Movable sliding block support device of intelligent detection system for appearance quality of concrete T-beam

Abstract

The invention relates to a movable sliding block support device of an intelligent detection system for appearance quality of a concrete T-beam. A sliding block slides up and down along a stand column to adjust the height of a camera, an extension rod horizontally extends out of a horizontal support and is used for adjusting the extension length of the camera, and the end of the extension rod is rotationally connected with a camera support to adjust the pitch angle of the camera; the sliding block, the extension rod and the camera base are respectively provided with a corresponding motor, and are controlled by a PLC control system, so that multi-dimensional movement and accurate control of the detection equipment in the vertical direction, the front-back direction and the angle are realized, and the intelligent detection system for the appearance quality of the precast beam can efficiently and accurately complete a detection task; a preferential experience playback and reward function algorithm is introduced to improve an SAC algorithm to serve as an intelligent obstacle avoidance algorithm, and telescopic movement of the camera is achieved through a movable sliding block support device.

Description

Technical Field

[0001] The invention belongs to the field of prefabricated beam appearance quality detection, and in particular relates to a mobile slider bracket device of a concrete T-beam appearance quality intelligent detection system. Background Art

[0002] The intelligent detection system for appearance quality defects of precast beams is designed to automatically detect the surface of precast beams by using computer vision, machine learning and deep learning technologies, so as to identify and evaluate various appearance defects such as cracks, bubbles, etc. The system consists of several key parts: image acquisition module, image preprocessing module, defect detection module and result display module.

[0003] The image acquisition module is implemented by mounting an industrial camera on the gantry to automatically capture and transmit images. However, in the prior art, due to the different sizes of beams, the safe distance between the camera and the beam is difficult to control. In addition, when there are fewer camera devices, it is difficult to effectively cover the entire appearance of the beam. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a mobile slider bracket device for an intelligent detection system for the appearance quality of a concrete T-beam.

[0005] The movable slider support device of the concrete T-beam appearance quality intelligent detection system is installed on the columns on both sides of the door frame. The movable slider support device includes: a slider, a horizontal support and an extension rod. The slider slides up and down along the column, one end of the horizontal support is fixed on the slider, and the extension rod is connected to a telescopic motor. The telescopic motor is used to control the extension rod to slide and extend from the end of the horizontal support away from the slider, and the end of the extension rod is rotatably connected to a camera support.

[0006] The camera is installed on a camera support, which is connected to a rotating motor. The rotating motor is used to control the rotation of the camera support to drive the pitch angle of the camera to change; the telescopic motor and the rotating motor are both connected to a PLC control system.

[0007] The method for intelligently detecting the appearance quality of a concrete T-beam using the movable slider support device comprises the following steps:

[0008] Step 1: The camera takes a picture of the T-beam appearance, calculates the distance between the camera and the T-beam, and calibrates the camera;

[0009] Step 2: Introduce the priority experience playback and reward function algorithm to improve the SAC algorithm as an intelligent obstacle avoidance algorithm; the intelligent obstacle avoidance algorithm controls the motor to adjust the extension length of the extension rod until the distance from the camera to the T-beam meets the requirements;

[0010] Step 3: Rotate the motor to control the camera support to rotate, adjust the camera pitch angle, and complete the inspection preparation;

[0011] Step 4: The gantry drives the movable slider bracket device to move along the length direction of the T-beam, and the camera takes an image of the appearance of the T-beam for defect detection.

[0012] Preferably, in step one, an optimization dispersal mechanism is introduced into the camera calibration internal parameter optimization method based on the particle swarm algorithm to perform calibration between the camera and the T-beam.

[0013] Preferably, in step 2, the priority experience replay randomly extracts a set of distances to obstacles in the environment, distances to target points, and angles as training samples during training. Before training, a priority is added to each sample. The priority of the sample is defined by the TD error. The larger the TD error, the higher the priority. The probability of the sample being sampled is regulated by the priority.

