Precision detection device and dynamic error compensation method

Abstract

The invention relates to the technical field of precision detection, in particular to a precision detection device and a dynamic error compensation method.The device comprises a conveying mechanism, a turnover clamping mechanism, a jacking mechanism, a detection mechanism and a manipulator, and the conveying mechanism is used for conveying a to-be-detected product and a detected product; the overturning clamping mechanism clamps the product after the manipulator places the product to be detected on the jacking mechanism and drives the product to overturn around the central axis of the product in the detection process, so that the detection of the front and back surfaces is realized; the jacking mechanism is located below the overturning and clamping mechanism and used for providing bottom supporting when the product is clamped and descending to give way during overturning so as to keep the posture stable. The detection mechanism is located above the turnover clamping mechanism and can move in a reciprocating mode and conduct non-contact detection above the product; the manipulator is responsible for taking, placing and sorting, so that double-sided automatic detection of the to-be-detected product in the same station is realized, and the detection precision, the detection efficiency and the consistency of front and back detection results are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of precision testing technology, specifically to a precision testing device and a dynamic error compensation method. Background Technology

[0002] With the rapid development of precision manufacturing, the quality requirements for product geometry and form and position tolerances are constantly increasing. In particular, the inspection of precision parameters such as product position, perpendicularity, and surface morphology has become an indispensable key process in production. Currently, precision inspection mainly relies on manual handling or semi-automatic inspection devices. During operation, the product needs to be fixed on a fixture, and then the inspection probe or vision equipment is used to complete the single-sided inspection item by item. For products with complex structures or high assembly precision requirements, inspecting only one side of the surface cannot fully reflect its true quality status. Therefore, it is necessary to achieve continuous and stable inspection of both sides of the product in the same inspection process.

[0003] However, due to limitations in product structure and testing processes, when testing two or more surfaces, existing technologies generally employ manual flipping or additional independent mechanical devices for posture adjustment. Such operations are not only cumbersome and time-consuming, significantly reducing testing efficiency; more importantly, the product is highly susceptible to positional shifts or spatial instability during the flipping process, leading to deviations between the testing points and theoretical coordinates. This, in turn, affects the repeatability and reliability of measurement data, making it difficult to meet the stringent consistency requirements of high-precision manufacturing.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a precision detection device and a dynamic error compensation method, thereby effectively solving the problems pointed out in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A precision detection device includes: a conveying mechanism, a flipping clamping mechanism, a lifting mechanism, a detection mechanism, and a robotic arm; The conveying mechanism is located on both sides of the flipping clamping mechanism and is used to convey the product to be tested and the product after testing. The flipping clamping mechanism is used to clamp the product after the robot arm places the product to be tested on the lifting mechanism, and to drive the product to flip along its own central axis during the detection process so that the front and back sides of the product can be detected in sequence. The lifting mechanism is located below the flipping clamping mechanism and is used to support the bottom of the product during the product being clamped and to lower itself when the product is flipped, so as to ensure the stability of the product's spatial posture during the non-flipping phase. The detection mechanism is configured to reciprocate above the flipping clamping mechanism and perform non-contact detection on the product when it moves directly above the product. The robotic arm is used to place the product to be tested into the clamping area of ​​the flipping clamping mechanism and to classify the product according to the test results.

[0007] Furthermore, the conveying mechanism includes a pull plate mechanism on one side and a conveyor belt mechanism on the other side; The tray-pulling mechanism is used to carry a tray containing multiple products to be tested, and pulls out the tray to move the products to be tested in the tray to the gripping position of the robot arm in sequence. The tray pulling mechanism can also be coordinated with the lifting mechanism to allow the tray to move up and down between different layers, thereby realizing the layer-by-layer feeding of multi-layer products; The conveyor belt mechanism includes a first conveyor belt and a second conveyor belt disposed on both sides of the device. The first conveyor belt corresponds to products that pass inspection, and the second conveyor belt corresponds to products that fail inspection.

[0008] Furthermore, the lifting mechanism includes a lifting plate, a support base, a guide rod, a lifting rod, and a driving component; The driving component is mounted on the support base, and the output end of the support base is connected to the lifting rod. The lifting rod can be movably passed through the support base; The lifting plate is disposed at the top of the lifting rod; At least two of the guide rods are disposed below the lifting plate, extend parallel to the lifting rod, and are movably inserted through the support base; The driving component drives the lifting rod to move along its axial direction, thereby causing the lifting plate to move toward or away from the bottom of the clamped product.

[0009] Furthermore, the flipping clamping mechanism includes a clamping arm, a clamping drive component, and a flipping drive component; The two clamping arms are arranged opposite each other for clamping the product from both sides; The clamping drive component is connected to the clamping arm and is used to drive the two clamping arms to move closer to each other to clamp the product or to move away from each other to release the product. The flipping drive component is connected to the clamping drive component and is used to drive the clamping arm and the clamped product to rotate around a horizontal axis to achieve product flipping.

