Intelligent highway pavement sensing and predicting method

Abstract

The invention relates to an intelligent highway pavement sensing and predicting method, which comprises the following steps of: constructing an acquisition layer, and acquiring basic data transmitted by a sensor network and pavement data acquired by a camera according to a data network constructed by the acquisition layer; constructing a verification network, and verifying the transmission conflict between the basic data acquired by the data network and the road surface data based on a plurality of acquisition perception parameters set by the verification network; and taking the flow characteristics and the pest and disease damage development trend characteristics as the prediction basis of the pavement crack and pit damage, obtaining the prediction data of the pavement crack and pit damage according to a set time period, and comparing the prediction data with the pavement data obtained in the corresponding time period to correct the processing model. Through the coding table and the synchronous clock technology, time-space alignment of multi-source data is achieved, synchronous association of pressure-image data is achieved, and the effectiveness of data during later data processing is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent road surface perception for expressways, and in particular to an intelligent road surface perception and prediction method for expressways. Background Art

[0002] Cracks, potholes and rutting on highway pavements are the basic causes of pavement damage, and the main causes of these factors are traffic volume, environment and overloading. Traffic volume and overloading lead to accelerated fatigue damage of the structural layer, and cause deformation diseases such as rutting and subsidence. Especially in high temperature environments, the compaction of asphalt pavements by heavy-loaded vehicles will aggravate oil spills and displacement. Therefore, studying the degree of damage to the pavement caused by these factors is the best way to improve the use of the pavement. At present, the basic situation of the pavement can be obtained by analyzing the pressure data of the roadbed surface and the actual image of the pavement. However, traditional data acquisition has a lot of randomness, and the collected pressure data and pavement images are sometimes not in the same acquisition cycle, resulting in inaccurate predictions. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a highway intelligent road surface perception and prediction method to solve the problems raised in the background technology.

[0004] The main purpose of this application is to provide a highway intelligent road surface perception and prediction method, including the following steps: Construct a collection layer, connect it to the sensor network installed in the highway and the camera for mobile detection of the highway road surface through the collection layer, and collect the basic data transmitted by the sensor network and the road surface data obtained by the camera according to the data network constructed by the collection layer.

[0005] A verification network is constructed to verify the transmission conflicts between basic data and road surface data collected by the data network based on a plurality of collection perception parameters set by the verification network.

[0006] Construct a processing model, mark the time series characteristics of the basic data and pavement data collected by the data network through the verification network, and after marking, obtain the flow characteristics of traffic flow and the development trend characteristics of pests and diseases by establishing a time series prediction model in the processing model; use the flow characteristics and the development trend characteristics of pests and diseases as the basis for the prediction of pavement cracks and potholes, and obtain the predicted data of pavement cracks and potholes according to the set time period, and compare the predicted data with the pavement data obtained in the corresponding time period to correct the processing model.

[0007] Furthermore, the following method is used to construct the collection layer: Divide the expressway to be monitored into several uniform areas, encode each area, and record the initial encoding table.

[0008] Set up a variety of sensors in each area to form a sensor combination unit. Call the initial encoding table and write the interface parameters of the various sensors and the access parameters of the master station module receiving the various sensors under the encoding of each area, so as to form a combined network unit inside the multiple sensor combination units, and form a sensor network with multiple sensor combination units having independent combined network units. Record the formed encoding network table.

[0009] Set up a mobile detection camera in each area. Call the encoding network table to encode and mark the camera in the encoding network table, so that the camera corresponds to the sensor network in the corresponding area. At the same time, form an application encoding table applied to the entire expressway monitoring based on the encoding network table.

[0010] Use the application encoding table as the framework for constructing the acquisition layer, set up an acquisition matrix, and set several acquisition modules under the acquisition matrix. Each acquisition module is configured with several acquisition units to correspond to the sensor network and camera in each area.

[0011] Further, the verification network includes: Verification control logic; A verification module, which is configured to: establish a data perception unit by obtaining the matching of the interface parameters of the sensor and the access parameters of the master station module receiving the sensor, and write the transmission channel established between the constructed data perception unit, the sensor, and the master station module into the verification control logic.