[0014] As a preferred method, in step 2, a continuous reward and penalty function is introduced. If the distance d from the camera to the target point at the current moment is t Less than threshold C d , then the camera is judged to have reached the target point, and a positive reward is given; if the minimum distance min for the camera to work x If the distance is less than the safety distance C0, a negative reward is given; the reward size is determined by (d t-1 -d t )×reward parameter, where (d t-1 -d t ) are assigned different reward parameters to positive and negative values.

[0015] Preferably, in step three, a tracking algorithm is used to calculate the center coordinates M(x, y) of the T-beam area to be photographed, and the deviation and direction between M(x, y) and the preset position of the camera support are calculated to control the rotation of the camera support.

[0016] Preferably, in step four, the gantry moves during camera shooting, and the slider slides up and down along the column so that the shooting range covers the surface of the T-beam; the up and down sliding speed of the slider and the camera's shooting frequency are set according to the size of the T-beam and the moving speed of the gantry.

[0017] The beneficial effects of the present invention are:

[0018] 1) The movable slider bracket device of the slider of the present invention slides up and down along the column to adjust the height of the camera. An extension rod is horizontally extended from the horizontal support to adjust the extension length of the camera. The end of the extension rod is rotatably connected to the camera support to adjust the pitch angle of the camera; and the slider, the extension rod and the camera base are respectively provided with corresponding motors, which are controlled by the PLC control system to realize multi-dimensional movement and precise control of the up and down, front and back and angle of the detection equipment, so that the prefabricated beam appearance quality intelligent detection system can efficiently and accurately complete the detection task.

[0019] 2) The PLC control system of the present invention introduces priority experience playback and reward function algorithm to improve the SAC algorithm as an intelligent obstacle avoidance algorithm, and realizes the telescopic movement of the camera through a movable slider bracket device, and automatically maintains a safe distance between the camera and the beam in the preparation stage before the detection begins. It can be suitable for intelligent detection of the appearance quality of beams of various sizes, has wide adaptability and scalability, and can meet the needs of different application scenarios.

[0020] 3) The present invention proposes a camera calibration internal parameter optimization method, which avoids falling into the local optimal solution by introducing an optimization dispersal mechanism into the camera calibration internal parameter optimization method based on the particle swarm algorithm, enhances the global search capability, and thus more effectively improves the calibration accuracy.

[0021] 4) The present invention designs a slider to slide up and down along the column, and sets the slider's up and down sliding speed and the camera's shooting frequency according to the size of the T-beam and the moving speed of the gantry, ensuring that the camera can cover and shoot all surfaces of the prefabricated beam to achieve comprehensive detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flow chart of an intelligent concrete T-beam appearance quality detection system of the present invention;

[0023] Figure 2 This is a diagram of the slider bracket structure;

[0024] Figure 3 It is a structural schematic diagram of the door frame;

[0025] Figure 4 This is a schematic diagram of the installation of the slider bracket on the door frame;

[0026] Figure 5 This is a schematic diagram of T-beam appearance quality inspection.

[0027] Explanation of the reference numerals: slide block 1, horizontal support 2, diagonal support 3, extension rod 4, camera support 5, camera 6, door frame 7, slide groove 8, guide groove 21. DETAILED DESCRIPTION

[0028] The present invention is further described below in conjunction with embodiments. The description of the following embodiments is only used to help understand the present invention. It should be noted that for ordinary persons in the art, without departing from the principle of the present invention, the present invention can also be modified in some ways, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

[0029] Embodiment 1

[0030] As another example, Figures 2 to 5As shown, the movable slider support device of the concrete T-beam appearance quality intelligent detection system is installed on the columns on both sides of the door frame 7, and the movable slider support device includes: a slider 1, a horizontal support 2 and an extension rod 4.

[0031] Each slider 1 is independently connected to a lifting motor for controlling the movable slider bracket device to slide up and down along the column as a whole. Figure 3 and 4 As shown, a slide groove 8 is provided on the opposite surfaces of the columns on both sides of the door frame 7, and two controlled movable slider bracket devices are provided in each slide groove 8. Through the cooperation of the two controlled movable slider bracket devices on the same side, the entire surface of the T-beam can be covered within the shooting range of the camera 6.