[0010] The present invention also includes a dynamic error compensation method for the accuracy detection device as described above, the steps of which include: The robot arm is controlled to transfer the product to be tested to the flipping clamping mechanism, and the flipping clamping mechanism is controlled to clamp the product. At the same time, the lifting mechanism is controlled to rise to support the product from the bottom. The detection mechanism is controlled to perform a first detection on one surface of the product; The lifting mechanism is controlled to descend and make way, then the flipping clamping mechanism is controlled to drive the product to flip, and after the flipping is completed, the lifting mechanism is controlled to rise again to support the product; The detection mechanism is controlled to perform a second detection on the other opposite surface of the product; During the detection process of at least one surface, dynamic error compensation is performed; the dynamic error compensation is based on a pre-generated digital model of the system error of the detection device, combined with real-time acquired device status data, to calculate the spatial pose deviation under the current detection state, and dynamically corrects the measurement coordinate system of the detection mechanism according to the spatial pose deviation. Based on the error-compensated test data, the product is determined to be qualified, and the robotic arm is controlled to move the product to the corresponding sorting position.

[0011] Furthermore, performing the dynamic error compensation specifically includes: The real-time flipping angle of the flipping clamping mechanism and the real-time support height of the lifting mechanism are obtained in real time. Based on the real-time flip angle and real-time support height, the corresponding theoretical pose error is obtained by querying the system error digital model. The real-time monitoring data of the auxiliary sensor is acquired to correct the theoretical pose error and generate a comprehensive pose error compensation amount. The measurement coordinate system of the detection mechanism is dynamically corrected based on the comprehensive pose error compensation amount.

[0012] Furthermore, a Kalman filter algorithm is used to fuse the real-time monitoring data of the auxiliary sensor to correct the theoretical pose error; the process model of the Kalman filter is constructed based on the gradient information of the digital model of the system error, and the observation model is constructed based on the measurement data of the auxiliary sensor. The comprehensive pose error compensation amount is optimally estimated through iterative calculation.

[0013] Furthermore, in the process model of the Kalman filter, dynamic state variables characterizing the motion speed and acceleration of the flipping clamping mechanism and the lifting mechanism are introduced to construct a time-varying state transition equation, so that the estimated comprehensive pose error compensation amount can predict and compensate for the dynamic hysteresis error caused by the acceleration and deceleration motion of the mechanism.

[0014] Furthermore, the system error digital model is pre-established through the following steps: The flipping clamping mechanism and the lifting mechanism are driven to operate at multiple discrete linkage pose points; At each of the aforementioned linked pose points, the six-dimensional deviation of the actual pose of the product clamping station relative to the theoretical pose is measured. Based on all the linked pose points and their corresponding six-dimensional deviations, a pose error mapping model covering the continuous workspace is generated through modeling, namely the system error digital model.

[0015] Furthermore, the pose error mapping model generated based on all linked pose points and their corresponding six-dimensional deviations is modeled using a neural network model. The neural network takes the flip angle and lifting height as inputs and the six-dimensional pose deviations as outputs, and learns the complex, nonlinear error transfer function of the detection device through training.

[0016] The technical solution of this invention can achieve the following technical effects: By setting up a flipping clamping mechanism to drive the product to flip and cooperating with a lifting mechanism to provide support during the non-flipping stage, the product's spatial posture is ensured to be stable. At the same time, by combining a dynamic error compensation method to correct the detection coordinate system, the positional shift and posture instability of the product during the flipping process can be effectively avoided, ensuring the accuracy of the detection points, improving the repeatability and reliability of the measurement data, and meeting the requirements of high-precision manufacturing for consistent detection. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Fig. 1 This is a schematic diagram of the precision detection device. Fig. 2 This is a schematic diagram of the flipping clamping mechanism and the lifting mechanism; Fig. 3 A flowchart of a dynamic error compensation method for a precision detection device. Fig. 4 A flowchart for performing the dynamic error compensation.

[0019] Reference numerals: 1. Conveying mechanism; 11. Pulling mechanism; 12. Conveyor belt mechanism; 121. First conveyor belt; 122. Second conveyor belt; 2. Tilting and clamping mechanism; 21. Clamping arm; 22. Clamping drive component; 23. Tilting drive component; 3. Lifting mechanism; 31. Lifting plate; 32. Support base; 33. Guide rod; 34. Lifting rod; 35. Drive component; 4. Detection mechanism; 41. Guide rail; 5. Robotic arm; 01. Products. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1: like Figs. 1-2 As shown: A precision detection device includes: a conveying mechanism 1, a flipping clamping mechanism 2, a lifting mechanism 3, a detection mechanism 4, and a robotic arm 5; The conveying mechanism 1 is located on both sides of the flipping clamping mechanism 2 and is used to convey the product to be tested 01 and the tested product 01. The flipping clamping mechanism 2 is used to clamp the product 01 to be tested after the robot arm 5 places it on the lifting mechanism 3, and to drive the product 01 to flip along its own central axis during the detection process so that the front and back sides of the product 01 can be detected in sequence. The lifting mechanism 3 is located below the flipping clamping mechanism 2. It is used to support the bottom of the product 01 during the clamping period and to lower itself when the product 01 is flipped, so as to ensure the spatial posture stability of the product 01 in the non-flipping stage. The detection mechanism 4 is configured to reciprocate above the flipping clamping mechanism 2 and perform non-contact detection on the product 01 when it moves directly above the product 01; The robotic arm 5 is used to place the product 01 to be tested into the clamping area of ​​the flipping clamping mechanism 2, and to classify the product 01 according to the test results.

