CN114323512B - Heavy-load vehicle identification method and system - Google Patents
Heavy-load vehicle identification method and system Download PDFInfo
- Publication number
- CN114323512B CN114323512B CN202111553384.6A CN202111553384A CN114323512B CN 114323512 B CN114323512 B CN 114323512B CN 202111553384 A CN202111553384 A CN 202111553384A CN 114323512 B CN114323512 B CN 114323512B
- Authority
- CN
- China
- Prior art keywords
- heavy
- vehicle
- vibration acceleration
- wavelet
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Abstract
The invention discloses a heavy-duty vehicle identification method and a system, comprising the following steps: collecting vibration acceleration response data of a main girder structure when a vehicle passes through a bridge; wavelet analysis is carried out on vibration acceleration response data of the main beam structure, the bridge crossing time of the heavy-duty vehicle is identified, and the wavelet characteristic value and the bridge crossing speed of the heavy-duty vehicle are extracted; and inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle to recognize the vehicle load. The advantages are that: the acceleration monitoring response data is simple and stable to obtain, displacement and strain monitoring data are not needed, and meanwhile, compared with a dynamic weighing method, the method has the advantages of simplicity and convenience in installation and maintenance, low cost and no damage to bridge structures; the method fully excavates the time-frequency information of the vibration acceleration data of the bridge structure, has the characteristics of high identification precision, rapidness, accuracy and real-time identification, and meets the monitoring requirement of heavy-duty vehicles in the bridge health monitoring system.
Description
Technical Field
The invention relates to a heavy-duty vehicle identification method and system, and belongs to the technical field of automatic recognition of heavy-duty vehicles.
Background
The bridge is taken as an important component of a transportation line, plays an important role in promoting national economic development and people's life, but the repeated action of the bridge super-heavy vehicle can cause damage to the whole or partial components of the bridge, so that the accurate understanding of the traffic condition of the heavy vehicle has important significance for evaluating the performance of the highway bridge.
The existing overweight vehicle identification method in the bridge health monitoring system is mainly a dynamic weighing method (BWIM), however, the BWIM method has the defects of high installation and maintenance cost, traffic sealing during installation, road surface damage, low precision and instability during multi-vehicle bridge passing, and the like, and on the other hand, the bridge health monitoring system collects a large amount of stable acceleration vibration data, and the information contained in the acceleration vibration data is not fully mined.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a heavy-duty vehicle identification method and a system.
In order to solve the technical problems, the invention provides a heavy-duty vehicle identification method, which comprises the following steps:
collecting vibration acceleration response data of a main girder structure when a vehicle passes through a bridge;
wavelet analysis is carried out on the vibration acceleration response data of the main beam structure, the bridge crossing time of the heavy-duty vehicle is identified, and the wavelet characteristic value and the vehicle speed are extracted;
and inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle to recognize the vehicle load.
Further, the collecting the vibration acceleration response data of the main beam structure when the vehicle passes through the bridge comprises the following steps:
vibration acceleration response data of the girder structure are collected through a vibration acceleration sensor arranged at the bottom of the girder when a vehicle passes through.
Further, the vibration acceleration sensor is provided with two vibration acceleration sensors which are respectively arranged at the bottom positions of different sections of the main beam, including but not limited to the bottom positions of different cross sections of the main beam, or the bottom positions of different cross sections of the main beam which are spanned by the same distance but have a distance of more than 100 m.
Further, the training process of the neural network model includes:
acquiring vibration acceleration response data of a heavy-duty vehicle with a history passing through a main beam structure and a corresponding vehicle-mounted vehicle;
performing wavelet analysis on vibration acceleration response data of the heavy-duty vehicle passing through the main beam structure, and extracting wavelet characteristic values and vehicle speed;
and taking the wavelet characteristic value and the vehicle speed as inputs, taking the corresponding vehicle as output, and training the dynamic recognition neural network model of the heavy-duty vehicle based on machine learning to obtain a trained dynamic recognition neural network model of the heavy-duty vehicle.
Further, the wavelet analysis is performed to extract a wavelet characteristic value and a vehicle speed, and the method comprises the following steps:
using continuous wavelet transform to transform vibration acceleration datax(t) Converted into CWT of time-frequency dataf, t)；
From time-frequency data CWTf, t) Time when heavy-duty vehicle passes through vibration acceleration sensor mounting section is identifiedT ins ；
Moment when heavy-duty vehicle passes through vibration acceleration mounting section based on identificationT ins And extracting wavelet characteristic values of a low frequency band, and carrying out correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the two identified vibration accelerations.