[0032] like Figure 2 As shown, one end of the horizontal support 2 is fixed on the slider 1, and the other end extends horizontally to the inner side of the door frame 7. The horizontal support 2 provides additional support to prevent the slider bracket from tilting or shaking during movement, thereby ensuring the stable operation of the camera 6. In a vibrating environment, the horizontal support 2 can effectively reduce and absorb vibrations to maintain the stability of the detection device, i.e., the camera 6. In this embodiment, an oblique support 3 is also provided below the horizontal support 2 to form a triangular structure to enhance the overall stability of the slider bracket and prevent tilting or deformation when moving or carrying heavy objects. In an environment with vibration or external force impact, the oblique support 3 can effectively absorb and disperse the impact force to protect the stability of the slider bracket and the detection equipment.

[0033] A guide groove 21 is provided on the upper surface of the horizontal support 2, and one end of the extension rod 4 is inserted into the guide groove 21. The extension rod 4 is connected to a telescopic motor, which is used to control the extension rod 4 to slide out from the end of the horizontal support 2 away from the slider 1 to adjust the extension amount of the camera 6. The end of the extension rod 4 is rotatably connected to the camera support 5.

[0034] The camera 6 is mounted on the camera support 5. The interface between the camera base 5 and the industrial camera should be universal to meet the adaptation of most models of industrial cameras. The camera support 5 is connected to a rotating motor, which is used to control the rotation of the camera support 5 to drive the precise control of the pitch angle of the camera 6, so that the camera 6 can be accurately aimed at the target area to meet different detection requirements.

[0035] Both the telescopic motor and the rotating motor are connected to the PLC control system, which can realize multi-dimensional movement and precise control of the up and down, front and back and angle of the detection equipment, so that the intelligent detection system for the appearance quality of precast beams can complete the detection task efficiently and accurately.

[0036] Embodiment 2

[0037] As another embodiment, this embodiment 2 proposes, on the basis of embodiment 1, a more specific movable slider bracket device of a concrete T-beam appearance quality intelligent detection system.

[0038] The slider 1 includes components such as a slider body, a guide device, a lubrication system and a limit device. The slider body is a high-strength steel rectangular structure, and the surface is anodized to enhance wear resistance and corrosion resistance. A high-precision ball bearing is installed inside the guide device to ensure smooth movement of the slider 1 on the slide 8. The lubrication system is provided with a lubricating oil tank inside and is equipped with an automatic lubrication device to ensure smooth operation during long-term use. The limit device installs limit switches at both ends of the slider 1 to prevent the slider 1 from exceeding the preset travel range and ensure safe operation.

[0039] The horizontal support 2 includes a horizontal support body and a guide groove 21. The guide groove 21 is designed with two-piece guide groove side plates and a C-shaped groove base, which can be flexibly adjusted according to the width of the extension rod 4. The slide bar is coated with a lubricating coating to reduce friction when the extension rod 4 moves and improve the smoothness of the movement.

[0040] The diagonal brace 3 comprises a diagonal brace body and a connecting device. The connecting device is arranged at both ends of the diagonal brace 3 and is fixed to the slide block 1 and the horizontal brace 2 by bolts.

[0041] The extension rod 4 includes an extension rod body and an electric telescopic device. Figure 2 As shown, the end of the extension rod 4 is harpoon-shaped, leaving space for the camera base 5 to rotate. The electric telescopic device is equipped with a telescopic motor, which is automatically extended and retracted by a screw drive. The electric telescopic device is connected to a PLC control system, and the safety distance between the detection equipment and the beam body is automatically controlled by a deep learning intelligent obstacle avoidance algorithm.

[0042] The camera base 5 includes a base body and a rotating mechanism. The base body adopts a modular interface, which is convenient for quick replacement and upgrading, and is adapted to different cameras 6 and application requirements. The rotating mechanism has a built-in high-precision ball bearing and is equipped with a rotating motor as an electric rotating device to achieve 180-degree free rotation.

[0043] It should be noted that the parts in this embodiment that are the same or similar to those in the first embodiment can be referenced to each other and will not be described in detail in this application.