[0012] Form acquisition perception parameters according to the data form of the sensor, and use the acquisition perception parameters as the first conflict verification (Ⅰ) of the data perception unit under the control of the verification control logic; by configuring a synchronous clock timing for the data perception unit, the data perception unit performs timing verification (Ⅱ) when monitoring the transmission of basic data and road surface data under the control of the verification control logic.

[0013] Further, establish a transmission channel between the sensor and the master station module by obtaining the interface parameters of the sensor and the access parameters of the master station module receiving the sensor.

[0014] Further, the verification control logic includes: a first logic for configuring the transmission channel and performing transmission control matching.

[0015] A second logic for identifying the acquisition perception parameters and applying and matching the data perception unit based on the acquisition perception parameters.

[0016] Complete the third logic for the first conflict verification of the data perception unit when monitoring data channel data transmission based on the second logic; and Complete the fourth logic for timing verification when transmitting the monitored basic data and road surface data based on the third logic.

[0017] Furthermore, the data forms of the sensors include: A marking layer, call the application coding table to obtain the coding of the sensor, and mark the transmission channel established between the sensor and the master station module with the coding of the sensor to form marking information.

[0018] A presentation layer, including the address information, data transmission format, and data structure of the sensor.

[0019] Furthermore, the timing verification includes: Parse the timestamp in the sensor data packet and verify the continuity of the timestamp.

[0020] Furthermore, the timing prediction model includes: Obtain the daily traffic flow data by sequentially obtaining the pressure data of the pressure sensor according to the clock timing, and use the pressure value distribution corresponding to the daily traffic flow data as the flow feature.

[0021] Use the temperature sensor and humidity sensor to obtain the periodic temperature average data and humidity average data of the roadbed, and use the periodic temperature average data and humidity average data as a reference and match them with the road surface cracks and pothole diseases obtained historically to obtain the development trend characteristics of pests and diseases.

[0022] Furthermore, obtain the high-frequency, medium-frequency, and low-frequency components of the pressure value in the flow feature within the set period according to the daily flow feature, use the random forest method to model the high-frequency, medium-frequency, and low-frequency components respectively to predict the disturbances of the road surface respectively, obtain the corresponding first disturbance features, obtain the proportion of each first disturbance feature within multiple periods, and when comparing with the actual road surface cracks and pothole diseases, obtain the contribution score of each first disturbance feature to the actual road surface cracks and pothole diseases according to the proportion of each first disturbance feature.

[0023] Furthermore, use the periodic temperature average data and humidity average data, use the random forest method to model the periodic temperature average data and humidity average data respectively to predict the disturbances of the road surface respectively, obtain the corresponding second disturbance features, and when comparing with the actual road surface cracks and pothole diseases, obtain the contribution score of each second disturbance feature to the actual road surface cracks and pothole diseases according to each second disturbance feature.

[0024] Through the coding table and synchronous clock technology, this application achieves the spatio-temporal alignment of multi-source data, enables the synchronous correlation of pressure-image data, and ensures the effectiveness of data during subsequent data processing.

[0025] This application adopts the multi-dimensional perturbation separation technology to greatly reduce the crack prediction error rate, and has obvious improvement compared with the error rate of a single-factor model. Brief Description of the Drawings

[0026] Figure 1 It is a flowchart of the method of the present invention.

[0027] Figure 2 It is a flowchart of the method for constructing the acquisition layer in the present invention. Detailed Embodiments

[0028] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0029] To solve the problems raised in the background technology, this application adopts the following technical solutions. It should be noted that the advantages of this application are: Adopting a "trinity" acquisition system of sensor network + mobile camera networking + coding mapping to build a multi-source data fusion architecture, enabling data from different source ends to be accurately transmitted to the corresponding site modules according to the allocated transmission channels. This provides a hardware foundation for the synchronous verification of source-end data.

[0030] In this application, the multi-source acquisition layer includes: traffic flow data: basic data such as vehicle flow, vehicle speed, and vehicle type distribution are collected through roadside sensors / cameras; road surface detection data: intelligent inspection vehicles are deployed to collect structured data such as crack width and pothole depth.

[0031] This application adopts a dual-logic dynamic verification mechanism to synchronously solve the problems of device interface matching and data transmission timing conflict. Through the coding table and synchronous clock technology, this application achieves the spatio-temporal alignment of multi-source data, enables the synchronous correlation of pressure-image data, and ensures the effectiveness of data during subsequent data processing.

[0032] This application adopts a multi-dimensional disturbance separation algorithm: through frequency domain decomposition and random forest weighting, the independent contribution quantification of traffic load and environmental factors is realized for the first time. Specifically, the high-frequency, medium-frequency and low-frequency components of the pressure value in the flow characteristics within the set period are obtained according to the daily flow characteristics. The high-frequency, medium-frequency and low-frequency components are modeled respectively using the random forest method to predict the disturbance of the road surface respectively, obtain the corresponding first disturbance characteristics, obtain the proportion of each first disturbance characteristic in multiple periods, and obtain the contribution score of each first disturbance characteristic to the actual road surface cracks and potholes according to the proportion of each first disturbance characteristic when comparing with the actual road surface cracks and potholes. The periodic temperature average data and humidity average data are modeled respectively using the random forest method to predict the disturbance of the road surface respectively, and obtain the corresponding second disturbance characteristics. When comparing with the actual road surface cracks and potholes, the contribution score of each second disturbance characteristic to the actual road surface cracks and potholes is obtained according to each second disturbance characteristic.

[0033] This application realizes a closed-loop self-correction system: combining prediction-measurement comparison with time series simulation, breaking through the limitations of traditional static models. Specifically, the time series characteristics of the basic data and road surface data collected by the data network are marked through the verification network. After marking, the flow characteristics of traffic flow and the development trend characteristics of pests and diseases are obtained by establishing a time series prediction model in the processing model; the flow characteristics and the development trend characteristics of pests and diseases are used as the basis for the prediction of road cracks and potholes, and the prediction data of road cracks and potholes are obtained according to the set time period, and the predicted data is compared with the road surface data obtained in the corresponding time period to correct the processing model.

[0034] Reference Figure 1 and Figure 2 , the present application provides a highway intelligent road surface perception and prediction method, comprising the following steps: Construct a collection layer, connect it to the sensor network installed in the highway and the camera for mobile detection of the highway road surface through the collection layer, and collect the basic data transmitted by the sensor network and the road surface data obtained by the camera according to the data network constructed by the collection layer.

[0035] A verification network is constructed to verify the transmission conflicts between basic data and road surface data collected by the data network based on a plurality of collection perception parameters set by the verification network.

[0036] Build a processing model. Mark the temporal features of the basic data and road surface data collected by the data network through a verification network. After marking, establish a temporal prediction model in the processing model to obtain the flow characteristics of traffic flow and the development trend characteristics of pests and diseases. Use the flow characteristics and the development trend characteristics of pests and diseases as the prediction basis for road surface cracks and pothole diseases, and obtain the prediction data of road surface cracks and pothole diseases according to the set time period. Compare the prediction data with the road surface data obtained in the corresponding time period to correct the processing model.

[0037] Further, the acquisition layer is constructed by the following method: Divide the highway to be monitored into several uniform areas, and encode each area; record the initial encoding table.

[0038] Set multiple sensors in each area to form a sensor combination unit. Call the initial encoding table and write the interface parameters of the multiple sensors and the access parameters of the master station module for receiving the multiple sensors under the encoding of each area, so as to form a combined network unit inside the multiple sensor combination units, and form a sensor network with multiple sensor combination units having independent combined network units; record the formed encoding network table.

[0039] Set a mobile detection camera in each area. Call the encoding network table to encode and mark the camera in the encoding network table, so that the camera corresponds to the sensor network in the corresponding area. At the same time, form an application encoding table for the monitoring of the entire highway based on the encoding network table.

[0040] Use the application encoding table as the framework for constructing the acquisition layer, set an acquisition matrix, and set several acquisition modules under the acquisition matrix. Each acquisition module is configured with several acquisition units to correspond to the sensor network and camera in each area.

[0041] In the above, this application adopts regionalized monitoring: divide the highway into uniformly encoded areas (such as one unit every 500 meters), and record the deployment positions of sensors and cameras through the encoding table to ensure that the data sources correspond one-to-one with the spatial positions. At the same time, multi-source device networking is adopted: the sensor combination unit (pressure, temperature and humidity sensors, etc.) and the mobile camera form a mapping relationship through the application encoding table, and automatically match multi-modal data in the same area and at the same time point during data acquisition.

[0042] Further, the verification network includes: Verification control logic; A verification module, which is configured to: establish a data perception unit by obtaining the matching of the interface parameters of the sensor and the access parameters of the master station module of the sensor, and write the established transmission channel between the constructed data perception unit, the sensor and the master station module into the verification control logic.

[0043] Form acquisition perception parameters according to the data form of the sensor, and use the acquisition perception parameters as the first conflict verification (Ⅰ) of the data perception unit under the control of the verification control logic; by configuring a synchronous clock timing for the data perception unit, the data perception unit monitors the transmission of basic data and road surface data under the control of the verification control logic for timing verification (Ⅱ).

[0044] In the above, the present application adopts dual-logic verification. Through interface parameter matching verification (the first conflict verification), by checking the consistency between the sensor interface protocol and the master station module reception protocol, data packet loss or format errors are avoided. Synchronous clock timing verification (the second conflict verification): Provide a unified clock reference for all devices through a clock chip, parse the data packet timestamp and verify its continuity. At the same time, a dynamic error correction mechanism is implemented. When a transmission conflict is detected, the verification network automatically triggers data retransmission or switches to an alternative transmission channel.

[0045] Further, establish a transmission channel between the sensor and the master station module by obtaining the interface parameters of the sensor and the access parameters of the master station module that receives the sensor.

[0046] Further, the verification control logic includes: a first logic for configuring the transmission channel and performing transmission control matching.

[0047] A second logic for identifying the acquisition perception parameters and applying and matching the data perception unit based on the acquisition perception parameters.

[0048] A third logic for completing the first conflict verification of the data perception unit when monitoring the data transmission of the data channel based on the second logic; and A fourth logic for performing timing verification when monitoring the transmission of basic data and road surface data based on the third logic.

[0049] Further, the data form of the sensor includes: A marking layer, which calls the application coding table to obtain the encoding of the sensor, and marks the established transmission channel between the sensor and the master station module with the encoding of the sensor to form marking information.

[0050] A presentation layer, which includes the address information, data transmission format, and data structure of the sensor.

[0051] Further, the timing verification includes: Parse the timestamp in the sensor data packet and verify the continuity of the timestamp.

[0052] Furthermore, the timing prediction model includes: Obtain the daily traffic flow data by sequentially acquiring the pressure data of the pressure sensor according to the clock timing, and use the pressure value distribution corresponding to the daily traffic flow data as the traffic flow characteristics.

[0053] Use the temperature sensor and humidity sensor to obtain the periodic temperature average data and humidity average data of the roadbed, and use the periodic temperature average data and humidity average data as a reference and match them with the road surface cracks and pothole diseases obtained historically to obtain the development trend characteristics of plant diseases and insect pests.

[0054] Furthermore, according to the daily traffic flow characteristics, obtain the high-frequency, medium-frequency, and low-frequency components of the pressure value in the traffic flow characteristics within a set period, use the random forest method to model the high-frequency, medium-frequency, and low-frequency components respectively to predict the disturbances of the road surface respectively, obtain the corresponding first disturbance characteristics, and obtain the proportion of each first disturbance characteristic within multiple periods. When comparing with the actual road surface cracks and pothole diseases, obtain the contribution score of each first disturbance characteristic to the actual road surface cracks and pothole diseases according to the proportion of each first disturbance characteristic.

[0055] Furthermore, use the periodic temperature average data and humidity average data, use the random forest method to model the periodic temperature average data and humidity average data respectively to predict the disturbances of the road surface respectively, obtain the corresponding second disturbance characteristics, and obtain the contribution score of each second disturbance characteristic to the actual road surface cracks and pothole diseases according to each second disturbance characteristic when comparing with the actual road surface cracks and pothole diseases.

[0056] In the above, traffic flow characteristic extraction: Pressure data spectrum decomposition: Decompose the pressure sensor data into high-frequency (instantaneous overload), medium-frequency (conventional traffic flow), and low-frequency (long-term load accumulation) components through wavelet transform, and model and predict the pavement fatigue degree respectively. Random forest dynamic weight allocation: Train the model according to historical data, and dynamically adjust the contribution weights of different frequency band pressure components to cracks and ruts (for example, the weight proportion of the high-frequency component can reach 60%). Specifically, wavelet transform frequency band decomposition can be used for signal stratification and feature extraction. For example, high-frequency component: Capture instantaneous overload signals above 100Hz (such as emergency braking, heavy load impact), medium-frequency component: Extract the vibration characteristics of conventional traffic flow at 10 - 100Hz (such as tire pressure fluctuations during uniform driving); low-frequency component: Reflect the long-term load accumulation effect below 10Hz (such as periodic changes in traffic flow, temperature deformation).

[0057] This application can also extract traffic flow characteristics in the following way.

[0058] 'High frequency': [Kurtosis, Pulse factor, Short-time energy variance, Zero-crossing rate], 'Medium frequency': [RMS, Spectrum centroid, Waveform factor, Autocorrelation decay rate], 'Low frequency': [Trend slope, Long-term energy integral, Periodicity intensity].

[0059] Quantitative analysis of environmental disturbances: Temperature-humidity coupling model: Establish a temperature-humidity-disease correlation matrix. For example, when high temperature (>35°C) is superimposed on high humidity (>80%), the softening rate of the asphalt layer increases by 2 times, significantly aggravating the rut depth. Second disturbance feature calibration: Identify the environmental threshold through the decision tree algorithm (such as a temperature mutation ≥10°C / hour triggering an alarm), and output the contribution score of environmental disturbances to the development of potholes.

[0060] Since this application adopts a closed-loop feedback system, the predicted data can be compared with the actual inspection images (such as crack width, pothole area) at the pixel level, and when the error exceeds the threshold, the model parameter iteration is triggered. Disease evolution map: Based on the LSTM network, a time series prediction is constructed to simulate the crack propagation path and pothole merging probability within the next 30 days, and a visual risk heat map is output.

[0061] The above are only some embodiments of this application, and do not limit the patent scope of this application. Any equivalent structural transformation made under the technical concept of this application by using the content of the specification and drawings of this application, or directly / indirectly applied in other related technical fields, is included in the patent protection scope of this application.

Claims

1. An intelligent road surface perception and prediction method for expressways, characterized in that The steps include: Construct a collection layer, connect the collection layer with the sensor network installed in the expressway and the camera for mobile detection of the expressway road surface, and collect the basic data transmitted by the sensor network and the road surface data obtained by the camera according to the data network constructed by the collection layer; Building a verification network to verify the transmission conflict between basic data and road surface data collected by the data network based on multiple collection perception parameters set by the verification network; Build a processing model, mark the time series characteristics of the basic data and road surface data collected by the data network through the verification network, and then establish a time series prediction model in the processing model to obtain the flow characteristics of traffic flow and the development trend characteristics of pests and diseases; The flow characteristics and the development trend characteristics of pests and diseases are used as the basis for predicting pavement cracks and potholes. The predicted data of pavement cracks and potholes are obtained according to the set time period. The predicted data are compared with the pavement data obtained in the corresponding time period to correct the processing model.

2. The intelligent pavement perception and prediction method for expressways according to claim 1, wherein, The following method is used to construct the collection layer: Divide the highway to be monitored into a number of uniform areas, and encode each area; record the initial coding table; Arrange multiple sensors in each area to form a sensor combination unit, call the initial coding table to write the interface parameters of the multiple sensors and the access parameters of the main station module receiving the multiple sensors into the coding of each area, so as to form a combination network unit inside the multiple sensor combination units, and form a sensor network with multiple sensor combination units having independent combination network units; Record the formed coding network table; A motion detection camera is set in each area, and the coding network table is called to code the camera in the coding network table so that the camera corresponds to the sensor network of the corresponding area, and at the same time, an application coding table for monitoring the entire highway is formed based on the coding network table; The application coding table is used as a framework for constructing the acquisition layer, and an acquisition matrix is set. Several acquisition modules are set under the acquisition matrix, and each acquisition module is configured with several acquisition units for the sensor network and camera corresponding to each area.

3. The intelligent pavement perception and prediction method for expressways according to claim 1, wherein The verification network comprises: Verify control logic; A verification module, the verification module is configured to: establish a data sensing unit by acquiring the matching of the interface parameters of the sensor and the access parameters of the main station module receiving the sensor, and write the transmission channel established between the constructed data sensing unit and the sensor and the main station module into the verification control logic; The sensor data form is used to form acquisition perception parameters, and the acquisition perception parameters are used as the first conflict verification of the data perception unit under the control of the verification control logic (I). By configuring a synchronous clock timing for the data perception unit, the data perception unit performs timing verification when monitoring the transmission of basic data and road surface data under the control of the verification control logic (II).

4. The intelligent pavement perception and prediction method for expressways according to claim 3, wherein, A transmission channel is established between the sensor and the master station module by acquiring the interface parameters of the sensor and receiving the access parameters of the master station module of the sensor.

5. The intelligent road surface perception and prediction method for expressways according to claim 3, characterized in that The verification control logic includes: a first logic for configuring a transmission channel and performing transmission control matching; a second logic for identifying acquired perception parameters, applying and performing data perception unit matching based on the acquired perception parameters; a third logic for completing the first conflict verification of the data perception unit during data transmission on the monitoring data channel based on the second logic; and a fourth logic for performing timing verification during the transmission of monitoring basic data and road surface data based on the third logic.

6. The intelligent pavement perception and prediction method for expressways according to claim 3, wherein, The data form of the sensor includes: a marking layer that calls the application coding table to obtain the encoding of the sensor, and marks the transmission channel established between the sensor and the master station module with the encoding of the sensor to form marking information; a presentation layer that includes the address information, data transmission format, and data structure of the sensor.

7. The intelligent road surface perception and prediction method for expressways according to claim 3, characterized in that The timing verification includes: parsing the timestamp in the sensor data packet to verify the continuity of the timestamp.

8. The intelligent pavement perception and prediction method for expressways according to claim 1, wherein The timing prediction model includes: obtaining traffic daily flow data by sequentially acquiring the pressure data of the pressure sensor according to the clock timing, and using the pressure value distribution corresponding to the daily flow data as the flow feature; obtaining the periodic temperature average data and humidity average data of the roadbed with the temperature sensor and humidity sensor, using the periodic temperature average data and humidity average data as a benchmark and performing corresponding matching with the road surface cracks and pothole diseases acquired historically to obtain the development trend characteristics of plant diseases and insect pests.

9. The intelligent road surface perception and prediction method for expressways according to claim 8, characterized in that Obtaining the high-frequency, medium-frequency, and low-frequency components of the pressure value in the flow feature within a set period according to the daily flow feature, respectively modeling the high-frequency, medium-frequency, and low-frequency components using the random forest method to predict the disturbances of the road surface respectively, obtaining the corresponding first disturbance features, obtaining the proportion of each first disturbance feature within multiple periods, and when comparing with the actual road surface cracks and pothole diseases, obtaining the contribution score of each first disturbance feature to the actual road surface cracks and pothole diseases according to the proportion of each first disturbance feature.

10. The intelligent road surface perception and prediction method for expressways according to claim 8, wherein Using the periodic temperature average data and humidity average data, respectively modeling the periodic temperature average data and humidity average data using the random forest method to predict the disturbances of the road surface respectively, obtaining the corresponding second disturbance features, and when comparing with the actual road surface cracks and pothole diseases, obtaining the contribution score of each second disturbance feature to the actual road surface cracks and pothole diseases according to each second disturbance feature.

Images