Further, the slave time frequency data CWT #f, t) Time when heavy-duty vehicle passes through vibration acceleration sensor mounting section is identifiedT ins Comprising:
CWT for continuous time-frequency dataf, t) Time domain of (a)tConverting envelope into discrete time frequency data CWTf, T)；
According to CWT of discrete time dataf, T) Based on a predetermined frequencyf high Identifying information about time domain when a vehicle passes through a vibration acceleration sensor mounting sectionTWave crest CWT max （f high , T ins ) WhereinT ins Corresponding to the peakEtching;
for the identified wave crest CWT max （f high , T ins ) Setting a threshold YZ, and identifying the moment when the heavy-duty vehicle passes through the installation section of the vibration acceleration sensorT ins 。
Further, the high frequencyf high Comprises:
selecting vibration acceleration data time period caused by light-load vehicle passing through the bridge, and performing continuous wavelet transformation to obtain time-frequency data CWT #f, t) Analyzing the high frequency sensitive regionf lower ~f upper The subscripts lower and upper represent the lower and upper limits,f high the selected range isf high <f lower 。
Further, the moment when the heavy-duty vehicle passes through the vibration acceleration sensor mounting section based on identificationT ins Extracting wavelet characteristic values of a low frequency band, and performing correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the two identified vibration acceleration sensors, wherein the method comprises the following steps:
based on corresponding moments of heavy-duty vehiclesT ins In the frequency spectrum CWT%f, T ins ) Spectrum value CWT of middle extracted low frequency band wave peakf low ，T ins ) Obtaining wavelet characteristic valuesY wave The method comprises the steps of carrying out a first treatment on the surface of the Low frequency band f low Representing the frequency spectrum CWT #f, T ins ) With respect to frequencyfA plurality of frequency values corresponding to peaks in the low frequency band, wavelet characteristic valuesY wave Also contains a plurality of spectral values;
wavelet characteristic values respectively identified for collected data of two vibration acceleration sensorsY wave1 AndY wave2 correlation analysis is carried out to identify different moments when the same heavy-duty vehicle passes through two sections successivelyT ins1 、T ins2 Speed of the vehicleV = D/( T ins2 -T ins1 ) Wherein, the method comprises the steps of, wherein,Dthe longitudinal distance of the cross section is set for two vibration acceleration sensors.
Further, wavelet analysis is carried out on the vibration acceleration response data of the main beam structure, and wavelet characteristic values and vehicle speeds are extracted; inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle, and recognizing the vehicle-mounted vehicle, wherein the method comprises the following steps of:
based on two vibration acceleration sensors, vibration acceleration response data to be identified are collected, wavelet analysis is respectively carried out, and wavelet characteristic values are respectively extractedY wave1 AndY wave2 according to wavelet characteristic valuesY wave1 AndY wave2 performing correlation analysis to obtain vehicle speedV；
Wavelet characteristic valuesY wave2 And vehicle speedVAnd inputting the overweight vehicle identification vehicle to a trained dynamic identification neural network model of the heavy-duty vehicle.
A heavy-duty vehicle identification system comprising:
the acquisition module is used for acquiring vibration acceleration response data of the main beam structure when the vehicle passes through the bridge;
the analysis module is used for carrying out wavelet analysis on the vibration acceleration response data of the main beam structure, identifying the bridge crossing time of the heavy-load vehicle, and extracting wavelet characteristic values and vehicle speed;
the recognition module is used for inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle and recognizing the vehicle load.
The invention has the beneficial effects that:
the heavy-duty vehicle identification method based on bridge vibration acceleration machine learning is simple and stable in acceleration monitoring response data acquisition, does not need displacement and strain monitoring data, and meanwhile has the advantages of being simple and convenient to install and maintain, low in cost and free of damage to bridge structures compared with a BWIM method.
The heavy-duty vehicle identification method based on bridge vibration acceleration machine learning fully excavates time-frequency information of bridge structure vibration acceleration data, has the characteristics of high identification precision, rapidness, accuracy and real-time identification, and can meet the requirement of heavy-duty vehicle monitoring in a bridge health monitoring system to a great extent.
According to the heavy-duty vehicle identification method based on bridge vibration acceleration machine learning, a plurality of vibration acceleration sensors can be used for identifying heavy-duty vehicles at the same time, and the situation that a single heavy-duty vehicle identification system is damaged to cause incorrect mining and missing mining is avoided.
