A laser optical axis calibration method and device based on diffraction image feature inversion of incident angle

Abstract

The application discloses a laser optical axis calibration method and device based on diffraction image feature inversion incident angle, the laser optical axis calibration method collects calibration diffraction images for a plurality of preset calibration tilt angles and calibration azimuth angles, extracts corresponding feature vectors, and constructs a mapping function in combination with first direction and second direction incident tilt angle components; current diffraction images are collected during calibration, two-dimensional incident tilt angle components of a laser beam are inverted according to the mapping function, and angle offset is obtained in combination with optical path distance, then two-dimensional physical translation of the laser beam is obtained according to the difference between the measured offset of the diffraction image and the angle offset, and the posture and position of the laser are adjusted accordingly. The device comprises a calibration component and a calibration execution component, the calibration execution component comprises a base, an optical axis adjusting seat, a posture adjusting mechanism, a position adjusting mechanism, a laser, a diffraction aperture plate, a detector and a processor.

Description

Technical Field

[0001] This invention belongs to the field of laser diffraction detection technology, specifically relating to a laser optical axis calibration method and device based on the inversion of the incident angle from diffraction image features. Background Technology

[0002] Laser diffraction is a technique for precisely measuring geometric dimensions by utilizing the diffraction phenomenon produced when a laser beam passes through a tiny aperture or slit. Its basic principle is as follows: when a coherent laser beam passes through a diffraction aperture, a diffraction image is formed in the far field. After the part under test is placed in the optical path, the slit formed between the part's edge and a reference object alters the distribution characteristics of the diffraction image. By comparing and analyzing the changes in the diffraction images before and after placing the part, the spacing dimensions at the gaps between parts can be deduced. This method has advantages such as being non-contact, highly accurate, and capable of measuring micrometer-level dimensions. It is widely used in various inspection scenarios in precision manufacturing, including but not limited to: edge gap detection of parts, thermal deformation measurement of components, warpage detection of disk-type parts, deformation assessment during machining, and measurement of shaft-hole fit offset.

[0003] In laser diffraction testing, the perpendicularity of the laser beam's incidence is a key factor determining measurement accuracy. The ultimate goal of calibration is to ensure the laser beam is perpendicular to the detector's receiving surface—because the diffraction pattern is ultimately received and analyzed by the detector, and the beam's tilt relative to the detector directly leads to pattern projection distortion. When the laser beam is incident on the diffraction aperture at an angle, the following adverse effects occur:

[0004] (1) Overall translation of the diffraction image: The tilted incident light causes the center of the diffraction image to deviate from the geometric center of the detector. Taking an optical path distance of L = 200 mm and a tilt angle of α = 0.01° as an example, the translation amount of the pattern center is... This translation can cause image analysis algorithms to misjudge the position of the stripe center, directly introducing gap measurement errors.

[0005] (2) Pattern asymmetry distortion: The tilted incident light disrupts the rotational symmetry of the diffraction aperture, causing the diffraction pattern to change from central symmetry to off-axis asymmetry, and the fringe spacing and dark ring position shift, further affecting the accuracy of gap measurement.

[0006] (3) Changes in light intensity distribution: The oblique incidence causes the aperture wall to partially block the light beam, resulting in a decrease in transmitted light intensity and uneven distribution, which reduces the signal-to-noise ratio and fringe contrast of the diffraction image.

[0007] It is worth noting that the changes in the diffraction patterns mentioned above (center shift, light intensity variation, pattern asymmetry) actually carry information about the incident angle—different tilt angles produce diffraction patterns with different characteristics. However, existing technologies do not utilize this information to extract angle data from the diffraction patterns themselves, but instead rely on external measurement tools to obtain the angle, making angle measurement and gap detection two independent processes.

[0008] In existing technologies, conventional methods to ensure perpendicular laser incidence mainly include:

[0009] (1) Machining surface as reference: The geometric relationship between the laser and the diffraction aperture plate is ensured by improving the machining accuracy of the mating surface of the positioning plate. However, the cumulative error after assembling multiple parts is usually on the order of 0.01~0.05mm, and the angular deviation generated on the lever arm of tens of millimeters far exceeds the requirements of micron-level detection. Once the machining is completed, this scheme cannot be adjusted and lacks the ability to compensate for machining and assembly errors.

[0010] (2) Fixed optical collimation: One-time calibration is performed using external optical instruments such as autocollimating telescopes or collimators. This method has high accuracy but the instruments are expensive, and real-time monitoring cannot be achieved after calibration. Once the device is affected by vibration or temperature changes and shifts, it needs to be recalibrated.