[0023] The following describes the control flow of the accuracy detection device in a complete detection cycle: Initialization and standby: The control system is initialized, the lifting mechanism 3 rises to the preset position, the clamping arm 21 of the flipping clamping mechanism 2 is in the open state, the robot arm 5 is in the origin position, and the pulling mechanism 11 sends the tray loaded with the product to be tested 01 to the predetermined loading layer. Feeding and positioning: The control system first drives the robot arm 5 to move to the gripping position of the pull plate mechanism 11, and accurately positions the product to be tested 01 through the visual positioning or mechanical positioning device at its end, and performs the gripping action. Clamping and support: The robot arm 5 places the product 01 on the support block of the lifting mechanism 3, which has been raised to the waiting position. Then, the control system simultaneously issues commands: a) The clamping drive component 22 of the flip clamping mechanism 2 is activated, driving the two clamping arms 21 to close and reliably clamp the product 01 from both sides; b) The drive component 35 of the lifting mechanism 3 continues to work, so that the support block is in close contact with the bottom surface of the product 01 with constant pressure, completing the clamping and support. The clamping confirmation sensor installed on the clamping arm 21 and the pressure sensor on the lifting mechanism 3 will feed back the confirmation signal to the control system. First surface inspection: After receiving the clamping completion signal, the control system drives the inspection mechanism 4, which is installed on the linear guide rail 41, to move directly above the product 01. The inspection mechanism 4, such as a 3D line laser sensor, scans and inspects the first surface of the product 01 according to the preset program, and transmits the original point cloud data to the control system in real time.

[0024] Flipping and Transformation: After the first side inspection is completed, the control system executes strict timing control: a) Control the driving component 35 of the lifting mechanism 3 to reverse, so that the support block descends to the lowest safe position, completely clearing the flipping path of product 01; b) After receiving the sensor signal that the lifting mechanism 3 has descended to the position, control the flipping drive component 23 of the flipping clamping mechanism 2 to start, driving the clamped product 01 to rotate precisely 180° around the horizontal axis; c) After receiving the encoder feedback signal that the flipping mechanism has rotated to the position, control the lifting mechanism 3 to rise again, so that the support block re-contacts and supports the bottom surface of the flipped product 01.

[0025] Second-side inspection and sorting: After the lifting mechanism 3 is re-supported in place, the control system drives the inspection mechanism 4 again to scan and inspect the second surface of product 01. After the inspection is completed, the control system plans the movement trajectory of the robot arm 5 according to the inspection results, such as qualified / unqualified, and controls it to grab product 01 and place it on the corresponding first conveyor belt 121 (qualified product) or second conveyor belt 122 (unqualified product).

[0026] Reset: After product 01 is removed, the flip clamping mechanism 2 opens, and all mechanisms reset to standby state, waiting for the next cycle.

[0027] Through the aforementioned explicit, sensor-feedback-based sequential control logic, strict coordination and interlocking between the clamping, supporting, flipping, detection, and sorting actions are ensured, achieving automatic, stable, and reliable operation of the device.

[0028] Through the coordinated arrangement of the conveying mechanism 1, the flipping and clamping mechanism 2, the lifting mechanism 3, the detection mechanism 4, and the robotic arm 5, the product under test 01 can complete continuous precision detection of both sides within the same detection process and the same detection station. This avoids the inefficiency and inconsistent positioning problems caused by relying on manual flipping or multiple clamping in existing technologies. The cooperation between the flipping and clamping mechanism 2 and the lifting mechanism 3 during the detection and flipping process ensures that the product 01 has reliable support in the non-flipping stage and sufficient clearance in the flipping stage. This allows the flipping action to be completed while ensuring the spatial stability of the product 01, improving the correlation and consistency between the detection results of the two sides. The detection mechanism 4 adopts a non-contact detection method and moves back and forth above the flipping and clamping mechanism 2, enabling double-sided detection to be completed under the same detection benchmark conditions. This effectively reduces the detection errors introduced by station switching or repeated positioning. Overall, the structural coordination achieves an organic unity of clamping, support, flipping, and detection, which not only improves detection accuracy and efficiency but also facilitates the automation and continuity of the detection process, exhibiting good stability.

[0029] As a preferred embodiment of the above, the conveying mechanism 1 includes a pull plate mechanism 11 on one side and a conveyor belt mechanism 12 on the other side; The tray pulling mechanism 11 is used to carry a tray loaded with multiple products 01 to be tested, and to pull out the tray so that the products 01 to be tested in the tray are moved sequentially to the gripping position of the robot arm 5. The tray pulling mechanism 11 can also be used in conjunction with the lifting mechanism to lift the tray between different layers, thereby realizing the layer-by-layer feeding of multi-layer products 01; The conveyor belt mechanism 12 includes a first conveyor belt 121 and a second conveyor belt 122 respectively disposed on both sides of the device. The first conveyor belt 121 corresponds to the qualified product 01, and the second conveyor belt 122 corresponds to the unqualified product 01.

[0030] Specifically, the conveying mechanism 1 includes a tray-pulling mechanism 11 on one side of the device and a conveyor belt mechanism 12 on the other side, used to continuously feed the products to be tested 01 and to divert the output after testing. The tray-pulling mechanism 11 is used to carry a tray loaded with multiple products to be tested 01, and by driving the tray to pull it out horizontally, the products to be tested 01 in the tray are moved sequentially to the gripping position of the robot arm 5 according to the arrangement order, thereby realizing continuous feeding without manual intervention for each item. At the same time, the tray-pulling mechanism 11 can also be set in conjunction with a lifting mechanism, which drives the tray to move up and down between different height layers, so that the products to be tested 01 in the multi-layer tray can enter the working layer by layer. The positioning further improves the continuity of material supply. After the inspection is completed, the product 01 is placed by the robot arm 5 according to the inspection results. The first conveyor belt 121 in the conveyor belt mechanism 12 is used to receive and transport the qualified product 01, and the second conveyor belt 122 is used to receive and transport the unqualified product 01, so that qualified and unqualified products can be distinguished and output inside the device. Through the combination of the pull plate mechanism 11, the multi-layer lifting material supply method and the double conveyor belt diversion structure, this embodiment enables the inspection device to achieve automatic feeding of batch products 01 and diversion of inspection results while maintaining a stable inspection rhythm, reducing manual intervention and improving overall operating efficiency and automation level.

[0031] In this embodiment, the lifting mechanism 3 includes a lifting plate 31, a support base 32, a guide rod 33, a lifting rod 34, and a driving component 35; The drive component 35 is mounted on the support base 32, and the output end of the support base 32 is connected to the lifting rod 34; The lifting rod 34 can be movably passed through the support base 32; The lifting plate 31 is located at the top of the lifting rod 34; At least two guide rods 33 are provided below the lifting plate 31, extend parallel to the lifting rod 34, and are movably inserted through the support base 32; The drive unit 35 drives the lifting rod 34 to move along its axial direction, thereby causing the lifting plate 31 to move toward or away from the bottom of the clamped product 01.

[0032] Specifically, the lifting mechanism 3 is located below the flipping clamping mechanism 2. It includes a lifting plate 31, a support base 32, guide rods 33, a lifting rod 34, and a driving component 35. It provides controllable support to the bottom of the clamped product 01 during the inspection process. The driving component 35 is mounted on the support base 32, and its output end is connected to the lifting rod 34, allowing the lifting rod 34 to reciprocate axially under the constraint of the support base 32. The lifting plate 31 is located at the top of the lifting rod 34 and moves vertically towards or away from the bottom of the product 01 as the lifting rod 34 moves. To ensure the stability of the lifting process, at least two guide rods 33 are provided below the lifting plate 31. The guide rods 33 are arranged parallel to the lifting rod 34 and movably pass through the support base 32. The coordinated guidance of the guide rod 33 ensures that the lifting plate 31 maintains stable linear motion while bearing the weight of the product 01 and coordinating with the flipping action. When the product 01 is clamped by the flipping clamping mechanism 2 and is being inspected, the drive component 35 drives the lifting rod 34 to move upward, so that the lifting plate 31 contacts the bottom of the product 01 and forms support. When the flipping clamping mechanism 2 flips the product 01, the drive component 35 controls the lifting rod 34 to move downward, so that the lifting plate 31 moves away from the bottom of the product 01 to make room. After the flipping is completed, it returns to the rising state, thus achieving orderly switching between support and rooming under different working conditions of inspection and flipping. Through the above structure and action coordination, the lifting mechanism 3 effectively improves the posture stability of the product 01 during the inspection process while ensuring lifting accuracy and structural stability.

[0033] The flipping clamping mechanism 2 includes a clamping arm 21, a clamping drive component 22, and a flipping drive component 23. Two clamping arms 21 are arranged opposite each other for clamping the product 01 from both sides; The clamping drive component 22 is connected to the clamping arm 21 and is used to drive the two clamping arms 21 to move closer to each other to clamp the product 01 or to move away from each other to release the product 01. The flipping drive component 23 is connected to the clamping drive component 22 and is used to drive the clamping arm 21 and the clamped product 01 to rotate around the horizontal axis to achieve the flipping of the product 01.