The heavy-duty vehicle identification method based on bridge vibration acceleration machine learning has wide application scene: 1. the invention can be applied to roads or municipal bridges with installed dynamic weighing systems and larger transverse rigidity, and has high practical value in the aspect of data checking and redundancy systems of bridge health monitoring systems; 2. when the dynamic weighing system is not installed on the highway or municipal bridge, the heavy-load vehicle identification method based on bridge vibration acceleration machine learning can replace the dynamic weighing system to monitor the overweight vehicle in real time; 3. as a nested heavy-duty vehicle recognition module, the neural network model trained by the heavy-duty vehicle recognition method based on bridge vibration acceleration machine learning can be used as a nested heavy-duty vehicle recognition module for recognition and monitoring of the heavy-duty vehicles at the positions of standard highway bridges, turn road junctions or bridge entrance openings which are of the same type and have the same span.
Drawings
FIG. 1 is a flow chart of a heavy-duty vehicle identification method based on bridge vibration acceleration machine learning;
FIG. 2 is a flowchart of wavelet feature value extraction;
fig. 3 is a vehicle speed identification flowchart.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
As shown in fig. 1, a heavy-duty vehicle identification method based on bridge vibration acceleration machine learning comprises the following steps:
the first step: acquiring vibration acceleration response of a main beam structure and corresponding vehicle-mounted data serving as sample data;
and a second step of: performing wavelet analysis on the vibration acceleration sample data;
and a third step of: identifying the time when the heavy-duty vehicle passes through the installation section of the vibration acceleration sensor;
fourth step: extracting wavelet characteristic values and vehicle speed of a heavy-duty vehicle;
fifth step: the wavelet characteristic value and the vehicle speed are used as inputs, the corresponding vehicle is used as an output, and the dynamic recognition neural network model of the heavy-duty vehicle is trained based on machine learning;
sixth step: performing wavelet analysis on vibration acceleration response data to be identified, and extracting wavelet characteristic values and vehicle speed;
seventh step: the wavelet characteristic values and the vehicle speed are input into a trained neural network model to identify overweight vehicles.
The bridge heavy-duty vehicle recognition technology based on vibration acceleration and machine learning comprises the following specific steps: and acquiring vibration acceleration response data acquired by a vibration acceleration sensor arranged at the bottom of the main beam when the heavy-duty vehicle passes through, and taking corresponding vehicle-mounted data as a sample.
Vibration acceleration sensor number and mounting position: the vibration acceleration sensor needs 2 vibration acceleration sensors, and according to bridge component information and material characteristics, the installation positions respectively comprise but are not limited to different cross section bottom positions of the main beam or different cross section bottom positions of the main beam with the same span interval of more than 100 m;
vehicle sample data sources: the vehicle-mounted sample data acquisition method corresponding to the vibration acceleration response comprises, but is not limited to, a bridge dynamic weighing system or a vehicle load used for calibration in a load test;
sample number: the sample data volume mainly collects vertical response and vehicle-mounted performance when the heavy-duty vehicle passes through the bridge, the sample data volume influences the accuracy of a vehicle-mounted recognition technology, and the recognition effect is better when the sample volume is larger.
The method for identifying the heavy-duty vehicle based on bridge vibration acceleration machine learning comprises the following specific steps: and carrying out wavelet analysis on the vibration acceleration sample data to obtain time-frequency data.
Wavelet transformation: wherein the wavelet analysis employs a continuous wavelet transform whose wavelet functions include, but are not limited to, an amor wavelet, vibration acceleration datax(t) Wavelet transformed into time-frequency data CWT #f, t) CWT is Continuous Wavelet Transform acronym.
Time-frequency data CWT #f, t) Is characterized in that: high resolution in the frequency domain at the low frequency band and high resolution in the time domain at the high frequency band.
The method for identifying the heavy-duty vehicle based on bridge vibration acceleration machine learning comprises the following specific steps: from time-frequency data CWTf, t) The time when the heavy-duty vehicle passes through the vibration acceleration sensor mounting section is identified.
Envelope is made: the heavy-duty vehicle has the characteristic of multiple axles, namely the same heavy-duty vehicle has CWT (continuous wave) on time frequencyf, t) Multiple time domains can appeartPeak of time-frequency data CWT%f, t) Time domain of (a)tConverting envelope into CWT%f, T) To eliminate the multi-axis effect of the heavy-duty vehicle.
Selecting a higher frequencyf high : CWT based on time-frequency dataf, T) The high frequency band has the characteristic of high resolution in the time domain, and a higher frequency is selectedf high (whereinfFrequency, higher index) based onf high Identifying information about time domain when a vehicle passes through a vibration acceleration sensor mounting sectiontWave crest CWT max （f high , T ins ) WhereinT ins The time corresponding to the peak (subscripts ins are the 3 letters preceding the instance).