[0011] (3) Manual adjustment with multiple studs: The laser beam is manually adjusted to be approximately vertical by changing the tilt angle of the optical axis adjustment seat using multiple studs. However, this method has a fundamental drawback: it can only adjust the angle and cannot quantitatively determine the actual tilt angle after adjustment. The operator can only determine the approximate vertical state through qualitative judgments such as "light transmission / light blocking", and cannot obtain the precise value of the tilt angle. Therefore, it is impossible to determine whether the adjustment has met the accuracy requirements, nor can the angle deviation be used to compensate and correct subsequent measurement results.

[0012] (4) External angle measurement tools: Use tools such as a level and a protractor to measure the tilt angle of the positioning plate before each inspection. This method can achieve quantitative measurement, but it requires additional tools and operating steps, which increases the inspection time and operational complexity, and is inefficient in multi-station mobile application scenarios.

[0013] In summary, the existing technology has the following shortcomings:

[0014] 1. Diffraction images already contain angle information, but existing technologies cannot extract it from them; angle measurement and gap detection are disconnected from each other.

[0015] 2. Purely mechanical solutions cannot compensate for machining and assembly errors, and are significantly affected by temperature changes;

[0016] 3. External optical collimation solutions are expensive, inconvenient, and cannot achieve real-time online monitoring;

[0017] 4. Manual adjustment with multiple studs is a "blind adjustment" method, which cannot quantitatively determine the actual tilt angle;

[0018] 5. External angle measurement tools increase the number of operation steps and complexity, and are inefficient in portable multi-station scenarios. Summary of the Invention

[0019] The technical problem to be solved by this invention is: addressing the difficulty of existing laser diffraction detection devices in quantitatively obtaining the incident angle and different axis offsets of the laser beam from diffraction images, relying on external tools, and difficulty in achieving joint calibration of angle and position. This invention provides a laser optical axis calibration method and device based on the inversion of the incident angle from diffraction image features, thereby realizing the quantitative identification and calibration of laser optical axis attitude deviation and position deviation.

[0020] In a first aspect, the present invention provides a laser optical axis calibration method based on inverting the incident angle from diffraction image features, the method comprising:

[0021] For multiple preset calibration tilt angles, the laser beam is positioned relative to the diffraction aperture to form the current preset calibration tilt angle, and the laser beam is positioned around the vertical axis to correspond to multiple calibration azimuth angles in sequence. At each calibration azimuth angle, a detector is used to acquire the calibration diffraction image formed by the laser beam passing through the diffraction aperture, and the first incident tilt angle component and the second incident tilt angle component of the calibration diffraction image are obtained according to the calibration tilt angle and the calibration azimuth angle. The corresponding feature vector is extracted from each calibration diffraction image.

[0022] Based on the first direction incident tilt angle component, the second direction incident tilt angle component, and the feature vector corresponding to each calibration diffraction image, a mapping function between the feature vector and the tilt angle component is constructed.

[0023] The detector acquires the current diffraction image formed by the laser beam passing through the diffraction aperture under the state to be calibrated, and extracts the feature vector of the current diffraction image; the feature vector of the current diffraction image is input into the mapping function to obtain the first incident tilt angle component and the second incident tilt angle component of the laser beam under the state to be calibrated.

[0024] Based on the incident tilt angle components in the first and second directions and the optical path distance, the angular offset of the current diffraction image in the first and second directions is obtained.

[0025] The measured offsets of the current diffraction image in the first and second directions are obtained, and the differences between the measured offsets in the first and second directions and the corresponding angular offsets are calculated respectively to obtain the physical translation of the laser beam in the first and second directions.

[0026] The laser's attitude and position are adjusted based on the incident tilt angle component and physical translation amount corresponding to the current diffraction pattern to complete the calibration of the laser optical axis.

[0027] Preferably, the feature vector includes the measured offset of the diffraction image in the first and second directions, as well as the central peak light intensity and image asymmetry; the measured offset is the offset of the centroid of the diffraction image relative to the calibration null position.

[0028] Preferably, the angular offset is obtained by multiplying the tangent of the corresponding incident tilt angle component by the optical path distance.

[0029] Preferably, the first incident tilt angle component is obtained by multiplying the cosine of the calibrated azimuth angle by the calibrated tilt angle; the second incident tilt angle component is obtained by multiplying the sine of the calibrated azimuth angle by the calibrated tilt angle.

