CN118260647A - Bridge damage assessment method based on bridge deck image recognition and bridge vibration perception - Google Patents
Bridge damage assessment method based on bridge deck image recognition and bridge vibration perception Download PDFInfo
- Publication number
- CN118260647A CN118260647A CN202410217628.0A CN202410217628A CN118260647A CN 118260647 A CN118260647 A CN 118260647A CN 202410217628 A CN202410217628 A CN 202410217628A CN 118260647 A CN118260647 A CN 118260647A
- Authority
- CN
- China
- Prior art keywords
- bridge
- vehicle
- load
- data
- power response
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000008447 perception Effects 0.000 title claims abstract description 17
- 230000004044 response Effects 0.000 claims abstract description 54
- 238000012544 monitoring process Methods 0.000 claims abstract description 39
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000013528 artificial neural network Methods 0.000 claims abstract description 8
- 238000005457 optimization Methods 0.000 claims abstract description 5
- 238000006073 displacement reaction Methods 0.000 claims description 16
- 238000010276 construction Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 9
- 230000001133 acceleration Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 230000003287 optical effect Effects 0.000 claims description 6
- 230000009471 action Effects 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 5
- 230000036541 health Effects 0.000 claims description 5
- 230000007613 environmental effect Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000005284 excitation Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 230000010365 information processing Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 238000004643 material aging Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Abstract
The invention discloses a bridge damage assessment method based on bridge deck image recognition and bridge vibration perception, which comprises the following steps: establishing a heavy-duty truck registration database, identifying and capturing truck vehicles running on a bridge deck, searching load information of corresponding vehicles in the heavy-duty truck registration database, acquiring power response time sequence data caused when the vehicles pass the bridge by utilizing power response monitoring sensing equipment, and finally generating specific vehicle-load-power response data; and constructing a vehicle load recognition model based on bridge vibration perception based on a deep neural network, training the specific vehicle-load-power response data to obtain a model after training and optimization, inputting structural power response data caused by an unknown vehicle into the vehicle load recognition model to output a predicted load of a random unknown vehicle, and comparing the predicted load with the known vehicle load in a database to judge whether the bridge is damaged. The invention can effectively realize the damage diagnosis of a large number of small and medium bridges and has low operation cost.
Description
Technical Field
The invention belongs to the field of damage assessment of existing bridge structures in an operation environment, and particularly relates to a bridge damage assessment method based on bridge deck image recognition and bridge vibration perception.
Background
The urban process supports the economical flight and the large-scale infrastructure establishment of China, in particular to bridge engineering used as a traffic network hub, on average, the construction of the bridge is over 4 ten thousand per year, and China becomes the center of world civil engineering and bridge construction after Europe and America and Japan of the generation.
The construction of the large-scale bridge in China starts from the later 90s of the last century, so that the structural behavior and the safety condition of the bridge are urgently evaluated for decades of service, the decision of repairing and reinforcing the bridge or dismantling the bridge for retirement needs scientific basis, and the construction and maintenance are advanced in China. With the rapid development of economic construction, the driving density and the vehicle load are rapidly increased. Under the drive of economic benefits, the trend of vehicle enlargement is increased, the number of overrun overload trucks is continuously increased, and under the temptation that more profits are exceeded and the larger profit is, the actual load is usually 100-400% of the nuclear load capacity of the vehicle. Such severe traffic density and overload conditions, and under the combined action of environmental erosion, material aging, and long-term effects of loading, damage accumulation and resistance decay will inevitably occur to the structure.
With the deep research of the digital technology, the bridge monitoring technology with the sensing function is continuously developed and widely applied. Advanced intelligent sensing equipment is adopted, and research such as bridge cluster online monitoring and data processing, safety early warning and state evaluation and maintenance decision and the like under the action of random traffic flow is developed by combining an existing artificial intelligent algorithm, so that health and safety evaluation and prediction of the existing bridge are realized. The method is characterized in that the current single bridge monitoring is forced to develop towards cluster monitoring, offline hysteresis information processing is developed towards real-time online information processing and fragment information analysis is developed towards comprehensive information analysis, the aims of monitoring intellectualization, network interconnection, dynamic real-time, comprehensive coverage and real virtualization are achieved, the science and technology of the service and safety guarantee of the urban major bridge adapting to economy and society are developed, public safety of cities is guaranteed, and the realization of digital city and urban bridge management informatization is promoted.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention aims to provide a bridge damage assessment method based on bridge deck image recognition and bridge vibration perception, which is used for establishing regional monitoring of bridge structure deformation in a road network hierarchy, establishing a long-term deformation evolution rule of a same-factor analysis structure for a single bridge and realizing recognition and diagnosis of the bearing capacity degradation condition of the bridge; and for the bridge cluster, screening and acquiring the same heavy vehicle based on the acquired vehicle information, and realizing the evaluation and prediction of the relative health states of different bridge structures.
The technical scheme adopted by the invention is as follows:
A bridge damage assessment method based on bridge deck image recognition and bridge vibration perception comprises the following steps:
s1, connecting a truck access registration information database of factories, construction sites or highway bayonets near bridge sites, and acquiring load information and license plate information of an access cargo truck to obtain a heavy-load truck registration database;
S2, setting power response monitoring sensing equipment on the bridge so as to acquire real-time power response data of the bridge structure under the action of environmental excitation;
s3, capturing the appearance and license plate information of a bridge deck running vehicle based on bridge deck monitoring equipment and an image recognition algorithm;
S4, aiming at the recognized truck vehicles in the step S3, searching load information of corresponding vehicles in a heavy truck registration database, and generating specific vehicle-load data; according to the specific vehicle passing time, the power response time sequence data caused when the vehicle passes through the bridge is obtained by utilizing the power response monitoring sensing equipment in the step S2, and the specific vehicle-load-power response data is finally generated;
s5, constructing a vehicle load recognition model based on bridge vibration sensing based on the deep neural network, taking the known load-corresponding power response data of the specific vehicle obtained in the step S4 as a training data set, inputting the training data set into the vehicle load recognition model for model training, and obtaining the vehicle load recognition model based on bridge vibration sensing after training and optimization;
S6, inputting structural power response data caused by the unknown vehicle into a vehicle load identification model, and outputting the predicted load of the random unknown vehicle;
S7, automatically identifying load information of vehicles randomly running through the bridge deck based on the vehicle load identification model, and comparing the load information with the known vehicle load in the heavy-duty truck registration database in the step S1 to judge whether the bridge is damaged.
Further, the power response monitoring sensing device in the step S2 comprises bridge deck vehicle graphic acquisition equipment, a vibration acceleration sensor and a displacement sensor, wherein the displacement sensor comprises at least one of a dynamic displacement optical sensor and a laser displacement sensor, and the bridge deck vehicle graphic acquisition equipment, the vibration acceleration sensor, the dynamic displacement optical sensor and the laser displacement sensor are respectively used for acquiring vehicle apparent identity information, bridge vibration data, bridge girder dynamic deflection data and bridge girder dynamic corner data of a randomly driven bridge deck vehicle.
Further, the vehicle load recognition model constructed in step S5 has an input and output formula expressed as:
Where a represents the vehicle load output, g (·) represents the activation function, w i represents the weight of the ith input signal, x i represents the ith input signal, the power response data obtained by the power response monitoring sensing device when the vehicle passes through the bridge is taken as the input signal data, b represents the bias, the bias term b is a key parameter of the activation function, it determines the initial position of the input signal in the activation function, the relationship of the bias in the formula is "+" sign, and the purpose of the bias is to increase the robustness of the enhancement model.
Further, x i represents bridge vibration data, and one of bridge girder dynamic deflection and bridge girder dynamic rotation angle, i represents the input number of such signals. For example, bridge vibration data and girder dynamic deflection data are used as input vector data, or bridge vibration data and girder dynamic rotation angle data are used as input vector data, and the monitoring data not only represent real dynamic response data of the structure, but also have a one-to-one mapping relation with the load of the vehicle.
Further, step S5 further includes a feature extraction and learning process of the model, including: the method comprises the steps of firstly decomposing EEMD based on an integrated empirical mode, decomposing dynamic response data obtained by dynamic response monitoring sensing equipment when a vehicle passes through a bridge into a plurality of modal components, removing components irrelevant to the vehicle load in the components, and obtaining residual modal components which are dynamic response monitoring data relevant to the vehicle load; then, based on the fully connected neural network FCN, establishing a mapping relation between dynamic response monitoring data related to the vehicle load and the actual vehicle load, and forming an identification model of the vehicle load; as the monitoring data is continuously accumulated, the self-repairing and optimizing of the model are realized.
Further, in step S7, when the error between the predicted vehicle load data and the known vehicle load in the heavy truck registration database is within 5%, the bridge health is determined, otherwise, the bridge damage is indicated.