Concrete embodimentsf high When determining, selecting the light-load vehicle<5 t) vibration acceleration data period caused by bridge crossing and converting continuous wavelet into time-frequency data CWT #f, t) Analyzing the high frequency sensitive regionf lower ~f upper (subscripts lower and upper represent lower and upper limits), specific parametersf high The selection is as follows:f high <f lower 。
setting a threshold value YZ: for the identified wave crest CWT max （f high , T ins ) Setting a threshold valueYZ (YZ is Chinese Pinyin Yuzhi first letter) initially identifying moment of heavy-duty vehicle passing through vibration acceleration sensor mounting sectionT ins 。
The heavy-duty vehicle identification method based on bridge vibration acceleration machine learning comprises the following specific steps: moment when heavy-duty vehicle passes through vibration acceleration sensor installation section based on identificationT ins And extracting wavelet characteristic values of a low frequency band, and carrying out correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the identified 2 vibration acceleration sensors.
Wavelet eigenvaluesY wave And (3) identification: the heavy-duty vehicle can cause the low-frequency vibration of the main beam, and the wavelet is in a low frequency bandf low Has the characteristic of high resolution in the frequency domain and is based on the corresponding moment of the heavy-duty vehicleT ins Can be used in the CWT frequency spectrumf, T ins ) Spectrum value CWT of middle extracted low frequency band wave peakf low ，T ins ) I.e. wavelet characteristic valuesY wave (wherein the subscript wave stands for wavelet).
Selecting lower frequency bandsf low : wherein the method comprises the steps off low (wherein the subscript low represents a low frequency) represents the spectrum CWT #f, T ins ) With respect to the frequency domainfA plurality of frequency values corresponding to peaks in the low frequency band, wavelet characteristic valuesY wave Also contains a plurality of spectral values.
Vehicle speedVAnd (3) identification: wavelet characteristic values of heavy-duty vehicle when passing through bridge are respectively identified by collecting data of 2 vibration acceleration sensorsY wave1 AndY wave2 correlation analysis is carried out to identify different moments when the same heavy-duty vehicle passes through two sections successivelyT ins1 、T ins2 Speed of the vehicleV = D/( T ins2 -T ins1 )。
Wherein the method comprises the steps ofDThe longitudinal distance of the section is set for 2 vibration acceleration sensors.
The method for identifying the heavy-duty vehicle based on bridge vibration acceleration machine learning comprises the following specific steps: wavelet eigenvaluesY wave And the vehicle speed is used as input, the corresponding vehicle is used as output, and the neural network model for dynamically identifying the heavy-duty vehicle is trained based on machine learning.
Wherein the wavelet eigenvalues of the inputY wave Selecting wavelet characteristic values identified by data collected by vibration acceleration sensor 2Y wave2 The vibration acceleration sensor 1 can be used for acquiring data of the vehicle speedVIs a function of the identification of the device.
The method for identifying the heavy-duty vehicle based on bridge vibration acceleration machine learning comprises the following specific steps: wavelet analysis is respectively carried out on the vibration acceleration response data to be identified, which are collected based on the vibration acceleration sensor 1 and the vibration acceleration sensor 2, and wavelet characteristic values are respectively extractedY wave1 、Y wave2 And vehicle speedV。
The method for identifying the heavy-duty vehicle based on bridge vibration acceleration machine learning comprises the following specific steps: wavelet characteristic value for identifying data acquired by vibration acceleration sensor 2Y wave2 And vehicle speedVInput to the trained neural network model identifies overweight vehicles.
The invention also provides a heavy-duty vehicle identification system, which comprises:
the acquisition module is used for acquiring vibration acceleration response data of the main beam structure when the vehicle passes through the bridge;
the analysis module is used for carrying out wavelet analysis on the vibration acceleration response data of the main beam structure, identifying the bridge crossing time of the heavy-duty vehicle, and extracting wavelet characteristic values and the vehicle speed;
the recognition module is used for inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle and recognizing the vehicle load.
In the embodiment, a left continuous bridge and a right continuous bridge in China are taken as an example for explanation, the span of a main bridge is 78+138+110+58m, the cross section of a Liang Shanfu bridge adopts a single-box double-chamber prestressed concrete box structure, the monitoring data of a vibration acceleration sensor 1 and a vibration acceleration sensor 2 at a midspan of a span 1 and a span 3 of east are selected as research objects, a vehicle speed axle instrument is a DCS-P2000 high-speed dynamic weighing system of a Beijing Zhi Lian Tianchi, the installation position is positioned at the upper bridge end, and the time span is 2021 year 8 month 1 day to 2021 year 8 month 31 day. The method comprises the following steps:
the first step: acquiring vibration acceleration response of a main beam structure and corresponding vehicle-mounted data as a sample;
vibration acceleration data acquired based on bridge health monitoring system at 2 sections of midspan base plates of 1 st span and 3 rd span of main beam are respectively as followsx 1 (t) Andx 2 (t) Sampling frequencyF=50 Hz, the time span for collecting sample data is 2021, 8 months 1, 0-23h; and meanwhile, the monitoring data of the dynamic weighing system are used as vehicle-mounted sample data.
And a second step of: performing wavelet analysis on the vibration acceleration sample data;
based on 2 section acceleration response datax 1 (t) Andx 2 (t) Respectively performing continuous wavelet transformation, and converting into CWT by using amor wavelet 1 (f, t) And CWT 2 (f, t) The low frequency band has the characteristics of high resolution in the frequency domain and high resolution in the time domain.
Wherein for the convenience of calculation, 1 day sample data volumex 1 (t) Andx 2 (t) All divided into 24 sections of data volume, each section of data volume is subjected to continuous wavelet transformation respectively for 1h time length, and the data volume is spliced into CWT 1 (f, t) And CWT 2 (f, t)。
And a third step of: identifying the time when the heavy-duty vehicle passes through the installation section of the vibration acceleration sensor;
at CWT i (f, t)，iTime domain of =1, 2tUp-conversion of envelope into CWT i (f, T) Parameter setting: both vibration acceleration sensors take high frequencyf high Threshold YZ of vibration acceleration sensor 1 =11 Hz 1 Threshold YZ of vibration acceleration sensor 2 =1.8 2 =3; respectively identifying time domainsTWave crest CWT max （f high , T ins ) WhereinT ins The moment when the heavy-duty vehicle passes through the vibration acceleration sensor mounting section.
Fourth step: extracting wavelet characteristic values and vehicle speeds of heavy-duty vehicles (figures 2 and 3);
parameter setting: low-frequency band of two vibration acceleration sensorf low =2.3 Hz to 3.5Hz; extracting wavelet characteristic valueY wave1 AndY wave2 identifying moment of same heavy-duty vehicle passing through two sections based on correlation analysisT ins1 AndT ins2 the heavy-duty vehicle is driven at a speedV=D/ (T ins2 -T ins1 ) WhereinDMounting cross-section longitudinal distance for two vibration acceleration sensorsD=316m。
Fifth step: the wavelet characteristic value and the vehicle speed are used as inputs, the corresponding vehicle is used as an output, and the dynamic recognition neural network model of the heavy-duty vehicle is trained based on machine learning;
wavelet characteristic value of data acquired by vibration acceleration sensor 2Y wave2 And an identified vehicle speedV200 groups of sample data are taken as input, 200 monitoring value samples of the dynamic weighing system corresponding to the input samples are taken as output, and the neural network model is trained based on machine learning.
Sixth step: performing continuous wavelet analysis on vibration acceleration response data to be identified, and extracting wavelet characteristic values and vehicle speed;
the time span of vibration acceleration data to be identified is 2021, 8 months, 1 to 2021, 8 months, 31 days, the vibration acceleration data to be identified is divided into data quantity per hour based on vibration acceleration response data acquired by the vibration acceleration sensor 1 and the vibration acceleration sensor 2, amor wavelets are respectively adopted for continuous wavelet analysis, and the moment when a heavy-duty vehicle passes through the mounting section of the two vibration acceleration sensors is extractedT ins1 AndT ins2 corresponding wavelet characteristic valueY wave1 、Y wave2 And vehicle speedV。
Seventh step: the wavelet characteristic values and the vehicle speed are input into a trained neural network model to identify overweight vehicles.
Wavelet characteristic value for identifying data acquired by vibration acceleration sensor 2Y wave2 And vehicle speedVInput to the trained neural network model identifies overweight vehicles.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (5)
1. A method of identifying a heavy-duty vehicle, comprising:
collecting vibration acceleration response data of a main girder structure when a vehicle passes through a bridge;
wavelet analysis is carried out on the vibration acceleration response data of the main beam structure, the bridge crossing time of the heavy-duty vehicle is identified, and the wavelet characteristic value and the bridge crossing speed of the heavy-duty vehicle are extracted;
inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle, and recognizing the vehicle load;
the training process of the neural network model comprises the following steps:
acquiring vibration acceleration response data of a heavy-duty vehicle with a history passing through a main beam structure and a corresponding vehicle-mounted vehicle;
performing wavelet analysis on vibration acceleration response data of the heavy-duty vehicle passing through the main beam structure, and extracting wavelet characteristic values and vehicle speed;
taking the wavelet characteristic value and the vehicle speed as inputs, taking the corresponding vehicle as output, and training the dynamic recognition neural network model of the heavy-duty vehicle based on machine learning to obtain a trained dynamic recognition neural network model of the heavy-duty vehicle;
the wavelet analysis is performed, the wavelet characteristic value and the vehicle speed are extracted, and the method comprises the following steps:
using continuous wavelet transform to transform vibration acceleration datax(t) Converted into CWT of time-frequency dataf, t)；
From time-frequency data CWTf, t) Medium recognition heavy-duty vehicleMoment when vehicle passes through vibration acceleration sensor mounting sectionT ins ；
Moment when heavy-duty vehicle passes through vibration acceleration mounting section based on identificationT ins Extracting wavelet characteristic values of a low frequency band, and carrying out correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the data acquired by the two identified vibration acceleration sensors;
the slave time-frequency data CWT #f, t) Time when heavy-duty vehicle passes through vibration acceleration sensor mounting section is identifiedT ins Comprising:
CWT for continuous time-frequency dataf, t) Time domain of (a)tConverting envelope into discrete time frequency data CWTf, T)；
According to CWT of discrete time dataf, T) Based on a predetermined frequencyf high Identifying information about time domain when a vehicle passes through a vibration acceleration sensor mounting sectionTWave crest CWT max （f high , T ins ) WhereinT ins The time corresponding to the wave crest;
for the identified wave crest CWT max （f high , T ins ) Setting a threshold YZ, and identifying the moment when the heavy-duty vehicle passes through the installation section of the vibration acceleration sensorT ins ；
High frequencyf high Comprises:
selecting vibration acceleration data time period caused by light-load vehicle passing through the bridge, and performing continuous wavelet transformation to obtain time-frequency data CWT #f, t) Analyzing the high frequency sensitive regionf lower ~f upper The subscripts lower and upper represent the lower and upper limits,f high the selected range isf high <f lower ；
The moment when the heavy-duty vehicle passes through the vibration acceleration sensor mounting section based on identificationT ins Extracting wavelet characteristic values of a low frequency band, performing correlation analysis to identify vehicle speed based on the wavelet characteristic values of the data acquired by the two vibration acceleration sensors, wherein the method comprises the following steps of：
Based on corresponding moments of heavy-duty vehiclesT ins In the frequency spectrum CWT%f, T ins ) Spectrum value CWT of middle extracted low frequency band wave peakf low ，T ins ) Obtaining wavelet characteristic valuesY wave The method comprises the steps of carrying out a first treatment on the surface of the Low frequency bandf low Representing the frequency spectrum CWT #f, T ins ) With respect to frequencyfA plurality of frequency values corresponding to peaks in the low frequency band, wavelet characteristic valuesY wave Also contains a plurality of spectral values;
wavelet characteristic values respectively identified for collected data of two vibration acceleration sensorsY wave1 AndY wave2 performing correlation analysis to identify different moments when the same heavy-duty vehicle passes through two sections sequentiallyT ins1 、T ins2 Speed of the vehicleV = D/( T ins2 -T ins1 ) Wherein, the method comprises the steps of, wherein,Dthe longitudinal distance of the mounting section for two vibration acceleration sensors, and subscripts 1,2 represent the numbers of the vibration acceleration sensors.
2. The method of claim 1, wherein the collecting vibration acceleration response data of the main beam structure of the vehicle passing the bridge comprises:
vibration acceleration response data of the girder structure are collected through a vibration acceleration sensor arranged at the bottom of the girder when a vehicle passes through.
3. The method for recognizing the heavy-duty vehicle according to claim 2, wherein the vibration acceleration sensor is provided in two, and is respectively installed at the bottom positions of different sections of the main beam.
4. The method for identifying the heavy-duty vehicle according to claim 3, wherein wavelet analysis is performed on the vibration acceleration response data of the main beam structure, the bridge crossing time of the heavy-duty vehicle is identified, and the wavelet characteristic value and the vehicle speed are extracted; inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle, and recognizing the vehicle-mounted vehicle, wherein the method comprises the following steps of:
based on two vibration acceleration sensors, vibration acceleration response data to be identified are collected, wavelet analysis is respectively carried out, and wavelet characteristic values are respectively extractedY wave1 AndY wave2 according to wavelet characteristic valuesY wave1 AndY wave2 performing correlation analysis to obtain vehicle speedV；
Wavelet characteristic valuesY wave2 And vehicle speedVAnd inputting the overweight vehicle identification vehicle to a trained dynamic identification neural network model of the heavy-duty vehicle.