[0044] Embodiment 3

[0045] As another embodiment, this embodiment 3, based on the first and second embodiments, proposes a detection method using the mobile slider bracket device of the concrete T-beam appearance quality intelligent detection system, including the following steps:

[0046] Step 1: Camera 6 takes a picture of the appearance of the T-beam, and calculates the distance between the camera 6 and the T-beam through coordinate system conversion;

[0047] Step 2: Introduce priority experience replay and reward function algorithm to improve the SAC algorithm as an intelligent obstacle avoidance algorithm;

[0048] Priority experience replay randomly extracts a set of distances to obstacles, distances to target points, and angles in the environment as training samples during training. Before starting training, each sample is assigned a priority. Samples with high priority are more likely to be sampled during sampling, thereby increasing the training speed. The priority of the sample is defined by the TD error. The larger the TD error, the higher the priority. TD error δ t The calculation formula is:

[0049] δ t =r t +γQ(s t+1 ,a t+1 )-Q(s t ,a t )

[0050] Among them, s t is the position of the end of the extension rod 4 and the camera 6 at time t, a t is the extension or retraction action of the extension rod 4 at time t, Q is the current state value, γ is the weight coefficient, and r t is the return value of the current state.

[0051] The probability of sample extraction is defined as:

[0052]

[0053] Among them, a is used to adjust the priority; p i =|δ i |+ε is the priority of the i-th sample, δ i is the TD error of the i-th sample, and ε is added to avoid the probability being 0.

[0054] The reward function determines how good or bad an action is in a certain state. The sparse reward problem is solved by designing a continuous reward and penalty function. The reward and penalty function is as follows:

[0055]

[0056] Among them, r arr Represents the positive reward for reaching the target point; d t Indicates the distance from camera 6 to the target point at the current moment; d t-1 Indicates the distance from camera 6 to the target point at the last moment; C d Indicates the threshold value of reaching the target point. If the value is less than this value, it means that the target point has been reached and a positive reward is given; r coll Indicates a negative reward for hitting an obstacle; min xIndicates the minimum distance at which the camera 6 works; C0 indicates the safe distance for collision with obstacles. If the minimum distance at which the camera 6 works is lower than this value, it means that the camera 6 has touched the obstacle, i.e., the T-beam;

[0057] The specific value of the reward is based on the distance difference from the target to the last time step (d t-1 -d t ) is calculated by multiplying the reward parameter. r1 and C r2 There are two reward parameters set. If the difference is positive, the reward is equal to the distance traveled multiplied by the parameter C r1 , otherwise multiply by parameter C r2 This reward mechanism motivates the camera 6 to reach the target position through positive rewards, punishes collision behavior through negative rewards, and rewards or punishes movement behavior according to the change in distance, thereby guiding the extension rod 4 to drive the camera 6 to move effectively during work.

[0058] The intelligent obstacle avoidance algorithm controls the motor to adjust the extension length of the extension rod 4 until the distance from the camera 6 to the T-beam meets the requirement.

[0059] Step 3: The rotating motor controls the camera support 5 to rotate, adjusts the pitch angle of the camera 6, and completes the detection preparation;

[0060] The system uses the tracking algorithm to calculate the center coordinates M(x,y) of the area to be photographed on the beam, calculates the deviation and direction between this position and the preset position of the camera support, and automatically controls the rotation of the support through PLC serial communication. Assuming that α and β are half of the horizontal and vertical viewing angles of the camera, and r is the slant distance between the beam and the camera, by controlling the horizontal azimuth and vertical elevation angle of the camera support θ, so that the center M is always within the constraint area. The tracking algorithm can be expressed as:

[0061] x m = r tanα

[0062] y m = r tanβ

[0063] Among them, x m ,y m Related to the resolution of the image taken by the camera, such as the image resolution is 750*600, x m =750 / 2,y m =600 / 2. The center of the beam area to be photographed is:

[0064]

[0065] y=rtanθ

[0066] Horizontal azimuth of camera mount and the vertical elevation angle θ can be expressed as:

[0067]

[0068] θ=tan -1 (ytanβ / y m )

[0069] Step 4: The gantry 7 drives the movable slider bracket device to move along the length direction of the T-beam, and the camera 6 takes an image of the appearance of the T-beam for defect detection.