The beneficial effects obtained by the invention are as follows:
1) The bridge damage assessment method based on bridge deck image recognition and bridge vibration perception adopts a low-cost monitoring hardware means, combines the bridge damage assessment technical process provided by the invention, combines the existing artificial intelligent algorithm, and can effectively realize damage diagnosis of a large number of small and medium bridges, and has low operation cost.
2) The trucks at factories, sites or highway bays near the bridge site come in and go out of the same batch of vehicles, and the weighing systems of the vehicles are consistent, so that the weight measurement errors of the same vehicles are almost negligible. In addition, the method generates the vehicle load identification model with strong robustness through multi-source and massive heavy vehicle load data, the vehicle load identification precision can be improved along with the increase of the data quantity, and the cost can be greatly reduced by utilizing the existing resources to acquire the load data of the loading and unloading goods wagon.
Drawings
Fig. 1 is a schematic flow chart of a bridge damage assessment method based on bridge deck image recognition and bridge vibration perception.
FIG. 2 is a schematic representation of the actual flow of the various parts of the process of the present invention.
Fig. 3 is a database of incomplete information of the deck random driving vehicles constructed by S1 and S2 in the process of the method of the present invention.
Fig. 4 is a set of lightweight monitoring hardware system suitable for middle and small bridges, which is proposed by the method of the invention.
Fig. 5 is a map of the dynamic response data of a bridge generated by S4 and S5 in the process of the method of the present invention for a specific known load vehicle.
Fig. 6 is a schematic flow chart of the method of the invention for evaluating bridge damage based on random vehicles.
Detailed Description
The invention will be further illustrated with reference to specific examples, but the scope of the invention is not limited thereto.
A bridge damage assessment method based on structural dynamic corner monitoring under random traffic flow is shown in fig. 1 and 2, and comprises the following steps:
S1, capturing appearance and license plate information of traffic flow randomly running through a bridge floor based on bridge floor monitoring equipment and an image recognition algorithm;
s2, connecting a truck access registration information database of factories, construction sites or highway bayonets near the bridge site, and acquiring load information and license plate information of the truck for accessing the cargo;
According to the steps S1 and S2, a non-complete information database of the bridge deck random running vehicles is constructed, and apparent identity information of the vehicles is captured and acquired based on bridge deck monitoring equipment; acquiring in-out vehicle registration data through factories, construction sites or highway bayonets near the bridge site; and combining the vehicle identification information and the load data to form an incomplete database of the bridge deck traveling vehicle, as shown in fig. 3.
S3, setting power response monitoring sensing equipment on the bridge so as to acquire real-time power response data of the bridge structure under the action of environmental excitation;
Step S3, a set of lightweight monitoring hardware system suitable for small and medium bridges is constructed, bridge power response data are obtained, and the power response monitoring sensing equipment comprises: the bridge deck vehicle drawing equipment (optional), the vibration acceleration sensor (optional), the dynamic displacement optical sensor and the laser displacement sensor (optional ) respectively acquire the apparent identity information of the randomly-driven bridge deck vehicle, the bridge vibration data, the bridge girder dynamic deflection data and the bridge girder dynamic corner data.
As shown in fig. 4, the vibration acceleration sensor is installed at the beam end of the bottom surface of the bridge girder and is respectively arranged right below the center lines of the two lanes; the movable displacement optical sensor is arranged at the midspan of the bottom surface of the bridge girder; the laser displacement sensor is arranged at the expansion joint of the beam end of the bottom surface of the bridge girder.
S4, searching load information of the corresponding vehicle in a heavy-duty truck registration database based on license plate information of the passing vehicle acquired by bridge deck drawing equipment, and generating specific vehicle-load data;
S5, acquiring power response time sequence data caused when the vehicle passes through the bridge by the method of the step S3 according to the specific vehicle passing time, and finally generating specific vehicle-load-power response data;
And S4 and S5, generating a map of the vehicle with a specific known load and the bridge dynamic response data according to the incomplete database of the random vehicle and the bridge dynamic response data acquired in the steps S2 and S3 by taking the vehicle identity information as an intermediate tie, as shown in FIG. 5.