5. A heavy-duty vehicle identification system, comprising:
the acquisition module is used for acquiring vibration acceleration response data of the main beam structure when the vehicle passes through the bridge;
the analysis module is used for carrying out wavelet analysis on the vibration acceleration response data of the main beam structure, identifying the bridge crossing time of the heavy-duty vehicle, extracting wavelet characteristic values and vehicle speed, and comprises the following steps:
a training process for a neural network model, comprising:
acquiring vibration acceleration response data of a heavy-duty vehicle with a history passing through a main beam structure and a corresponding vehicle-mounted vehicle;
performing wavelet analysis on vibration acceleration response data of the heavy-duty vehicle passing through the main beam structure, and extracting wavelet characteristic values and vehicle speed;
taking the wavelet characteristic value and the vehicle speed as inputs, taking the corresponding vehicle as output, and training the dynamic recognition neural network model of the heavy-duty vehicle based on machine learning to obtain a trained dynamic recognition neural network model of the heavy-duty vehicle;
the wavelet analysis is performed, the wavelet characteristic value and the vehicle speed are extracted, and the method comprises the following steps:
using continuous wavelet transform to transform vibration acceleration datax(t) Converted into CWT of time-frequency dataf, t)；
From time-frequency data CWTf, t) Time when heavy-duty vehicle passes through vibration acceleration sensor mounting section is identifiedT ins ；
Moment when heavy-duty vehicle passes through vibration acceleration mounting section based on identificationT ins Extracting wavelet characteristic values of a low frequency band, and carrying out correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the data acquired by the two identified vibration acceleration sensors;
the slave time-frequency data CWT #f, t) Time when heavy-duty vehicle passes through vibration acceleration sensor mounting section is identifiedT ins Comprising:
CWT for continuous time-frequency dataf, t) Time domain of (a)tConverting envelope into discrete time frequency data CWTf, T)；
According to CWT of discrete time dataf, T) Based on a predetermined frequencyf high Identifying information about time domain when a vehicle passes through a vibration acceleration sensor mounting sectionTWave crest CWT max （f high , T ins ) WhereinT ins The time corresponding to the wave crest;
for the identified wave crest CWT max （f high , T ins ) Setting a threshold YZ, and identifying the moment when the heavy-duty vehicle passes through the installation section of the vibration acceleration sensorT ins ；
High frequencyf high Comprises:
selecting vibration acceleration data time period caused by light-load vehicle passing through the bridge, and performing continuous wavelet transformation to obtain time-frequency data CWT #f, t) Analyzing the high frequency sensitive regionf lower ~f upper The subscripts lower and upper represent the lower and upper limits,f high the selected range isf high <f lower ；
The moment when the heavy-duty vehicle passes through the vibration acceleration sensor mounting section based on identificationT ins Extracting wavelet characteristic values of a low frequency band, and performing correlation analysis to identify the vehicle speed based on the wavelet characteristic values of the acquired data of the two vibration acceleration sensors, wherein the method comprises the following steps:
based on heavy-duty vehicle correspondenceTime of dayT ins In the frequency spectrum CWT%f, T ins ) Spectrum value CWT of middle extracted low frequency band wave peakf low ，T ins ) Obtaining wavelet characteristic valuesY wave The method comprises the steps of carrying out a first treatment on the surface of the Low frequency bandf low Representing the frequency spectrum CWT #f, T ins ) With respect to frequencyfA plurality of frequency values corresponding to peaks in the low frequency band, wavelet characteristic valuesY wave Also contains a plurality of spectral values;
wavelet characteristic values respectively identified for collected data of two vibration acceleration sensorsY wave1 AndY wave2 performing correlation analysis to identify different moments when the same heavy-duty vehicle passes through two sections sequentiallyT ins1 、T ins2 Speed of the vehicleV = D/( T ins2 -T ins1 ) Wherein, the method comprises the steps of, wherein,Dthe longitudinal distance of the mounting sections of the two vibration acceleration sensors is set, and subscripts 1 and 2 represent the numbers of the vibration acceleration sensors;
the recognition module is used for inputting the wavelet characteristic value and the vehicle speed into a pre-trained dynamic recognition neural network model of the heavy-duty vehicle and recognizing the vehicle load.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111553384.6A CN114323512B (en) | 2021-12-17 | 2021-12-17 | Heavy-load vehicle identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111553384.6A CN114323512B (en) | 2021-12-17 | 2021-12-17 | Heavy-load vehicle identification method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114323512A CN114323512A (en) | 2022-04-12 |
| CN114323512B true CN114323512B (en) | 2023-06-16 |
Family
ID=81053182
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111553384.6A Active CN114323512B (en) | 2021-12-17 | 2021-12-17 | Heavy-load vehicle identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114323512B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112100721A (en) * | 2020-09-07 | 2020-12-18 | 浙江大学 | Bridge structure modal identification method based on mobile crowd sensing and deep learning |
| CN112461358A (en) * | 2020-11-23 | 2021-03-09 | 合肥工业大学 | Bridge modal parameter identification method based on instantaneous frequency of vehicle-bridge system |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467564B (en) * | 2010-11-12 | 2013-06-05 | 中国科学院烟台海岸带研究所 | Remote sensing image retrieval method based on improved support vector machine relevance feedback |
| CN102607680A (en) * | 2012-03-07 | 2012-07-25 | 四川升拓检测技术有限责任公司 | Vibration-based rapid detection method for vehicle load identification for bridges |
| CN103076393B (en) * | 2012-12-28 | 2015-06-03 | 清华大学 | Knocking scan type bridge damage detecting system |
| CN107766877B (en) * | 2017-09-27 | 2020-05-22 | 华南理工大学 | Method for dynamically identifying overweight vehicle in bridge monitoring system |
| CN108763763B (en) * | 2018-05-28 | 2021-10-19 | 东南大学 | Bridge structure strain response abnormity early warning method |
| CN109839175B (en) * | 2019-01-23 | 2021-09-03 | 同济大学 | Bridge live load optimizing and recognizing system |
| CN111144039B (en) * | 2019-12-04 | 2023-04-18 | 东南大学 | Train dynamic weighing system and weighing method based on deep learning |
| CN112529842B (en) * | 2020-11-23 | 2023-02-07 | 天津大学 | Multi-excitation fusion plate structure damage identification method based on wavelet packet energy |
| CN112665695B (en) * | 2020-12-16 | 2021-11-12 | 四川农业大学 | Axle information recognition early warning method and system based on bridge transverse strain |
-
2021
- 2021-12-17 CN CN202111553384.6A patent/CN114323512B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112100721A (en) * | 2020-09-07 | 2020-12-18 | 浙江大学 | Bridge structure modal identification method based on mobile crowd sensing and deep learning |
| CN112461358A (en) * | 2020-11-23 | 2021-03-09 | 合肥工业大学 | Bridge modal parameter identification method based on instantaneous frequency of vehicle-bridge system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114323512A (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | EMD-based stochastic subspace identification of structures from operational vibration measurements | |
| CN105241660A (en) | High-speed rail large-scale bridge performance evaluation method based on health monitoring data | |
| CN107953888B (en) | Road type identification method and system | |
| CN111667688B (en) | Comprehensive traffic information analysis method and system based on pavement vibration signals | |
| CN102539098A (en) | Bridge dynamic load testing method based on neural network technology | |
| CN104331595B (en) | The mobile principal component correlation coefficient process of bridge damnification early warning | |
| CN104850676B (en) | A kind of random traffic flow simulation analogy method of highway bridge | |
| CN102254068A (en) | Multi-scale analyzing method for buffeting response of large-span bridge | |
| CN101950477A (en) | Method and device for processing traffic information | |
| CN111460551A (en) | Simulation, classification and compilation method of two-dimensional vehicle overrun load spectrum | |
| CN103471865A (en) | Train suspension system failure isolation method based on LDA method | |
| CN110321616B (en) | Three-span tower fault diagnosis method based on wavelet transformation and attention model | |
| CN102890750A (en) | Data analysis method for transportation safety recorder | |
| CN116881819A (en) | Stay cable working state monitoring method based on isolated forest | |
| CN113935384B (en) | Bridge health monitoring method and system for self-adaptive decomposition and recognition of signals | |
| CN110867075A (en) | Method for evaluating influence of road speed meter on reaction behavior of driver under rainy condition | |
| CN114323512B (en) | Heavy-load vehicle identification method and system | |
| Qin et al. | Output-only modal analysis based on improved empirical mode decomposition method | |
| CN105404708A (en) | Actually measured acceleration-based outside period motivation and evaluation method for long-span bridge structures | |
| CN114993434B (en) | Vehicle weight identification system and method based on EMD energy entropy of acoustic signals | |
| CN115481669A (en) | Method and system for quickly separating live load strain monitoring signals of highway and railway dual-purpose bridge | |
| CN105702028B (en) | A kind of traffic state judging system based on motor vehicle audio signal | |
| CN112026856B (en) | Intelligent calculation method and system for running speed of rail transit train | |
| CN112446092A (en) | Vehicle structural part damage testing method, device, equipment and storage medium | |
| Silva et al. | Traffic monitoring using a structural health monitoring system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |