A multi-source data fusion road surface monitoring method and system

Abstract

The application discloses a kind of multi-source data fusion's road surface monitoring method and system, it is related to wisdom traffic infrastructure safety monitoring field, the method includes: to road surface monitoring data is preprocessed, based on the road surface monitoring data after preprocessing is carried out structure feature extraction and normalization processing, obtain road surface structure health index data;The historical road surface monitoring data obtained is converted into road surface structure probability distribution function, determines initial road surface structure weight, and the initial road surface weight is dynamically optimized;According to road surface structure health index data and the road surface structure weight after optimization calculates road surface health index, and carries out road surface risk early warning, based on road surface risk early warning result carries out damage factor identification, obtains leading damage factor.The application can visualize road surface health index, accurately feedback the risk and wear degree of road, reduce the risk brought by road damage, and can independently judge the main damage factor of road surface health, identify potential safety hazard.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent transportation infrastructure safety monitoring, in particular to a multi-source data fusion road surface monitoring method and system. BACKGROUND

[0002] With the rapid development of transportation industry, especially the surge in bulk cargo transportation and traffic flow, the highway pavement gradually damages under the influence of traffic load and environmental factors, and the safety, stability and efficiency of highway transportation are facing great challenges. The traditional traffic management mainly relies on manual patrol and regular inspection, which cannot accurately find potential problems and sudden conditions on the road in real time. In order to improve the intelligent level of traffic management and reduce the safety hazards caused by traffic events, bad weather and road damage, perception monitoring technology emerges as the times require. Especially on the national highway and expressway with high flow and high risk, it is particularly important to build an integrated and automated intelligent perception system.

[0003] Although some perception monitoring facilities have been applied on some highways, most of these devices are limited to monitoring in a certain aspect, such as traffic flow or weather conditions, and lack comprehensive data integration and intelligent analysis capabilities and multi-source data collaborative analysis capabilities, which cannot comprehensively evaluate the road health status. Therefore, there is an urgent need for a new intelligent perception monitoring system that can comprehensively and real-time monitor the traffic conditions, weather conditions, road health and traffic events of the highway and provide decision support through intelligent analysis.

[0004] At present, no effective solution has been proposed for the problems in the related art. SUMMARY

[0005] In view of the problems in the related art, the present application proposes a multi-source data fusion road surface monitoring method and system to overcome the above technical problems existing in the prior art.

[0006] To this end, the specific technical solutions adopted by the present application are as follows:

[0007] According to an aspect of the present application, a multi-source data fusion road surface monitoring method is provided, which comprises the following steps:

[0008] S1, acquiring road surface monitoring data and pre-processing the road surface monitoring data to obtain pre-processed road surface monitoring data, and based on the pre-processed road surface monitoring data, performing structure feature extraction and normalization processing to obtain road surface structure health index data;

[0009] S2, convert the obtained historical road surface monitoring data into a road surface structure probability distribution function, determine an initial road surface structure weight according to the road surface structure probability distribution function, and dynamically optimize the initial road surface weight to obtain an optimized road surface structure weight;

[0010] S3, calculate a road surface health index according to the road surface structure health index data and the optimized road surface structure weight, and perform road surface risk early warning in combination with a preset health index threshold, identify a dominant damage factor based on a road surface risk early warning result.

[0011] Further, road surface monitoring data is obtained, and the road surface monitoring data is preprocessed to obtain preprocessed road surface monitoring data, and structure feature extraction and normalization processing are performed based on the preprocessed road surface monitoring data to obtain road surface structure health index data, including:

[0012] S11, obtain road surface monitoring data based on a precision time protocol, and use an exponential smoothing method of a low-pass filter to smooth the road surface monitoring data to obtain preprocessed road surface monitoring data;

[0013] S12, structure feature extraction is performed on the preprocessed road surface monitoring data to obtain road surface structure feature data, and the road surface structure feature data is normalized by using a standardized score to obtain a road surface structure influence factor;

[0014] S13, the road surface structure influence factor is converted into a health index by using a linear inverse transformation method to obtain road surface structure health index data.

[0015] Further, the preprocessed road surface monitoring data includes preprocessed road surface stress and strain data, preprocessed road surface pressure data, preprocessed road surface temperature and humidity data, and preprocessed road surface settlement data;

[0016] The road surface structure feature data includes road surface strain feature data, road surface load damage strength feature data, road surface temperature amplitude feature data, and road surface settlement feature data;

[0017] The preprocessed road surface stress and strain data is subjected to a root mean square operation to obtain road surface strain feature data;

[0018] The preprocessed road surface pressure data is subjected to load damage strength extraction to obtain road surface load damage strength feature data;

[0019] The preprocessed road surface temperature and humidity data is subjected to a difference calculation of the maximum value and the minimum value to obtain road surface temperature amplitude feature data;

[0020] The preprocessed road surface settlement data is subjected to deviation threshold truncation and relative deviation average processing to obtain road surface settlement feature data.