S6, constructing a vehicle load recognition model based on bridge vibration perception based on a deep neural network;
And step S6, building a vehicle load identification model of the bridge dynamic response signal based on the neural network. Firstly, a characteristic extraction and learning process of a model comprises the following steps: the method comprises the steps of firstly decomposing EEMD based on an integrated empirical mode, decomposing dynamic response data obtained by dynamic response monitoring sensing equipment when a vehicle passes through a bridge into a plurality of modal components, removing components irrelevant to the vehicle load in the components, and obtaining residual modal components which are dynamic response monitoring data relevant to the vehicle load; then, based on the fully connected neural network FCN, establishing a mapping relation between dynamic response monitoring data related to the vehicle load and the actual vehicle load, and forming an identification model of the vehicle load; as the monitoring data is continuously accumulated, the self-repairing and optimizing of the model are realized.
For a constructed vehicle load identification model, the input and output formulas of the model can be expressed as follows:
Wherein a represents the load output of the vehicle, g (-) represents the activation function, w i represents the weight of the ith input signal, x i represents the ith input signal, the power response data obtained by the power response monitoring sensing equipment when the vehicle passes through the bridge is taken as the input signal data, and i represents the input quantity of the signals. For example, bridge vibration data and bridge girder dynamic deflection data are used as input vector data, or bridge vibration data and bridge girder dynamic rotation angle data are used as input vector data. b denotes the bias.
S7, a training data set of known load-corresponding power response of the specific vehicle is manufactured, the training data set is input into a vehicle load recognition model for model training, and the vehicle load recognition model based on bridge vibration perception after training and optimization is obtained;
S8, inputting structural power response data caused by an unknown vehicle into a vehicle load identification model, and outputting a predicted load of the random unknown vehicle;
and S8, bridge power response data caused by the random running vehicles with unknown loads are used as a learning set to be input into the vehicle load recognition model after training and optimization, and predicted load information of the random running vehicles is output.
S9, automatically identifying load information of vehicles randomly running through the bridge deck based on the vehicle load identification model, and comparing the load information with the known vehicle load in the truck registration information base to judge whether the bridge is damaged.
In the operation process of the bridge, step S9, the bridge power response data caused by random vehicles finally output actual load data of the vehicles through a vehicle load identification model, and the actual load data is compared and analyzed with the actual registration data to judge whether the bridge structure is damaged, when the error between the predicted vehicle load data and the known vehicle load in the heavy truck registration database is within 5%, the bridge health is judged, otherwise, the bridge is damaged, as shown in fig. 6.
What has been described in this specification is merely an enumeration of possible forms of implementation for the inventive concept and may not be considered limiting of the scope of the present invention to the specific forms set forth in the examples.
Claims (6)
1. A bridge damage assessment method based on bridge deck image recognition and bridge vibration perception is characterized by comprising the following steps:
s1, connecting a truck access registration information database of factories, construction sites or highway bayonets near bridge sites, and acquiring load information and license plate information of an access cargo truck to obtain a heavy-load truck registration database;
S2, setting power response monitoring sensing equipment on the bridge so as to acquire real-time power response data of the bridge structure under the action of environmental excitation;
s3, capturing the appearance and license plate information of a bridge deck running vehicle based on bridge deck monitoring equipment and an image recognition algorithm;
S4, aiming at the recognized truck vehicles in the step S3, searching load information of corresponding vehicles in a heavy truck registration database, and generating specific vehicle-load data; according to the specific vehicle passing time, the power response time sequence data caused when the vehicle passes through the bridge is obtained by utilizing the power response monitoring sensing equipment in the step S2, and the specific vehicle-load-power response data is finally generated;
s5, constructing a vehicle load recognition model based on bridge vibration sensing based on the deep neural network, taking the known load-corresponding power response data of the specific vehicle obtained in the step S4 as a training data set, inputting the training data set into the vehicle load recognition model for model training, and obtaining the vehicle load recognition model based on bridge vibration sensing after training and optimization;
S6, inputting structural power response data caused by the unknown vehicle into a vehicle load identification model, and outputting the predicted load of the random unknown vehicle;
S7, automatically identifying load information of vehicles randomly running through the bridge deck based on the vehicle load identification model, and comparing the load information with the known vehicle load in the heavy-duty truck registration database in the step S1 to judge whether the bridge is damaged.
2. The bridge damage assessment method based on bridge deck image recognition and bridge vibration perception as claimed in claim 1, wherein the dynamic response monitoring sensing device in the step S2 comprises bridge deck vehicle image acquisition equipment, a vibration acceleration sensor and a displacement sensor, the displacement sensor comprises at least one of a dynamic displacement optical sensor and a laser displacement sensor, and the bridge deck vehicle image acquisition equipment, the vibration acceleration sensor, the dynamic displacement optical sensor and the laser displacement sensor are respectively used for acquiring vehicle apparent identity information, bridge vibration data, bridge girder dynamic deflection data and bridge girder dynamic corner data of a randomly driven bridge deck vehicle.