[0034] Specifically, in this embodiment, the clamping drive component 22 can be a linear drive element, such as a cylinder, electric push rod, or servo cylinder. Its output end is connected to the clamping arm 21 via a linkage mechanism or directly to provide precise clamping force and opening / closing stroke. The flipping drive component 23 can be a rotary cylinder, servo motor, or stepper motor. Its output shaft is fixedly connected to the mounting base of the clamping drive component 22, thereby driving the entire clamping unit, including the clamping drive component 22, the clamping arm 21, and the clamped product 01, to perform a precise 180-degree flipping motion. Preferably, the flipping drive component 23 can integrate an encoder for real-time feedback and control of the flipping angle to ensure consistency of the front and back detection positions. The clamping surfaces of the two clamping arms 21 can... The device is equipped with a flexible pad or contoured slot to adapt to the shape of product 01 and prevent surface damage. During the flipping process, the clamping drive component 22 maintains the clamping state, ensuring that product 01 is always under reliable constraint throughout the flipping path, avoiding positional displacement due to inertia or self-weight. After the flipping is completed, the flipping clamping mechanism 2 continues to clamp product 01 to cooperate with the detection mechanism 4 to detect the flipped surface. After the detection is completed, the clamping arm 21 is controlled to open and release product 01. Through the coordinated setting of clamping drive and flipping drive, this embodiment achieves controllability and consistency of product 01 flipping action while ensuring clamping stability, which is beneficial to improving the posture stability and detection reliability during double-sided detection.

[0035] Example 2: Although the aforementioned precision testing device achieves automated testing of both sides of product 01 through the coordination of flipping clamping and lifting support, in actual operation, the inherent errors of the mechanical system, such as the radial runout of the flipping shaft, the gap of the guide mechanism, and the repeatability deviation, will still appear with changes in the motion posture. These errors will directly cause slight, nonlinear drift in the measurement coordinate system of the testing mechanism 4, such as the 3D vision sensor. This makes it impossible to complete high-precision testing of the front and back sides of the same product 01, such as position, perpendicularity, and curvature measurements, under a unified and stable benchmark, thus affecting the absolute accuracy of the final test results and the consistency between the front and back data.

[0036] To this end, this application further proposes a dynamic error compensation method, which aims to actively identify and offset the pose error introduced by the above-mentioned mechanical system in real time from the control and algorithm level, thereby achieving measurement performance that exceeds the mechanical manufacturing precision based on the hardware capabilities of the device. The core of this method is to construct and utilize a digital model that characterizes the comprehensive error of the system and to perform real-time compensation during the detection process.

[0037] The present invention also includes a dynamic error compensation method for the accuracy detection device as described above, such as... Fig. 3 As shown, the steps include: S10: Control the robot arm 5 to transfer the product 01 to be tested to the flipping clamping mechanism 2, and control the flipping clamping mechanism 2 to clamp the product 01, while controlling the lifting mechanism 3 to rise to support the product 01 from the bottom; S20: Control and testing mechanism 4 performs the first inspection on one surface of product 01; S30: Control the lifting mechanism 3 to descend and make way, then control the flipping clamping mechanism 2 to drive the product 01 to flip, and after the flipping is completed, control the lifting mechanism 3 to rise again to support the product 01; S40: Control the detection mechanism 4 to perform a second detection on the other opposite surface of product 01; S50: During the detection process of at least one surface, dynamic error compensation is performed; the dynamic error compensation is based on the pre-generated digital model of the system error of the detection device, combined with the real-time acquired device status data, to calculate the spatial pose deviation under the current detection state, and dynamically correct the measurement coordinate system of the detection mechanism 4 according to the spatial pose deviation. S60: Based on the error-compensated detection data, determine whether product 01 is qualified, and control the robot arm 5 to transfer product 01 to the corresponding sorting position.

[0038] Specifically, the systematic error digital model describes the systematic errors that the detection device may generate under different operating conditions. It can be pre-generated during equipment debugging or calibration. This digital model may include the installation error of the detection mechanism 4, the rotation error of the flipping clamping mechanism 2, the lifting error of the lifting mechanism 3, and the geometric deviation relationships between each motion axis, and is stored in the control system in the form of a mathematical model or lookup table. Device status data is collected in real time during the detection process. This data may include the rotation angle information of the flipping clamping mechanism 2, the current height position of the lifting mechanism 3, the position information of the detection mechanism 4 on the guide structure, and the device's working posture or environmental parameters. The control system integrates the real-time status data with the systematic error digital model to calculate the measurement coordinate system of the detection mechanism 4 relative to the ideal coordinate system under the current detection state. Spatial pose deviation; while collecting raw detection data, the control system of the detection mechanism 4 dynamically corrects the measurement coordinate system, so that error compensation is completed in the generation stage of the detection data; or after the detection is completed, the raw detection data can be uniformly compensated to obtain the compensated detection results. Based on the compensated detection data, the geometric accuracy parameters of product 01 are judged, and qualified or unqualified detection results are generated. Based on this, the robot arm 5 is controlled to perform corresponding sorting actions; through the synergy of the system error digital model and the real-time device status data, the detection coordinate system is dynamically corrected during the flipping and multi-condition detection of product 01, which effectively reduces the impact of mechanism movement and posture changes on detection accuracy, and makes the double-sided detection results have higher consistency and reliability, which is particularly suitable for high-precision, automated continuous detection scenarios.

[0039] To achieve high-precision dynamic error compensation, an error model is first constructed through system calibration: the driving flipping and lifting mechanism traverses multiple pose combinations, and high-precision instruments are used to measure the deviation between the actual position and the theoretical position. After collecting multiple sets of pose deviation data, a numerical fitting method is used to generate a digital model describing the distribution of system errors in the entire workspace. In real-time detection, the system queries this model to obtain the theoretical error under the current pose and introduces a Kalman filter algorithm for optimization: using the theoretical error as the initial estimate, combined with real-time monitoring data from auxiliary sensors such as vibration and deformation, the dynamic process model of the built-in mechanism motion state prediction is used for iterative optimal estimation. Finally, a comprehensive compensation amount that can simultaneously compensate for geometric system errors and motion dynamic hysteresis errors is output, and the reference of the detection coordinate system is corrected in real time accordingly.

[0040] As a preferred embodiment of the above, in step S50, such as Fig. 4 As shown, performing dynamic error compensation specifically includes: S51: Real-time acquisition of the real-time flipping angle of the flipping clamping mechanism 2 and the real-time support height of the lifting mechanism 3; S52: Based on the real-time flip angle and real-time support height, query the system error digital model to obtain the corresponding theoretical pose error; S53: Acquire real-time monitoring data from auxiliary sensors, correct theoretical pose error, and generate comprehensive pose error compensation amount; S54: Based on the comprehensive pose error compensation amount, the measurement coordinate system of the detection mechanism 4 is dynamically corrected.

[0041] Specifically, firstly, the real-time flipping angle of the flipping clamping mechanism 2 and the real-time support height of the lifting mechanism 3 are acquired in real time. The flipping angle can be obtained by the angle feedback unit of the flipping drive component 23, and the support height can be obtained by the displacement feedback device of the lifting mechanism 3. Subsequently, the control system uses the real-time flipping angle and real-time support height as state parameters to query the pre-established system error digital model, thereby obtaining the theoretical pose error corresponding to the current detection condition. This theoretical pose error is used to characterize the offset relationship of the measurement coordinate system of the detection mechanism 4 relative to the ideal measurement state. On this basis, the real-time monitoring data of the auxiliary sensors is further acquired, and the theoretical pose error is corrected according to the real-time monitoring data to compensate for the dynamic influence introduced by factors such as device vibration, environmental changes, or structural elastic deformation, thereby generating a comprehensive pose error compensation amount. Finally, the control system dynamically corrects the measurement coordinate system of the detection mechanism 4 according to the comprehensive pose error compensation amount, so that the detection data collected by the detection mechanism 4 in the current detection state is mapped to the compensated coordinate system.

[0042] The implementation of the dynamic error compensation method relies on a pre-established and stored digital model of system error in the control system. This model is essentially a function or dataset describing the mechanical system of the device, mainly the flipping clamping mechanism 2 and the lifting mechanism 3, and the mapping relationship between the actual spatial position and the ideal theoretical position of their clamping station under different postures.

[0043] In this embodiment, in step S53, the Kalman filter algorithm is used to fuse the real-time monitoring data of the auxiliary sensor to correct the theoretical pose error. The process model of the Kalman filter is constructed based on the gradient information of the digital model of the system error, and the observation model is constructed based on the measurement data of the auxiliary sensor. The comprehensive pose error compensation amount is optimally estimated through iterative calculation.

[0044] Specifically, after obtaining the real-time flipping angle of the flipping clamping mechanism 2 and the real-time support height of the lifting mechanism 3, and querying the theoretical pose error from the system error digital model, the control system uses the theoretical pose error as the initial estimate for Kalman filtering. The process model of Kalman filtering is constructed based on the gradient information of the system error digital model, used to describe the evolution trend of the pose error with changes in flipping angle and support height, giving the error estimation in the prediction stage a clear physical meaning. The observation model of Kalman filtering is constructed based on the real-time measurement data of auxiliary sensors, used to incorporate actual deviations caused by device vibration, attitude changes, or environmental factors into the estimation process. Through iterative calculations of prediction and updating, Kalman filtering dynamically corrects the theoretical pose error, optimally estimates the comprehensive pose error compensation amount under the current detection state, and dynamically corrects the measurement coordinate system of the detection mechanism 4 based on this compensation amount, thereby effectively improving the stability and accuracy of the detection results under complex working conditions.

[0045] In the process model of Kalman filtering, dynamic state variables representing the speed and acceleration of the flipping clamping mechanism 2 and the lifting mechanism 3 are introduced to construct a time-varying state transition equation, so that the estimated comprehensive pose error compensation can predict and compensate for the dynamic hysteresis error caused by the acceleration and deceleration of the mechanism.