[0021] Further, the acquired historical road surface monitoring data is converted into a road surface structure probability distribution function, an initial road surface structure weight is determined according to the road surface structure probability distribution function, and the initial road surface weight is dynamically optimized to obtain an optimized road surface structure weight, including:

[0022] S21, probability distribution fitting is performed on the acquired historical road surface monitoring data to obtain a road surface structure probability distribution function;

[0023] S22, calculating the road surface structure dispersion degree according to the road surface structure probability distribution function, and determining the initial road surface structure weight based on the calculation result of the road surface structure dispersion degree;

[0024] S23, updating and optimizing the initial road surface weight by using a nonlinear penalty function to obtain the optimized road surface structure weight.

[0025] Further, the expression for calculating the road surface structure dispersion degree is:

[0026] ;

[0027] ;

[0028] In the formula, E j represents the calculation result of the road surface structure dispersion degree; P j represents the road surface structure probability distribution function; j represents the serial number of the road surface structure probability distribution function; N represents the total number of data samples of the historical road surface monitoring data; w j represents the initial road surface structure weight.

[0029] Further, the expression for updating and optimizing is:

[0030] ;

[0031] ;

[0032] In the formula, represents the optimized road surface structure weight; u i ( t ) represents the nonlinear penalty function; λ i represents the attenuation coefficient; w i ( t ) represents the initial road surface structure weight; θ i represents the historical failure rate of the sensor; μi，N total amount of data samples representing the pre-processed road surface monitoring data N mean value of the total amount of data samples representing the pre-processed road surface monitoring data σ i，N total amount of data samples representing the pre-processed road surface monitoring data N standard deviation of the total amount of data samples representing the pre-processed road surface monitoring data x i t represents the road structure impact factor.

[0033] Further, the road health index is calculated according to the road structure health index data and the optimized road structure weight, and the road risk early warning is performed in combination with the preset health index threshold value, the damage factor identification is performed based on the road risk early warning result, and the dominant damage factor includes:

[0034] S31, based on the optimized road structure weight, the road structure health index data is weighted and summed to obtain the road health index;

[0035] S32, the road health index is compared with the preset health index threshold value, and the road risk early warning is performed according to the comparison result;

[0036] S33, based on the road risk early warning result, the damage attribution analysis and screening are performed by using the road structure health index data and the preset road health baseline value, and the dominant damage factor is obtained.

[0037] Further, the road risk early warning according to the comparison result includes:

[0038] when the comparison result is that the road health index exceeds the upper limit of the preset health index threshold value, it is determined that the road structure is in a dangerous state, and a danger early warning is triggered;

[0039] when the comparison result is that the road health index is within the preset health index threshold value range, it is determined that the road structure is in a risk state, and a preliminary risk early warning is triggered;

[0040] when the comparison result is that the road health index is lower than the lower limit of the preset health index threshold value, it is determined that the road structure is in a normal state, and no road risk early warning is triggered.

[0041] Further, based on the road risk early warning result, the damage attribution analysis and screening are performed by using the road structure health index data and the preset road health baseline value, and the dominant damage factor includes:

[0042] S331, when the road risk early warning result is that the early warning is triggered, a straight line path is constructed according to the road structure health index data and the preset road health baseline value, and a gradient integral calculation is performed based on the straight line path construction result, and the attribution score is determined through the gradient integral calculation result;

[0043] ​S332, the attribution score calculation result is normalized by percentage to obtain a damage contribution degree, and a key damage factor is determined according to the damage contribution degree to obtain a dominant damage factor.

[0044] According to another aspect of the present application, a multi-source data fusion road surface monitoring system is provided, comprising a road surface structure health index acquisition module, a road surface structure weight determination module and a damage factor identification module.

[0045] The road surface structure health index acquisition module is configured to acquire road surface monitoring data, pre-process the road surface monitoring data to obtain pre-processed road surface monitoring data, and extract and normalize structural features based on the pre-processed road surface monitoring data to obtain road surface structure health index data.

[0046] The road surface structure weight determination module is configured to convert the acquired historical road surface monitoring data into a road surface structure probability distribution function, determine an initial road surface structure weight based on the road surface structure probability distribution function, and dynamically optimize the initial road surface weight to obtain an optimized road surface structure weight.

[0047] The damage factor identification module is configured to calculate a road health index based on the road surface structure health index data and the optimized road surface structure weight, perform road risk early warning in combination with a pre-set health index threshold, and identify a dominant damage factor based on the road risk early warning result.