[0070] Embodiment 4

[0071] As another embodiment, this embodiment 4 proposes a more specific detection method based on the embodiment 3.

[0072] In step 1, the camera ranging algorithm is specifically as follows:

[0073] The image captured by camera 6 is calibrated using the image coordinate system, camera coordinate system and world coordinate system respectively. The pixel point of the center of the camera 6 lens in the image coordinate system and the actual point of the pixel point in the world coordinate system are used as the proportional medium. The distance value from camera 6 to the T-beam is solved through coordinate system conversion combined with camera 6 parameters.

[0074] Furthermore, the present invention improves the camera calibration internal parameter optimization method based on the particle swarm algorithm, expands the search space of the particle swarm by introducing an optimization dispersal mechanism, avoids falling into the local optimal solution, and enhances the global search capability, thereby more effectively improving the calibration accuracy. The basic particle swarm algorithm first randomly initializes a group of particles, and then updates the particles in each iteration cycle. In an n-dimensional search space, the current position of the i-th particle is X i =(X i1 ,X i2 ,X i3 ...X in ) T , its generalized velocity is V i =(V i1 ,V i2 ,V i3 ……V in ) T , then the particle update equation is expressed as:

[0075]

[0076] Among them, p best is the historical optimal position of the particle; g best is the optimal position of the entire particle swarm; k is the current number of iterations; w is the inertia weight, which is used to adjust the convergence direction; c1 and c2 are the self- and social learning rates, respectively, which are used to adjust the iteration step; e1 and e2 are independent random numbers in the interval (0, 1).

[0077] The basic particle swarm algorithm is improved, an optimized dispersal mechanism is introduced, and the average value of the difference between the two nearest adjacent iterations is used to measure the degree of convergence that changes with the number of iterations. Its expression is:

[0078]

[0079] Among them, s(k) is the convergence degree of iteration k; g best (ka) is the global optimal value after iteration (ka) times; z is a weight coefficient of 3 to 4.

[0080] By judging whether s(k) is less than a certain value ε, it can be determined whether the dispersal action needs to be triggered. After the dispersal action is triggered, the particle swarm is assigned to a new spatial range. As the number of iterations increases, the global search space range after each dispersal becomes smaller and smaller, so the degree of dispersal action also decreases accordingly. The degree of dispersal action can be expressed as:

[0081]

[0082] Where d represents the degree of dispersal action; k represents the number of iterations when the dispersal action is triggered. The new position relationship of the particles can be expressed as:

[0083] X i (k+1)=dX i (k)

[0084] Among them, X i (k) is the current position of the i-th particle; X i (k+1) is the new position of the particle after being dispersed.

[0085] In step 4, the gantry 7 moves while the camera 6 is shooting, and the slider 1 slides up and down along the column so that the shooting range covers the surface of the T-beam; the up and down sliding speed of the slider 1 and the shooting frequency of the camera 6 are set according to the size of the T-beam and the moving speed of the gantry 7.

[0086] It should be noted that the parts in this embodiment that are the same or similar to those in Embodiment 3 can be referenced to each other and will not be described in detail in this application.

[0087] Embodiment 5

[0088] As another embodiment, this embodiment 5 also proposes a method for installing and using the mobile slider bracket device of the concrete T-beam appearance quality intelligent detection system, including the following steps:

[0089] S1. Install the guide device to the slider 1 and add a proper amount of lubricating oil to the lubrication system.

[0090] S2. Install the slider 1 to the slide groove 8, and install fixed limit switches at both ends of the slider 1.

[0091] S3. Install the guide groove side plate to the inner side of the guide groove 21, install the C-shaped groove base to the bottom of the guide groove 21, fix the slide bar to the side plate of the guide groove 21, and apply a lubricating coating.

[0092] S4. Fix the front end of the horizontal support 4 to the slider 1 by bolts.

[0093] S5. Set the diagonal brace connection device at both ends of the diagonal brace 3 and fix it on the slider 1 and the horizontal brace 2 by bolts.

[0094] S6. Install the extension rod 4 body to the slide bar, install the telescopic motor to the front end of the guide groove 21 and connect to the extension rod 4 body, and connect the telescopic motor to the PLC control system.

[0095] S7, install the rotating mechanism of the camera base 5 to the end of the extension rod 4, connect the camera base 5 to the rotating motor, connect the rotating motor to the PLC control system, and install the camera 6 to the camera base 5.

[0096] S9, the intelligent obstacle avoidance algorithm identifies the safe distance of the detection equipment; the system triggers the PLC control system and the electric telescopic device to position the extension rod 4 to the working position.

[0097] S10, the system triggers the PLC control system to adjust the pitch angle of the camera base 5.

[0098] S11, the system calculates and sets the moving speed of the slider 1 and the shooting frequency of the camera 6, the gantry 7 translates, the camera 6 takes pictures of the prefabricated beam body and transmits them back, and detects and identifies the appearance quality defects of the beam body.

[0099] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

Claims

1. A mobile slider support device for an intelligent detection system for the appearance quality of concrete T-beams. It is characterized in that Installed on the columns on both sides of the door frame, the movable slider bracket device includes: a slider, a horizontal support and an extension rod. The slider slides up and down along the column, one end of the horizontal support is fixed on the slider, and the extension rod is connected to a telescopic motor. The telescopic motor is used to control the extension rod to slide and extend from the end of the horizontal support away from the slider, and the end of the extension rod is rotatably connected to a camera support; The camera is installed on a camera support, which is connected to a rotating motor. The rotating motor is used to control the rotation of the camera support to drive the pitch angle of the camera to change; the telescopic motor and the rotating motor are both connected to a PLC control system.

2. An intelligent detection method for the appearance quality of a concrete T-beam using the movable slider support device described in claim 1, It is characterized in that The following steps are involved: Step 1: The camera takes a picture of the appearance of the T-beam, calculates the distance between the camera and the T-beam, and calibrates the camera; Step 2: Introduce the priority experience playback and reward function algorithm to improve the SAC algorithm as an intelligent obstacle avoidance algorithm; the intelligent obstacle avoidance algorithm controls the motor to adjust the extension length of the extension rod until the distance from the camera to the T-beam meets the requirements; Step 3: Rotate the motor to control the camera support to rotate, adjust the camera pitch angle, and complete the inspection preparation; Step 4: The gantry drives the movable slider bracket device to move along the length direction of the T-beam, and the camera takes an image of the appearance of the T-beam for defect detection.

3. The intelligent detection method for appearance quality of concrete T-beams according to claim 2, It is characterized in that In step 1, an optimization dispersal mechanism is introduced into the camera calibration internal parameter optimization method based on particle swarm algorithm to calibrate the camera and T-beam.

4. The intelligent detection method for appearance quality of concrete T-beams according to claim 2, It is characterized in that In step 2, the priority experience replay randomly extracts a set of distances to obstacles in the environment, distances to target points, and angles as training samples during training. Before training, a priority is added to each sample. The priority of the sample is defined by the TD error. The larger the TD error, the higher the priority. The sampling probability of the sample is regulated by the priority.

5. The intelligent detection method for appearance quality of concrete T-beams according to claim 2, It is characterized in that In step 2, a continuous reward and penalty function is introduced. If the distance d from the camera to the target point at the current moment is t Less than threshold C d , then the camera is judged to have reached the target point, and a positive reward is given; if the minimum distance min for the camera to work x Less than the safety distance C 0 , a negative reward is given; the reward size is determined by (d t-1 -d t )×reward parameter, where (d t-1 -d t ) are assigned different reward parameters to positive and negative values.

6. The intelligent detection method for appearance quality of concrete T-beams according to claim 2, It is characterized in that In step three, the tracking algorithm is used to calculate the center coordinates M(x, y) of the T-beam area to be photographed, the deviation and direction between M(x, y) and the preset position of the camera support are calculated, and the rotation of the camera support is controlled.

7. The intelligent detection method for appearance quality of concrete T-beams according to claim 2, It is characterized in that In step 4, the gantry moves during camera shooting, and the slider slides up and down along the column so that the shooting range covers the surface of the T-beam; the up and down sliding speed of the slider and the camera shooting frequency are set according to the size of the T-beam and the moving speed of the gantry.

Images