[0046] Specifically, the angular velocity and angular acceleration of the flipping clamping mechanism 2 can be obtained from the feedback information of the flipping drive component 23, and the velocity and acceleration of the lifting mechanism 3 can be calculated based on the displacement change rate of the lifting rod 34. The control system combines the above velocity and acceleration parameters with the pose error parameters to form an extended state vector, and constructs a state transition equation that changes with time. This allows the Kalman filter to predict the error change based on the current acceleration and deceleration trend of the mechanism during the prediction stage. On this basis, the Kalman filter uses the theoretical pose error as the initial estimate and updates it by combining the real-time monitoring data of the auxiliary sensors. Finally, it obtains a comprehensive pose error compensation amount that reflects the dynamic working condition characteristics. In this way, the detection system can compensate for the error caused by dynamic lag in advance during the movement of the mechanism, so that the error compensation changes from a simple ex-post correction to a combination of prediction and compensation, thereby significantly improving the stability and accuracy of the detection results under high-speed operation conditions.

[0047] As a preferred embodiment of the above, in step S50, the digital model of the system error is pre-established through the following steps: A51: Drive the flipping clamping mechanism 2 and the lifting mechanism 3 to operate at multiple discrete linkage pose points; A52: At each linkage pose point, measure the six-dimensional deviation of the actual pose of the product 01 clamping station from the theoretical pose. A53: Based on all the linked pose points and their corresponding six-dimensional deviations, a pose error mapping model covering the continuous workspace is generated through modeling, namely, a system error digital model.

[0048] Specifically, the control mechanism 2 and the lifting mechanism 3 operate collaboratively at multiple discrete linkage pose points. Each linkage pose point is determined by the flipping angle of the flipping mechanism 2 and the support height of the lifting mechanism 3, allowing the product 01 clamping station to cover different posture combinations throughout the workspace. At each linkage pose point, the actual pose of the product 01 clamping station is measured using high-precision measurement methods and compared with the theoretical pose to obtain six-dimensional pose deviation data, including deviations in three translational directions and three rotational directions. After collecting the six-dimensional pose deviations of all linkage pose points, a pose error mapping model covering the continuous workspace is generated based on each linkage pose point and its corresponding six-dimensional pose deviation through interpolation, regression, or mapping modeling, thus forming a systematic error digital model. The systematic error digital model established in this way can output the corresponding theoretical pose error under different combinations of flipping angles and support heights, providing a reliable data foundation for subsequent dynamic error compensation and effectively improving the overall detection accuracy and consistency in the double-sided inspection process.

[0049] In this embodiment, a pose error mapping model is generated based on all linked pose points and their corresponding six-dimensional deviations, and a neural network model is used for modeling. The neural network takes the flip angle and lifting height as input and the six-dimensional pose deviation as output, and learns the complex, nonlinear error transfer function of the detection device through training.

[0050] Specifically, the control flipping clamping mechanism 2 and the lifting mechanism 3 operate collaboratively at multiple discrete linkage pose points. Each linkage pose point is determined by the flipping angle of the flipping clamping mechanism 2 and the support height of the lifting mechanism 3, allowing the product 01 clamping station to cover different posture combinations throughout the workspace. At each linkage pose point, the actual pose of the product 01 clamping station is measured using high-precision measurement methods and compared with the theoretical pose to obtain six-dimensional pose deviation data, including deviations in three translational directions and three rotational directions. After collecting the six-dimensional pose deviations of all linkage pose points, a pose error mapping model covering the continuous workspace is generated based on each linkage pose point and its corresponding six-dimensional pose deviation through interpolation, regression, or mapping modeling. This forms a systematic error digital model. The systematic error digital model established in this way can output the corresponding theoretical pose error under different combinations of flipping angles and support heights, providing a reliable data foundation for subsequent dynamic error compensation and effectively improving the overall detection accuracy and consistency in the double-sided inspection process.

[0051] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A precision detection device, characterized in that, include: Conveying mechanism, flipping and clamping mechanism, lifting mechanism, detection mechanism and robotic arm; The conveying mechanism is located on both sides of the flipping clamping mechanism and is used to convey the product to be tested and the product after testing. The flipping clamping mechanism is used to clamp the product after the robot arm places the product to be tested on the lifting mechanism, and to drive the product to flip along its own central axis during the detection process so that the front and back sides of the product can be detected in sequence. The lifting mechanism is located below the flipping clamping mechanism and is used to support the bottom of the product during the product is clamped and to lower itself when the product is flipped, so as to ensure the stability of the product's spatial posture during the non-flipping phase. The detection mechanism is configured to reciprocate above the flipping clamping mechanism and perform non-contact detection on the product when it moves directly above the product. The robotic arm is used to place the product to be tested into the clamping area of ​​the flipping clamping mechanism and to classify the product according to the test results.

2. The precision detection device according to claim 1, characterized in that, The conveying mechanism includes a pull plate mechanism on one side and a conveyor belt mechanism on the other side. The tray-pulling mechanism is used to carry a tray containing multiple products to be tested, and pulls out the tray to move the products to be tested in the tray to the gripping position of the robot arm in sequence. The tray pulling mechanism can also be coordinated with the lifting mechanism to allow the tray to move up and down between different layers, thereby realizing the layer-by-layer feeding of multi-layer products; The conveyor belt mechanism includes a first conveyor belt and a second conveyor belt disposed on both sides of the device. The first conveyor belt corresponds to products that pass inspection, and the second conveyor belt corresponds to products that fail inspection.

3. The precision detection device according to claim 1, characterized in that, The lifting mechanism includes a lifting plate, a support base, a guide rod, a lifting rod, and a driving component; The driving component is mounted on the support base, and the output end of the support base is connected to the lifting rod. The lifting rod can be movably passed through the support base; The lifting plate is disposed at the top of the lifting rod; At least two of the guide rods are disposed below the lifting plate, extend parallel to the lifting rod, and are movably inserted through the support base; The driving component drives the lifting rod to move along its axial direction, thereby causing the lifting plate to move toward or away from the bottom of the clamped product.

4. The precision detection device according to claim 1, characterized in that, The flipping clamping mechanism includes a clamping arm, a clamping drive component, and a flipping drive component; The two clamping arms are arranged opposite each other for clamping the product from both sides; The clamping drive component is connected to the clamping arm and is used to drive the two clamping arms to move closer to each other to clamp the product or to move away from each other to release the product. The flipping drive component is connected to the clamping drive component and is used to drive the clamping arm and the clamped product to rotate around a horizontal axis to achieve product flipping.

5. A dynamic error compensation method for the accuracy detection device as described in claim 1, characterized in that the steps include... include: The robot arm is controlled to transfer the product to be tested to the flipping clamping mechanism, and the flipping clamping mechanism is controlled to clamp the product. At the same time, the lifting mechanism is controlled to rise to support the product from the bottom. The detection mechanism is controlled to perform a first detection on one surface of the product; The lifting mechanism is controlled to descend and make way, then the flipping clamping mechanism is controlled to drive the product to flip, and after the flipping is completed, the lifting mechanism is controlled to rise again to support the product; The detection mechanism is controlled to perform a second detection on the other opposite surface of the product; During the detection process of at least one surface, dynamic error compensation is performed; the dynamic error compensation is based on a pre-generated digital model of the system error of the detection device, combined with real-time acquired device status data, to calculate the spatial pose deviation under the current detection state, and dynamically corrects the measurement coordinate system of the detection mechanism according to the spatial pose deviation. Based on the error-compensated test data, the product is determined to be qualified, and the robotic arm is controlled to move the product to the corresponding sorting position.

6. The dynamic error compensation method for the precision detection device according to claim 5, characterized in that, Performing the dynamic error compensation specifically includes: The real-time flipping angle of the flipping clamping mechanism and the real-time support height of the lifting mechanism are obtained in real time. Based on the real-time flip angle and real-time support height, the corresponding theoretical pose error is obtained by querying the system error digital model. The real-time monitoring data of the auxiliary sensor is acquired to correct the theoretical pose error and generate a comprehensive pose error compensation amount. The measurement coordinate system of the detection mechanism is dynamically corrected based on the comprehensive pose error compensation amount.

7. The dynamic error compensation method for the precision detection device according to claim 6, characterized in that, The Kalman filter algorithm is used to fuse the real-time monitoring data of the auxiliary sensor to correct the theoretical pose error. The process model of the Kalman filter is constructed based on the gradient information of the digital model of the system error, and the observation model is constructed based on the measurement data of the auxiliary sensor. The comprehensive pose error compensation amount is optimally estimated through iterative calculation.

8. The dynamic error compensation method for the precision detection device according to claim 7, characterized in that, In the Kalman filter process model, dynamic state variables characterizing the speed and acceleration of the flipping clamping mechanism and the lifting mechanism are introduced to construct a time-varying state transition equation, so that the estimated comprehensive pose error compensation amount can predict and compensate for the dynamic hysteresis error caused by the acceleration and deceleration of the mechanism.

9. The dynamic error compensation method for the precision detection device according to claim 5, characterized in that, The systematic error digital model is pre-established through the following steps: The flipping clamping mechanism and the lifting mechanism are driven to operate at multiple discrete linkage pose points; At each of the aforementioned linked pose points, the six-dimensional deviation of the actual pose of the product clamping station relative to the theoretical pose is measured. Based on all the linked pose points and their corresponding six-dimensional deviations, a pose error mapping model covering the continuous workspace is generated through modeling, namely the system error digital model.

10. The dynamic error compensation method for the precision detection device according to claim 9, characterized in that, The pose error mapping model is generated based on all linked pose points and their corresponding six-dimensional deviations, and a neural network model is used for modeling. The neural network takes the flip angle and lifting height as input and the six-dimensional pose deviation as output, and learns the complex, nonlinear error transfer function of the detection device through training.

Images