CN111986238B - Concrete arch dam modal shape identification method based on unmanned aerial vehicle video shooting - Google Patents

Abstract

The invention discloses a method for identifying the mode shape of a concrete arch dam based on video shooting of an unmanned aerial vehicle. ; Obtain the video amplified by the first four-order motion of the original dam vibration video according to the natural frequencies of the first four orders of the arch dam; The change of the center of mass coordinates before and after the vibration of each marked point is obtained, and the maximum value of the relative coordinate change of each marked point is normalized, which is the first fourth-order mode shape of the arch dam crest. Through the non-contact method, the present invention can realize the approximate full-field measurement of the vibration shape of the arch dam crest, with high spatial resolution, and can provide support for subsequent dam model update and damage identification.

Description

A method for identifying mode shapes of concrete arch dams based on UAV video shooting

technical field

The invention belongs to the technical field of dynamic characteristic analysis of concrete arch dams, finite element model updating and damage identification, and particularly relates to a modal vibration mode identification method of concrete arch dams based on video shooting of unmanned aerial vehicles.

Background technique

As an important dam type in water conservancy and hydropower projects, arch dams have made great contributions to the development of the national economy and society. However, during their service period, due to extreme environmental loads (such as earthquakes, floods, etc.), construction defects, poor management and maintenance, and aging after long-term use, many dams have various diseases and hidden dangers, which may lead to structural instability. Or strength damage, which seriously affects the normal operation and benefit of the project. Therefore, regular or real-time health monitoring of arch dams is of great significance to ensure their safe operation.

The dynamic parameters (frequency, damping and mode shape) of the structure can be identified through the dynamic test, so as to realize the health monitoring and damage identification of the dam. At present, there are mainly contact and non-contact methods for dynamic testing of dam structures. In the contact method, a large number of sensors (such as accelerometers) need to be fixed to the dam structure for vibration measurement. While these sensors are highly reliable, their installation is time-consuming and labor-intensive. In addition, these sensors only provide sparse discrete point measurements and can only obtain low spatial resolution, which is far from sufficient for application scenarios such as model updating and damage identification. Compared with the traditional contact vibration measurement method, the non-contact camera measurement technology does not require additional sensors, which is more convenient and efficient, and at the same time has higher spatial resolution, and can achieve approximately full field vibration measurement.

Invention content:

In order to overcome the above-mentioned defects of the background technology, the present invention provides a method for identifying the mode shape of a concrete arch dam based on video shooting of an unmanned aerial vehicle. In order to solve the problems of low rate and other problems, it is proposed to realize the approximate full-field vibration measurement of the dam body through the non-contact camera measurement technology. This technology does not require additional sensors, which is more convenient and efficient, and has higher spatial resolution.

In order to solve the above-mentioned technical problems, the adopted technical scheme of the present invention is:

A method for identifying modal mode shapes of a concrete arch dam based on video shooting by drones, comprising:

Step 1: Lay markers on the crest of the dam;

Step 2, obtain the dam crest image through the pan-tilt camera on the drone, and obtain the dam crest vibration video;

Step 3, intercepting the dam crest vibration video with a preset duration, selecting several sub-regions along the marked points of the dam crest, extracting the local motion phase information of each sub-region, and then obtaining its local motion;

Step 4: Perform singular value decomposition on the power spectral density of the motion signal of each sub-region to obtain the first fourth-order natural frequency of the arch dam;

Step 5, obtaining the amplified video of the first four-order motion of the original dam vibration video according to the first four-order natural frequency of the arch dam;

Step 6: Select the maximum deformation frame of the dam crest in the magnified video of the original dam's front fourth-order motion, identify the coordinates of the centroid of each marked point in the image, obtain the change of the centroid coordinates before and after the vibration of each marked point, and perform a calculation on the maximum value of the relative coordinate change of each marked point. Normalized processing is the first four modal mode shapes of the arch dam crest.

Preferably, the marking points arranged on the top of the dam are evenly spaced, and the spacing between the marking points is at least meters.

Preferably, the number of sub-regions is 9, and the size of each sub-region is 20×20 pixels.

Preferably, the local motion phase information of each sub-region extracted in step 3 is obtained by a steerable pyramid algorithm.

Preferably, the natural frequencies of the first four orders of the arch dam in step 4 are: the first four peaks of the first order singular values after the power spectral density of the motion signal of each sub-region is decomposed into singular values.

Preferably, in step 5, the method for obtaining the amplified video of the first four-order motion of the original dam vibration video according to the first four-order natural frequency of the arch dam includes:

The first four-order natural frequencies of the arch dam are taken as the center frequencies of the four band-pass filters, the frequency bandwidth is set to be 0.5Hz, and the four amplification factors are set to 10, 25, 50 and 100 in turn. Through the steerable pyramid algorithm The video signal is decomposed into local spatial amplitude and phase signals, and then the phase signals are filtered, amplified and reconstructed in turn to obtain the amplified video of the first four-order motion of the original dam vibration video.

The beneficial effects of the present invention are: the present invention can realize the approximate full-field measurement of the vibration shape of the arch dam crest through the non-contact method, and its spatial resolution is high, which can provide support for subsequent dam model update and damage identification. The identification process of the invention is fast and convenient, no contact acceleration or speed sensor is required, and the modal mode shape can be obtained only through the captured dam crest vibration video, thereby saving a lot of manpower, material and financial resources.

Description of drawings

1 is a flow chart of a method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of the vibration process of the arch dam crest photographed by a drone according to an embodiment of the present invention.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and embodiments.

The invention discloses a modal mode shape identification technology of a concrete arch dam based on unmanned aerial vehicle video shooting. Dam natural vibration frequency identification, video motion amplification processing, dam mode shape identification. The working principle of the present invention is as follows: the arbitrary motion process of the arch dam structure can be approximated as the linear superposition of the main mode shapes of the first several orders, and the phase-based video amplification technology is used to convert the specific frequency band (including a certain order natural frequency) The motion is amplified to highlight the contribution of the corresponding order mode shape, so as to realize the acquisition of the modal mode shape of the arch dam.

The dam crest marking point 3 adopts a red waterproof sticker with a radius of 0.3m, so as to ensure that the marking point in the dam crest vibration video shot has sufficient clarity.

The integrated pan-tilt camera 2 pre-installed on the drone 1 is a 10-bit color depth 4K resolution camera with a sampling frequency of 60Hz, so as to ensure the accurate capture of the tiny vibration of the dam crest.

The video of the vibration of the 4 dam crest of the concrete arch dam is only cut for 20s to reduce the subsequent video processing time.

In the natural frequency identification step of the arch dam, a series of sub-regions need to be selected on average along the dam crest marking points, and the local motion phase information of each sub-region is extracted through the complex domain filter, so as to estimate its local motion, and finally identified by the frequency domain decomposition method. Arch Dam Natural Frequency.

During video motion amplification processing, the amplification frequency band bandwidth is 0.5Hz, and the amplification factor is set between 10 and 100.

The phase-based video motion amplification technology mainly uses the complex domain steerable pyramid filter bank to decompose the video signal into local spatial amplitude and phase, and then filters, amplifies and reconstructs the phase signal to obtain the motion amplification video. The UAV control platform and UAV flight line are shown in Figure 2.

A method for identifying modal mode shapes of a concrete arch dam based on video shooting by drones, comprising:

Step 1: Arrange marking points on the dam crest; the marking points arranged on the dam crest are arranged evenly at intervals, and the interval between each marking point is at least 3 meters.

Step 2, obtain the dam crest image through the pan-tilt camera on the drone, and obtain the dam crest vibration video;

Step 3, intercept the dam crest vibration video of the preset duration, select several sub-areas along the marked points of the dam crest, extract the local motion phase information of each sub-area, and then obtain the local motion of each sub-area; the number of sub-areas is 9, The size of each sub-region is 20×20 pixels.

The local motion phase information of each sub-region is extracted by the steerable pyramid algorithm.

Step 4: Perform singular value decomposition on the power spectral density of the motion signal of each sub-region to obtain the first-order natural frequency of the arch dam; the first-order natural frequency of the arch dam is: the power spectral density of each sub-region motion signal is subjected to singular value decomposition , the first four peaks of the first-order singular values.

Step 5, obtaining the amplified video of the first four-order motion of the original dam vibration video according to the first four-order natural frequency of the arch dam;

The method of obtaining the amplified video of the original dam vibration video of the first four-order motion according to the first four-order natural frequency of the arch dam includes:

The first four-order natural frequencies of the arch dam are taken as the center frequencies of the four band-pass filters, the frequency bandwidth is set to be 0.5Hz, and the four amplification factors are set to 10, 25, 50 and 100 in turn. Through the steerable pyramid algorithm The video signal is decomposed into local spatial amplitude and phase signals, and then the phase signals are filtered, amplified and reconstructed in turn to obtain the amplified video of the first four-order motion of the original dam vibration video.

Step 6: Select the maximum deformation frame of the dam crest in the magnified video of the original dam's front fourth-order motion, identify the coordinates of the centroid of each marked point in the image, obtain the change of the centroid coordinates before and after the vibration of each marked point, and perform a calculation on the maximum value of the relative coordinate change of each marked point. Normalized processing is the first four modal mode shapes of the arch dam crest.

The technical solutions of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

The maximum dam height of concrete arch dam 4 is 66.8m, the dam crest is 179m long, 6m wide, and the dam bottom thickness is 21.4m. The maximum central angle is 90.383° and the minimum central angle is 6.059°. The valley of the dam site is "V"-shaped, and the dam body is roughly arranged symmetrically. In order to identify its modal shape, a DJI Inspire 1 drone1 was used to photograph the vibration process of the dam:

Step 1: Layout of dam crest marking points

In this example, 66 red circular waterproof stickers are set as marking points with a radius of 0.3 meters, which are evenly pasted on the crest of the arch dam along the axis of the dam, and the spacing between each marking point is 3 meters.

Step 2: Drone Flight Control

Install the propeller blade, power battery and integrated gimbal camera on the drone body. The camera is a 10-bit color depth 4K resolution camera with a sampling frequency of 60Hz. Plan the flight route of the drone, and the drone will be controlled by strictly trained technicians to fly directly above the crest of the arch dam and hover.

Step 3: Video shooting of dam crest vibration

Adjust the camera angle so that it is vertically aligned with the dam crest plane, and then adjust the drone height and camera focal length to make the entire dam crest area into the camera's field of view, and shoot the dam crest while ensuring the clarity of the dam crest marking points. Vibration video under discharge load.

Step 4: Video Motion Estimation

The dam crest vibration video with a duration of 20 s was intercepted and imported into the computer for processing. Nine sub-areas were selected on average along the dam crest marking points, and the sub-area size was 20 × 20 pixels. This estimates its local motion.

Step 5: Dam Natural Frequency Identification

The power spectral density of the motion signal in each sub-region is decomposed into singular value, and the first four peak points of the first-order singular value are the first-order natural frequencies of the arch dam.

Step 6: Video Motion Zoom Processing

The first four-order natural frequencies of the dam are taken as the center frequencies of the four band-pass filters, the frequency bandwidths are selected as 0.5Hz, and the amplification factors are selected as 10, 25, 50, and 100, respectively. The video signal is decomposed into local spatial amplitude and phase by the steerable pyramid algorithm, and then the phase signal is filtered, amplified and reconstructed, so as to obtain the amplified video of the first four-order motion of the original dam vibration video.

Step 7: Dam Mode Shape Identification

Select the frame with the largest deformation of the dam crest in the magnified video of each order, identify the coordinates of the centroid of each marked point in the image through image processing technology, calculate the change of the coordinates of the centroid before and after the vibration of each marked point, and normalize the maximum value of the relative coordinate change. The first four modal modes of the arch dam crest are obtained.

It should be understood that, for those skilled in the art, improvements or changes can be made according to the above description, and all these improvements and changes should fall within the protection scope of the appended claims of the present invention.

Claims (5)

1. The utility model provides a concrete arch dam modal shape recognition method based on unmanned aerial vehicle video shooting which characterized in that includes:

step 1, arranging marking points on the top of a dam;

step 2, acquiring a dam crest image through a cloud deck camera on the unmanned aerial vehicle, and acquiring a dam crest vibration video;

step 3, intercepting the dam crest vibration video with preset duration, selecting a plurality of sub-regions along the mark points of the dam crest, and extracting local motion phase information of each sub-region to further obtain local motion of each sub-region;

step 4, performing singular value decomposition on the power spectral density of the motion signals of each sub-area to obtain four-order natural frequency in front of the arch dam;

step 5, acquiring a video obtained by amplifying the front four-order motion of the original dam vibration video according to the front four-order natural frequency of the arch dam;

the method for acquiring the video amplified by the front four-order motion of the original dam vibration video according to the front four-order natural frequency of the arch dam in the step 5 comprises the following steps: respectively taking the front four-order natural frequency of the arch dam as the central frequency of four band-pass filters, setting the frequency bandwidth to be 0.5Hz, sequentially setting four amplification factors to be 10, 25, 50 and 100, decomposing a video signal into a local spatial amplitude and a phase signal through a steerable pyramid algorithm, and then sequentially filtering, amplifying and reconstructing the phase signal to obtain the video amplified by the front four-order motion of the original dam vibration video;

and 6, selecting a maximum frame of dam crest deformation in the original four-order motion amplification video in front of the dam, acquiring the barycentric coordinates of each mark point in the image and the variation of the barycentric coordinates of each mark point before and after vibration, and normalizing the maximum value of the variation of the relative coordinates of each mark point, namely the four-order modal shape of the front of the dam top of the arch dam.

2. The method for identifying the modal shape of the concrete arch dam based on unmanned aerial vehicle video shooting is characterized by comprising the following steps: the mark points distributed on the top of the dam are uniformly distributed at intervals, and the interval between every two mark points is at least 3 meters.

3. The method for identifying the modal shape of the concrete arch dam based on unmanned aerial vehicle video shooting is characterized by comprising the following steps: the number of the sub-regions is 9, and the size of each sub-region is 20 × 20 pixels.

4. The method for identifying the modal shape of the concrete arch dam based on unmanned aerial vehicle video shooting is characterized by comprising the following steps: the step 3 of extracting the local motion phase information of each sub-region is obtained by a steerable pyramid algorithm.

5. The method for identifying the modal shape of the concrete arch dam based on the video shooting of the unmanned aerial vehicle according to claim 1, wherein the four-order natural frequency before the arch dam in the step 4 is as follows: and after the power spectral density of the motion signal of each subregion is subjected to singular value decomposition, the first four peaks of the singular value of the first order are obtained.

Images