[0048] The present application has the following advantages:

[0049] The present application can monitor road surface stress, strain, temperature, humidity, pressure and other key structural data in real time, visualize road health index, accurately feedback road risk and wear degree, improve the comprehensiveness and accuracy of monitoring, identify potential safety hazards in time, reduce the occurrence of safety accidents, effectively solve the defects of data isolation and delayed response in traditional monitoring, eliminate the delay of cloud platform, ensure timely alarm of safety information, improve the timeliness and safety of highway facility maintenance, reduce the risk caused by road damage, and autonomously judge the main damage factors of road health, identify potential safety hazards, provide intelligent decision support for highway management personnel, guide repair work, reduce unnecessary maintenance cost, and improve the scientificity and accuracy of road maintenance management. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description only need to explain the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained based on these drawings without any creative effort.

[0051] Figure 1 is a flow chart of a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0052] Figure 2 is a principle block diagram of a multi-source data fusion road surface monitoring system according to an embodiment of the present application;

[0053] Figure 3 is a multi-source data fusion flow chart in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0054] Figure 4 is a sensor road surface layout diagram in a multi-source data fusion road surface monitoring method according to an embodiment of the present application.

[0055] Figure 5 is one of road surface monitoring data diagrams in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0056] Figure 6 is another of road surface monitoring data diagrams in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0057] Figure 7 is a third of road surface monitoring data diagrams in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0058] Figure 8 is a fourth of road surface monitoring data diagrams in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0059] Figure 9 is a dynamic weight data diagram corresponding to road surface monitoring data in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0060] Figure 10 is a road surface health index and early warning threshold diagram in a multi-source data fusion road surface monitoring method according to an embodiment of the present application;

[0061] Figure 11 is a damage factor probability diagram in a multi-source data fusion road surface monitoring method according to an embodiment of the present application.

[0062] In the drawings:

[0063] 1, road surface structure health index acquisition module; 2, road surface structure weight determination module; 3, damage factor identification module. DETAILED DESCRIPTION

[0064] To further illustrate the embodiments, the present application provides drawings, which are part of the disclosure of the present application, mainly used to illustrate the embodiments, and can explain the operating principle of the embodiments in conjunction with the related description of the specification. With reference to these contents, those skilled in the art should understand other possible implementations and advantages of the present application.

[0065] According to an embodiment of the present application, a multi-source data fusion road surface monitoring method and system are provided.

[0066] The present application will be further described in conjunction with the drawings and specific embodiments, as shown in Figure 1 and Figure 3 According to an embodiment of the present application, a multi-source data fusion road surface monitoring method is provided, which comprises:

[0067] S1, obtaining road surface monitoring data, and preprocessing the road surface monitoring data to obtain preprocessed road surface monitoring data, based on the preprocessed road surface monitoring data, structure feature extraction and normalization processing are performed to obtain road surface structure health index data.

[0068] Specifically, obtaining road surface monitoring data, and preprocessing the road surface monitoring data to obtain preprocessed road surface monitoring data, based on the preprocessed road surface monitoring data, structure feature extraction and normalization processing are performed to obtain road surface structure health index data, comprising:

[0069] S11, obtaining road surface monitoring data based on the precision time protocol, and using the exponential smoothing method of the low-pass filter to smooth the road surface monitoring data to obtain preprocessed road surface monitoring data;

[0070] S12, structure feature extraction is performed on the preprocessed road surface monitoring data to obtain road surface structure feature data, and the road surface structure feature data is normalized by using the standardization score to obtain the road surface structure influence factor.

[0071] Specifically, the preprocessed road surface monitoring data includes: preprocessed road surface stress and strain data, preprocessed road surface pressure data, preprocessed road surface temperature and humidity data, and preprocessed road surface settlement data;

[0072] The road surface structure feature data includes: road surface strain feature data, road surface load damage strength feature data, road surface temperature amplitude feature data, and road surface settlement feature data;

[0073] The root mean square calculation is performed on the preprocessed pavement stress-strain data to obtain pavement strain characteristic data.

[0074] The load failure strength is extracted from the preprocessed pavement pressure data to obtain pavement load failure strength characteristic data.

[0075] The difference between the maximum and minimum values ​​of the preprocessed road surface temperature and humidity data is calculated to obtain the road surface temperature amplitude characteristic data.

[0076] The preprocessed pavement settlement data were truncated by a deviation threshold and averaged by relative deviation to obtain pavement settlement characteristic data.

[0077] S13. Use the linear inverse transformation method to convert the pavement structure influencing factors into health indicators to obtain pavement structure health indicator data.

[0078] Specifically, a cluster of intelligent sensing devices is deployed on the road surface, including concrete thermo-hygrometers, asphalt thermo-hygrometers, earth pressure gauges, concrete strain gauges, asphalt strain gauges, settlement gauges, etc. These sensors are connected to a data acquisition unit to collect real-time, high-frequency structural data such as road surface strain, pressure, settlement, and temperature and humidity—i.e., road surface monitoring data. All data is processed locally through the data acquisition unit's embedded operating system and multi-source data fusion algorithm. The data acquisition unit has a built-in high-precision clock and uses the Precise Time Protocol (PTP) as the unified time source for all sensors. During data acquisition, each data packet from each sensor is timestamped with a uniform millisecond-level timestamp, ensuring strict alignment of all data on the timeline. Noise is removed from the raw sensor data (i.e., road surface monitoring data) using an exponential smoothing method with a low-pass filter, resulting in a smoothed trend sequence after noise removal. z i ( t This refers to the pre-processed road surface monitoring data, in which... α For smoothing coefficients, h i ( t )for t Actual monitoring values ​​during the period z i ( t -1) represents the smoothed value from the previous time step. The calculation expression for the preprocessed pavement monitoring data is as follows:

[0079] ;

[0080] In the formula, z i ( t This represents the smoothed trend sequence after noise removal, i.e., the preprocessed road surface monitoring data; α Indicates the smoothing coefficient; h i (t ) represents t the actual monitoring value of the period, i.e. the road surface monitoring data; z i ( t -1) represents the smoothed value of the last moment.

[0081] The structural features reflecting the road surface state are extracted from the data type trend sequence z i ( t ), including the root mean square value ε ( t ) in the pre-processed road surface stress and strain data, i.e. road surface strain characteristic data, wherein, W represents the maximum moment of the sample data, S represents the total sample data amount, ε i represents the stress value of the sampling point; the road load damage intensity I ( t ) in the pre-processed road surface pressure data, i.e. road load damage intensity characteristic data, wherein, P ( t - i ) represents the current pressure value of the road surface; the temperature daily amplitude reflecting the thermal cycle intensity in the pre-processed road surface temperature and humidity data A T , i.e. road surface temperature amplitude characteristic data, wherein, k ∈(1, 2, 3...24), represents the past 24 hours; the road surface daily average settlement speed C V ( t ) in the pre-processed road surface settlement data, i.e. road surface settlement characteristic data, wherein M =30, represents a month period, V avg represents the overall average speed reduction, V ( t - i ) represents the daily settlement amplitude. The calculation expression of the road surface structure characteristic data is:

[0082] ;

[0083] ;

[0084] ;

[0085] ;

[0086] According to the Z-score standardization formula, the characteristic values of the four kinds of data ε ​t ), I ( t ), A T , C V ( t After normalization, the corresponding impact factors were obtained. x 1( t ), x 2( t ), x 3( t )and x 4( t ), namely, the pavement structure influencing factor, among which y i ( t ) represent the feature values ​​of the four types of data, μ i This represents the mean of the corresponding data. σ i The standard deviation of the corresponding data represents the standard deviation of the road surface. The larger the influence factor, the greater the road surface load and the more severe the damage. Therefore, a health index is obtained for the corresponding structural data. f i ( t ), i ∈(1, 2, 3, 4). The formula for calculating health indicators is:

[0087] ;

[0088] ;

[0089] In the formula, x i ( t () indicates the pavement structure influencing factor; y i ( t ) represents the feature values ​​of the four types of data; μ i This represents the mean of the corresponding data; σ i This represents the standard deviation of the corresponding data; f i This represents the health indicators of the corresponding structured data.

[0090] S2. Convert the acquired historical pavement monitoring data into a pavement structure probability distribution function, determine the initial pavement structure weights based on the pavement structure probability distribution function, and dynamically optimize the initial pavement weights to obtain the optimized pavement structure weights.

[0091] Specifically, the obtained historical road surface monitoring data is converted into a road surface structure probability distribution function, the initial road surface structure weight is determined according to the road surface structure probability distribution function, and the initial road surface weight is dynamically optimized to obtain the optimized road surface structure weight, including:

[0092] S21, probability distribution fitting is performed on the obtained historical road surface monitoring data to obtain a road surface structure probability distribution function;

[0093] S22, calculating the road surface structure dispersion degree according to the road surface structure probability distribution function, and determining the initial road surface structure weight based on the calculation result of the road surface structure dispersion degree.

[0094] Specifically, the expression for calculating the road surface structure dispersion degree is:

[0095] ;

[0096] ;

[0097] In the formula, E j represents the calculation result of the road surface structure dispersion degree; P j represents the road surface structure probability distribution function; j represents the serial number of the road surface structure probability distribution function; N represents the total number of data samples of the historical road surface monitoring data; w j represents the initial road surface structure weight.

[0098] S23, the initial road surface weight is updated and optimized by using a nonlinear penalty function to obtain the optimized road surface structure weight.

[0099] Specifically, the expression for updating and optimizing is:

[0100] ;

[0101] ;

[0102] In the formula, represents the optimized road surface structure weight; u i ( t ) represents a nonlinear penalty function; λ i represents a decay coefficient; w i ( t ) represents the initial road surface structure weight; θ i represents the historical failure rate of the sensor; μ i，Ntotal number of data samples representing pretreated pavement monitoring data N ; σ i，N total number of data samples representing pretreated pavement monitoring data N ; x i ( t ) represents a pavement structure influencing factor.

[0103] Specifically, historical data of strain, pressure, settlement and temperature and humidity in a period of time (i.e. historical pavement monitoring data) are respectively converted into probability distribution functions P 1, P 2, P 3, P 4, and the dispersion thereof is calculated E j , wherein N represents the total number of data samples, j ∈(1, 2, 3, 4). The information dispersion of data E j is smaller, the more information, the greater the information fluctuation, the stronger the discrimination of the data, and the higher the weight, and vice versa. Thus, the initial pavement structure weight is calculated, including the initial stress and strain data weight w 1, the initial pressure weight w 2, the initial temperature and humidity weight w 3, the initial pavement settlement weight w 4. The calculation expression of dispersion and initial pavement structure weight is:

[0104] ;

[0105] .

[0106] The initial pavement structure weight of each data type w j ( t ) is dynamically adjusted. When abnormal fluctuations occur in a sensor data, the weight thereof and the real-time uncertainty of each data source are reduced through a nonlinear penalty function u i ( t ) to obtain an optimized weight , so as to avoid the influence of unreliable data on the overall judgment. The optimized weight is determined by the instantaneous fluctuation of data and the long-term reliability of the sensor, wherein θ i is the historical failure rate of the sensor, σ i，N and μ i，N is the total number of data samples of the data sourceN the mean and standard deviation of the values of the health index of the pavement structure in the past time period; λ is a decay coefficient, determined based on long-term performance of the sensor, such as historical number of failures and failure time performance, λ ∈(0, 1). The calculation expression of the optimized pavement structure weight is:

[0107] ;

[0108] ;

[0109] S3, calculating a pavement health index based on the pavement structure health index data and the optimized pavement structure weight, and performing pavement risk early warning in combination with a preset health index threshold, and identifying a dominant damage factor based on a pavement risk early warning result.

[0110] Specifically, the calculating a pavement health index based on the pavement structure health index data and the optimized pavement structure weight, and performing pavement risk early warning in combination with a preset health index threshold, and identifying a dominant damage factor based on a pavement risk early warning result includes:

[0111] S31, performing weighted summation on the pavement structure health index data based on the optimized pavement structure weight to obtain a pavement health index;

[0112] S32, comparing the pavement health index with the preset health index threshold, and performing pavement risk early warning according to a comparison result.

[0113] Specifically, the performing pavement risk early warning according to a comparison result includes:

[0114] when the comparison result is that the pavement health index exceeds an upper limit of the preset health index threshold, determining that the pavement structure is in a dangerous state, and triggering a danger early warning;

[0115] when the comparison result is that the pavement health index is within a range of the preset health index threshold, determining that the pavement structure is in a risk existing state, and triggering a preliminary risk early warning;

[0116] when the comparison result is that the pavement health index is lower than a lower limit of the preset health index threshold, determining that the pavement structure is in a normal state, and not triggering a pavement risk early warning.

[0117] S33, performing damage attribution analysis and screening based on the pavement risk early warning result, using the pavement structure health index data and a preset pavement health baseline value, to obtain a dominant damage factor.

[0118] Specifically, the performing damage attribution analysis and screening based on the pavement risk early warning result, using the pavement structure health index data and a preset pavement health baseline value, to obtain a dominant damage factor includes:

[0119] S331、When the road surface risk early warning result is a trigger warning, a straight line path is constructed according to the road surface structure health index data and the preset road surface health baseline value, and a gradient integral calculation is performed based on the straight line path construction result, and an attribution score is determined through the gradient integral calculation result;

[0120] S332, the attribution score calculation result is normalized by percentage to obtain a damage contribution degree, and a key damage factor is determined according to the damage contribution degree to obtain a dominant damage factor.

[0121] Specifically, the road surface structure health index data obtained by the above steps f i ( t ) and the optimized road surface structure weight Calculate the real-time road surface health index RHI( t ), f i ( t ) includes f 1( t ), f 2( t ), f 3( t ), f 4( t ), including , , , , if the calculated RHI exceeds the threshold range, that is, exceeds the preset health index threshold, it indicates that the current road surface health state is at risk, and real-time local alarm will be given through the display screen and sound signal mode, and warning information will be sent to road users and management personnel. The calculation expression of the road surface health index is:

[0122] ;

[0123] When the road surface health index is lower than the safety threshold (i.e. the preset health index threshold), the system will also automatically identify the dominant damage factor that causes the health index to decrease. By setting the reference value T of the road surface health state, the health index f i ( t ) of the current monitoring data to the preset road surface health baseline value T is calculated to obtain the attribution score G i , wherein x is an interpolation constant, indicating the length of the path constructed between the current input and the baseline. Finally, the calculated attribution score is normalized to percentage, that is, the contribution degree of each damage factor Ci determining C i The largest feature is the most critical damage factor.

[0124] ;

[0125] ;

[0126] In the formula, x represents an interpolation constant, also represents the length of the path constructed between the current input and the baseline; C i represents the contribution degree of each damage factor.

[0127] As Figure 2 shown, according to another embodiment of the present application, a multi-source data fusion pavement monitoring system is provided, which comprises a pavement structure health index acquisition module 1, a pavement structure weight determination module 2, and a damage factor identification module 3.

[0128] The pavement structure health index acquisition module 1 is used to acquire pavement monitoring data, and pre-process the pavement monitoring data to obtain pre-processed pavement monitoring data, and perform structure feature extraction and normalization processing based on the pre-processed pavement monitoring data to obtain pavement structure health index data.

[0129] The pavement structure weight determination module 2 is used to convert the acquired historical pavement monitoring data into a pavement structure probability distribution function, determine an initial pavement structure weight according to the pavement structure probability distribution function, and dynamically optimize the initial pavement weight to obtain an optimized pavement structure weight.

[0130] The damage factor identification module 3 is used to calculate a pavement health index according to the pavement structure health index data and the optimized pavement structure weight, and perform pavement risk early warning in combination with a preset health index threshold, and identify a dominant damage factor based on the pavement risk early warning result.

[0131] In order to facilitate understanding of the above technical solutions of the present application, the working principle or operation mode of the present application in the actual process will be described in detail below.

[0132] In practical applications, the arrangement of the monitoring facility can be customized according to the characteristics of the road surface, the road is divided into equal intervals along the longitudinal direction, a fixed area test is selected, and a measuring point is randomly selected and marked in the measuring area, and a sensor is arranged to obtain data. Specifically, it includes: using buried asphalt strain gauge, concrete strain gauge, temperature and humidity meter, soil pressure cell and other sensor methods to monitor pavement layer, roadbed layer stress, strain and other pavement and roadbed structure data. The asphalt strain sensor is buried at the bottom of the underlying layer along the center line of the left wheel track of the vehicle, and 8 asphalt strain gauges are installed in a 2x2 matrix arrangement; the concrete strain gauge is arranged along the center line of the road track, and the arrangement method is the same as that of the asphalt strain gauge, and 8 concrete strain gauges are respectively buried in the middle and bottom of the base layer; the soil pressure cell is buried in the road center line below the pavement structure layer, in addition, the base layer and the roadbed below 1m, 1.5m, 2m also need to be arranged, a total of 20. The sensor for monitoring the temperature and humidity environment information of the pavement is installed on the hard shoulder outside the road, of which 18 temperature sensors are buried in the pavement layer, the underlying layer, the middle and bottom of the base layer, the subbase and the roadbed below 1m, 1.5m, 2m; 14 humidity sensors are arranged, and the arrangement position is similar to that of the temperature sensor, but no need to be arranged in the upper layer and the underlying layer. The layered settlement meter is installed directly below the center line of the road, and is arranged at intervals, each interval is about 1m, as shown in Figure 4 .

[0133] The above-mentioned buried sensor equipment is connected to a high-performance acquisition instrument, which collects real-time pavement strain, pressure, temperature, settlement and other structure data, and all data are processed locally through the operating system and multi-source data fusion algorithm embedded in the acquisition instrument. The acquisition instrument uses PTP as the unified time source of all sensors, marks the unified millisecond timestamp for each sensor data, ensures that all data are strictly aligned on the time axis, and avoids fusion deviation caused by time synchronization error. The trend sequence of the original sensor data is shown in Figures 5-8 .

[0134] According to the smooth trend sequence of the four data structure types of stress, strain, pressure, temperature and humidity, and settlement, the corresponding characteristic values are extracted, and the Z-Score standardization formula is used for normalization processing, the influence factor of the corresponding data on the pavement structure health is calculated, the greater the influence factor, the more serious the damage to the pavement health, and the health index of the corresponding pavement monitoring structure data is calculated, as shown in Figure 9According to the historical data samples of the original data of each sensor, the data is converted into a probability distribution, and the dispersion degree of the data source is calculated. The greater the dispersion degree of the data, the smaller the fluctuation degree of the data information, the smaller the influence degree on the road surface, and the smaller the weight of the corresponding data type in the calculation of the road health index RHI. The information amount of the data is converted into an initial objective weight, and the dynamic weight is obtained according to the instantaneous volatility of the data and the long-term historical performance of the sensor, as shown in the following formula: Figure 9

[0135] According to the four kinds of monitoring data, the corresponding health indexes and weights, the current road health index RHI is calculated in real time, the current health risk condition of the road is judged, and a multi-level response mechanism is established, as shown in the following formula: Figure 10 Figure 10 Figure 11 The system publishes warning information according to the range of the current road health index, when RHI is greater than or equal to 0.8, it means that the road running condition is good, and the light is always green; when 0.5 < RHI < 0.8, it means that the road has risks, the light displays yellow warning and emits 70dB intermittent buzzing sound; when RHI is less than 0.5, it means that the road has a sudden safety accident, the alarm device flashes 3Hz red light and accompanies 105dB urgent buzzing sound, and immediately informs the management personnel. The system can identify the dominant damage factors through attribution analysis, automatically identify the dominant factors leading to the deterioration of the road surface, and optimize the road maintenance strategy, as shown in the following formula: Figure 11 Figure 11

[0136] The present application realizes real-time monitoring of road surface strain, temperature and humidity, and settlement data changes through the intelligent perception facility of multi-source data fusion, and simultaneously deploys high-performance acquisition instruments to collect sensor data and locally processes all data. The data fusion algorithm uses embedded multi-source data fusion algorithm to normalize different types of data indicators, dynamically calculates the road health index in real time, and provides early warning information in the first time. The alarm can be sent through the display screen and the sound and light device within 1s when the road surface appears disease, and the full-automatic, high-precision real-time monitoring and millisecond-level risk warning of the road surface running state are comprehensively realized. The present application effectively solves the defects of data isolation and response delay in traditional monitoring, eliminates the delay of the cloud platform, ensures the safety information without delay alarm, improves the timeliness and safety of highway facility maintenance, and reduces the risk caused by road damage.

[0137] Adopt intelligent sensing facilities, all-weather real-time monitoring of pavement stress and strain, temperature and humidity and pressure and other key structure data, fusion analysis of various monitoring data, visualization of pavement health index, accurate feedback of road risk and wear degree, improve the comprehensiveness and accuracy of monitoring, identify potential safety hazards in time, reduce the occurrence of safety accidents. Adopt local edge computing, without relying on the cloud, reduce communication delay and bandwidth pressure, can issue early warning when the road surface risks occur, and can independently judge the main damage factors of road health, identify potential safety hazards, provide intelligent decision support for highway management personnel, guide repair work, reduce unnecessary maintenance cost, improve the scientific nature and accuracy of road maintenance management.

[0138] The above merely describes preferred embodiments of the present application and is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for pavement monitoring by multi-source data fusion, characterized in that, The method comprises: S1, acquiring road surface monitoring data, and preprocessing the road surface monitoring data to obtain preprocessed road surface monitoring data, performing structural feature extraction and normalization processing based on the preprocessed road surface monitoring data to obtain road surface structure health index data; The S1 comprises: S11, acquiring road surface monitoring data based on the precision time protocol, and performing smoothing processing on the road surface monitoring data by using the exponential smoothing method of a low-pass filter to obtain preprocessed road surface monitoring data; S12, performing structural feature extraction on the preprocessed road surface monitoring data to obtain road surface structure feature data, and performing normalization processing on the road surface structure feature data by using a standardization score to obtain a road surface structure influence factor; S13, performing health index conversion on the road surface structure influence factor by using a linear inverse transformation method to obtain road surface structure health index data; The preprocessed road surface monitoring data comprises preprocessed road surface stress and strain data, preprocessed road surface pressure data, preprocessed road surface temperature and humidity data, and preprocessed road surface settlement data; The road surface structure feature data comprises road surface strain feature data, road surface load damage strength feature data, road surface temperature amplitude feature data, and road surface settlement feature data; The preprocessed road surface stress and strain data is subjected to a root mean square operation to obtain the road surface strain feature data; The preprocessed road surface pressure data is subjected to load damage strength extraction to obtain the road surface load damage strength feature data; The preprocessed road surface temperature and humidity data is subjected to a difference calculation of the maximum value and the minimum value to obtain the road surface temperature amplitude feature data; The preprocessed road surface settlement data is subjected to deviation threshold truncation and relative deviation average processing to obtain the road surface settlement feature data; S2, converting the acquired historical road surface monitoring data into a road surface structure probability distribution function, determining an initial road surface structure weight according to the road surface structure probability distribution function, and dynamically optimizing the initial road surface weight to obtain an optimized road surface structure weight; The S2 comprises: S21, performing probability distribution fitting on the acquired historical road surface monitoring data to obtain a road surface structure probability distribution function; S22, calculating a road surface structure dispersion degree according to the road surface structure probability distribution function, and determining an initial road surface structure weight based on the calculation result of the road surface structure dispersion degree; S23, performing optimization and update on the initial road surface weight by using a nonlinear penalty function to obtain an optimized road surface structure weight; The expression for calculating the road surface structure dispersion degree is: ； ； In the formula, E j represents the calculation result of the discrete degree of the pavement structure; P j represents the pavement structure probability distribution function; j represents the serial number of the pavement structure probability distribution function; N represents the total number of data samples of the historical pavement monitoring data; w j represents the initial pavement structure weight; The expression for optimization and update is: ； ； In the formula, represents the optimized pavement structure weight; u i ( t ) represents a nonlinear penalty function; λ i represents a decay coefficient; w i ( t ) represents an initial pavement structure weight; θ i represents a sensor historical failure rate; μ i，N represents the mean of the total amount of data samples of the pretreated pavement monitoring data N ; σ i，N represents the standard deviation of the total amount of data samples of the pretreated pavement monitoring data N ; x i ( t ) represents a pavement structure influence factor; S3, calculating a road surface health index according to the road surface structure health index data and the optimized road surface structure weight, performing road surface risk early warning in combination with a preset health index threshold, and identifying a dominant damage factor based on a road surface risk early warning result. 2.The multi-source data fusion pavement monitoring method of claim 1, wherein, The calculation of the road surface health index according to the road surface structure health index data and the optimized road surface structure weight, the road surface risk early warning in combination with the preset health index threshold, and the identification of the dominant damage factor based on the road surface risk early warning result comprise: S31, based on the optimized pavement structure weight, the pavement structure health index data is weighted and summed to obtain a pavement health index; S32, compare the pavement health index with the preset health index threshold, and perform pavement risk warning according to the comparison result; S33, based on the pavement risk warning result, the pavement structure health index data and the preset pavement health baseline value are used for damage attribution analysis and screening to obtain a dominant damage factor. 3.The multi-source data fusion method of claim 2, wherein, The pavement risk warning according to the comparison result comprises: When the comparison result is that the pavement health index exceeds the upper limit of the preset health index threshold, it is determined that the pavement structure is in a dangerous state, and a danger warning is triggered; When the comparison result is that the pavement health index is within the preset health index threshold range, it is determined that the pavement structure is in a risk state, and a preliminary risk warning is triggered; When the comparison result is that the pavement health index is lower than the lower limit of the preset health index threshold, it is determined that the pavement structure is in a normal state, and no pavement risk warning is triggered. 4.The multi-source data fusion method of claim 3, wherein, The damage attribution analysis and screening based on the pavement risk warning result, using the pavement structure health index data and the preset pavement health baseline value, to obtain a dominant damage factor, comprises: S331, when the pavement risk warning result is to trigger warning, a straight line path is constructed according to the pavement structure health index data and the preset pavement health baseline value, and gradient integral calculation is performed based on the straight line path construction result, and the attribution score is determined by the gradient integral calculation result; S332, the attribution score calculation result is normalized by percentage to obtain a damage contribution degree, and the key damage factor is determined according to the damage contribution degree to obtain a dominant damage factor.

5. A multi-source data fusion pavement monitoring system for implementing the multi-source data fusion pavement monitoring method of any one of claims 1-4, characterized in that, The multi-source data fusion pavement monitoring system comprises: a pavement structure health index acquisition module, a pavement structure weight determination module and a damage factor identification module; The pavement structure health index acquisition module is used for acquiring pavement monitoring data, and pre-processing the pavement monitoring data to obtain pre-processed pavement monitoring data, and performing structure feature extraction and normalization processing based on the pre-processed pavement monitoring data to obtain pavement structure health index data; The pavement structure weight determination module is used for converting the obtained historical pavement monitoring data into a pavement structure probability distribution function, determining an initial pavement structure weight according to the pavement structure probability distribution function, and dynamically optimizing the initial pavement weight to obtain an optimized pavement structure weight; The damage factor identification module is used for calculating a pavement health index according to the pavement structure health index data and the optimized pavement structure weight, and performing pavement risk warning in combination with a preset health index threshold, and identifying a damage factor based on the pavement risk warning result to obtain a dominant damage factor.

Images