3. The bridge damage assessment method based on bridge deck image recognition and bridge vibration perception as claimed in claim 1, wherein the vehicle load recognition model constructed in step S5 has an input and output formula expressed as:
Where a represents the vehicle load output, g (·) represents the activation function, w i represents the weight of the ith input signal, x i represents the ith input signal, the power response data obtained by the power response monitoring sensing device when the vehicle passes the bridge is taken as the input signal data, and b represents the bias.
4. A bridge damage assessment method based on bridge deck image recognition and bridge vibration perception according to claim 3, wherein x i represents bridge vibration data and one of bridge girder dynamic deflection and bridge girder dynamic rotation angle, i represents the input number of such signals.
5. A bridge damage assessment method based on bridge deck image recognition and bridge vibration perception as recited in claim 3, wherein step S5 further comprises a feature extraction and learning process of the model, comprising: the method comprises the steps of firstly decomposing EEMD based on an integrated empirical mode, decomposing dynamic response data obtained by dynamic response monitoring sensing equipment when a vehicle passes through a bridge into a plurality of modal components, removing components irrelevant to the vehicle load in the components, and obtaining residual modal components which are dynamic response monitoring data relevant to the vehicle load; then, based on the fully connected neural network FCN, establishing a mapping relation between dynamic response monitoring data related to the vehicle load and the actual vehicle load, and forming an identification model of the vehicle load; as the monitoring data is continuously accumulated, the self-repairing and optimizing of the model are realized.
6. The bridge damage assessment method based on bridge deck image recognition and bridge vibration perception according to claim 1, wherein in step S7, when the error between the predicted vehicle load data and the known vehicle load in the heavy truck registration database is within 5%, the bridge health is judged, otherwise, the bridge damage is indicated.
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118260647A true CN118260647A (en) | 2024-06-28 |
Family
ID=
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111667688B (en) | Comprehensive traffic information analysis method and system based on pavement vibration signals | |
| Mei et al. | Towards smart cities: crowdsensing-based monitoring of transportation infrastructure using in-traffic vehicles | |
| Hou et al. | Cyber-physical system architecture for automating the mapping of truck loads to bridge behavior using computer vision in connected highway corridors | |
| CN113031562B (en) | Hidden danger risk early warning method for single-column pier bridge passing freight vehicle | |
| CN110702195B (en) | Method and device for monitoring running state of load-carrying vehicle | |
| CN114005278A (en) | Intelligent monitoring and early warning system and method for highway infrastructure group | |
| CN116881819B (en) | Stay cable working state monitoring method based on isolated forest | |
| Yu et al. | Short-term traffic flow prediction based on deep learning network | |
| CN113033976A (en) | Reliable filtering design method of urban road system based on event trigger mechanism | |
| CN106327630A (en) | High-and-low-speed no-parking overload control detection system | |
| Jin et al. | Predictive Analytics of In‐Service Bridge Structural Performance from SHM Data Mining Perspective: A Case Study | |
| Getachew | Traffic load effects on bridges, statistical analysis of collected and Monte Carlo simulated vehicle data | |
| CN116289444A (en) | Intelligent pavement disease real-time monitoring and predicting device and method | |
| CN115326178A (en) | Method and system for actively warning overload of bridge | |
| CN117436821B (en) | Method, device and storage medium for generating traffic accident diagnosis report | |
| CN112781888B (en) | System and method for testing a vehicle | |
| CN118260647A (en) | Bridge damage assessment method based on bridge deck image recognition and bridge vibration perception | |
| Arellano et al. | A model of truck speed profiles on short upward slopes | |
| CN110853164B (en) | Road network damage-based traffic charging method, system, medium and charging equipment | |
| Zhou et al. | A hybrid virtual–real traffic simulation approach to reproducing the spatiotemporal distribution of bridge loads | |
| CN110909405A (en) | Bridge structure optimization method and system based on vehicle load and intelligent equipment | |
| CN115563020B (en) | Method and system for generating danger test scene, electronic device and storage medium | |
| CN114333331B (en) | Method and system for identifying vehicle passing information and vehicle weight of multi-lane bridge | |
| DE112020006572T5 (en) | Movement assistance device, movement assistance learning device and movement assistance method | |
| Kazinski et al. | Contribution Determination of the Statistical Loading of the Crossroads by Means of the YOLO5 Neural Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication |