is respectively performed, and an abnormal area is determined;

[0101] The convolution kernel of the image is traversed for the abnormal area, and the convolution kernel is an abnormal convolution kernel constructed by big data;

[0102] According to the traversal result, an abnormal coordinate and a disease feature set are determined.

[0103] Further, the road surface intelligent management and control module 16 includes the following execution steps:

[0104] The road surface archive is parsed to obtain construction record information and environmental record information;

[0105] The abnormal coordinate is taken as a mapping comparison node, and the construction record information, the environmental record information, and the disease feature set are associated analyzed to construct an association identifier;

[0106] The disease feature set is taken as basic data, and the association identifier is taken as constraint data to perform maintenance strategy matching.

[0107] Further, the present application embodiment further includes a repeated collection instruction module, which includes the following execution steps:

[0108] According to the disease feature set, an abnormal value is analyzed;

[0109] If there is an abnormal value that meets the preset requirement of the disease feature, a repeated collection instruction is generated according to the abnormal coordinate corresponding to the disease feature;

[0110] The repeated collection instruction is re-executed to perform rich collection.

[0111] Any step of the above-mentioned method can be stored in a computer memory without limitation as computer instructions or programs, and can be called and recognized by a computer processor without limitation to realize any method in the present application embodiment, and no redundant limitation is made here.

[0112] Further, the first or second possible not only represents the order relationship, but also can represent a specific concept, and / or refers to the single or all selection between multiple elements. Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the scope of the application. Thus, if these modifications and variations of the present application fall within the scope of the present application and equivalent technology, then the present application is intended to include these modifications and variations.

Claims

1. An intelligent management and control method for cement concrete pavement cracks, characterized in that, The method comprises: Setting a sensitive switching parameter, which is a switching response parameter collected from the road surface, and the sensitive switching parameter is set by collecting basic data of an infrared sensor and an imaging sensor; Collecting the motion of a collection vehicle at a predetermined speed, pre-collecting road surface data by a pre-collection unit, constructing an initial reference data set, and a decision unit, the pre-collection unit, the infrared sensor, and the imaging sensor are arranged on the collection vehicle and are in communication with each other; Initializing the decision unit by the sensitive switching parameter, inputting the initial reference data set into the decision unit, and outputting decision data; Selecting the infrared sensor and / or the imaging sensor to collect image data according to the decision data, and constructing an image data set; Identifying defects in the image data set by the decision unit, recording abnormal coordinates and disease feature sets according to the identification results; Calling a road file, matching maintenance strategies by the road file, the abnormal coordinates, and the disease feature sets, and intelligently managing and controlling the road surface according to the matching results; The method further comprises: After the initial reference data set is input into the decision unit, performing light evaluation on the initial reference data set by a light evaluation sub-network; Generating light intensity evaluation data, and taking the light intensity evaluation data as first decision data; Generating light uniformity evaluation data, and taking the light uniformity evaluation data as second decision data; Sending the first decision data and the second decision data to the decision unit, and outputting the decision data; Determining whether the decision data is independent work of the imaging sensor; If the decision data is independent work of the imaging sensor, performing median value calculation on the first decision data; Taking the light intensity median value calculation result as an auxiliary control parameter, and sending it to the imaging sensor; Taking the auxiliary control parameter as a light measurement reference to control the imaging sensor to complete image data collection; If the decision data is mixed work with the imaging sensor, generating a data segmentation instruction; According to the data segmentation instruction, performing two-level segmentation on the first decision data; Taking the average value of the high brightness level in the two-level segmentation result as an imaging light measurement reference, and taking the average value of the low brightness level in the two-level segmentation result as an infrared control reference to control the infrared sensor and the imaging sensor to complete image data collection; If the image data set is double-layer data, generating a segmentation contour line according to the two-level segmentation result to determine a standard image area and an infrared image area; Taking the calibrated gray value of the road surface as a center comparison gray value to perform binary comparison of the standard image area and the infrared image area respectively, and determining an abnormal area; Performing convolution kernel traversal on the abnormal area, and the convolution kernel is an abnormal convolution kernel constructed by big data; Determining abnormal coordinates and disease feature sets according to the traversal result.

2. The method of claim 1, wherein, The method further comprises: Analyzing the road file to obtain construction record information and environmental record information; The abnormal coordinates are taken as mapping comparison nodes to perform correlation analysis on the construction record information, the environment record information and the disease feature set, and to construct correlation identification; The disease feature set is taken as basic data, and the correlation identification is taken as constraint data to perform maintenance strategy matching.

3. The method of claim 1, wherein, The method further comprises: Performing abnormal value analysis according to the disease feature set; If there is an abnormal value satisfying a preset requirement of the disease feature, generating a repeated collection instruction according to the abnormal coordinates corresponding to the disease feature; Re-executing the rich collection through the repeated collection instruction.

4. An intelligent management and control system for cracks in cement concrete pavement, characterized in that, The method comprises: A sensitive switching parameter module, which is configured to set a sensitive switching parameter, the sensitive switching parameter being a switching response parameter for collecting the road surface, and the sensitive switching parameter being set by basic data of an infrared sensor and an imaging sensor; An initial reference data module, which is configured to control a collection vehicle to move at a predetermined speed, and to pre-collect road surface data through a pre-collection unit to construct an initial reference data set, and the pre-collection unit, the infrared sensor and the imaging sensor being arranged on the collection vehicle and being communicatively connected to each other; A decision data output module, which is configured to initialize the decision unit by using the sensitive switching parameter, and to input the initial reference data set into the decision unit to output decision data; An image data collection module, which is configured to select the infrared sensor and / or the imaging sensor to collect image data by using the decision data, and to construct an image data set; A data defect identification module, which is configured to identify defects in the image data set by using the decision unit, and to record abnormal coordinates and a disease feature set according to an identification result; A road surface intelligent management and control module, which is configured to call a road surface file, to perform maintenance strategy matching by using the road surface file, the abnormal coordinates and the disease feature set, and to perform intelligent management and control of the road surface according to a matching result.

Images