[0030] Preferably, the mapping function is constructed as follows:

[0031] Using the feature vectors corresponding to each calibrated diffraction image as input, and the corresponding first-direction incident tilt angle components and second-direction incident tilt angle components as supervision labels, a training sample set is constructed. The angle inversion model is trained using the training sample set, so that the angle inversion model establishes a mapping relationship from the feature vectors to the first-direction incident tilt angle components and second-direction incident tilt angle components.

[0032] Preferably, the angle inversion model is one of a neural network model, a linear model, a polynomial model, or a lookup table interpolation model.

[0033] Preferably, after obtaining the incident tilt angle component of the laser beam in the state to be calibrated, if the total incident tilt angle of the laser beam is greater than the preset angle threshold, the attitude of the laser is initially adjusted according to the incident tilt angle component, and the current diffraction image is re-acquired after the attitude adjustment to perform incident tilt angle inversion.

[0034] Secondly, the present invention provides a laser optical axis calibration device for retrieving the incident angle based on diffraction image features, which is used to perform the above-described laser optical axis calibration method; the laser optical axis calibration device includes a calibration component and a calibration execution component;

[0035] The calibration assembly includes a calibration support plate and a goniometer; the calibration support plate is used to connect the goniometer and the worktable; the goniometer is used to drive the calibration execution assembly to rotate around the vertical axis, so that the laser beam corresponds to multiple calibration azimuth angles in sequence;

[0036] The calibration execution component includes a base, an optical axis adjustment seat, an attitude adjustment mechanism, a position adjustment mechanism, a laser, a diffraction plate, a detector, and a processor;

[0037] The optical axis adjustment seat is mounted on the base via an attitude adjustment mechanism; the attitude adjustment mechanism is used to adjust the attitude of the optical axis adjustment seat; the laser is mounted on the optical axis adjustment seat via a position adjustment mechanism; the position adjustment mechanism is used to adjust the position of the laser in the first and second directions; the diffraction plate is disposed on the light-emitting side of the laser and has a diffraction aperture for the laser beam to pass through; the detector is disposed on the side of the diffraction plate away from the laser and is used to collect the diffraction image formed by the laser beam passing through the diffraction aperture;

[0038] The processor is connected to the detector and is used to obtain the incident tilt angle component and physical translation of the laser beam based on the diffraction image, and to determine the adjustment parameters of the attitude adjustment mechanism and the position adjustment mechanism based on the incident tilt angle component and the physical translation.

[0039] Preferably, the attitude adjustment mechanism includes multiple adjusting studs and corresponding locking nuts; the adjusting studs are spaced apart between the base and the optical axis adjustment seat; by changing the support height of each adjusting stud on the optical axis adjustment seat, the tilt attitude of the optical axis adjustment seat in the first direction and the second direction is adjusted.

[0040] Preferably, the aperture of the diffraction aperture is 0.1 mm to 0.5 mm; the laser is a semiconductor laser.

[0041] The beneficial effects of this invention are:

[0042] 1. This invention utilizes the angular information carried in the diffraction image itself to directly decode the two-dimensional tilt angle of the laser beam from the feature vector of the diffraction image through a pre-calibrated mapping function. This enables diffraction detection to have angle measurement capabilities in addition to gap detection. After calibration, the current tilt state can be known without the need for external angle measurement tools (level, goniometer, etc.) during each detection. At the same time, this invention uses the diffraction image features to invert the incident tilt angle components of the first and second directions, and then decouples the two-dimensional physical translation of the laser beam based on the difference between the measured offset and the angular offset. This allows for the determination of attitude adjustment and position adjustment, achieving joint calibration of laser optical axis angular deviation and position deviation.

[0043] 2. This invention employs an azimuth scanning calibration method. By adjusting the studs to set a fixed tilt amount, the angle measuring stage can be rotated around the vertical axis to scan all azimuth directions, obtaining effective two-dimensional tilt angles. Changing the tilt amount and repeating the scan can cover the complete two-dimensional angle space. This method, using a single-axis angle measuring stage and a four-stud mechanism, achieves systematic calibration data acquisition of the two-dimensional angle space with relatively low hardware costs.

[0044] 3. This invention arranges independent adjusting studs at the four corners of the optical axis adjusting seat. Adjustment in any direction can be achieved through combinations of different studs. Once the mapping function decodes the current tilt angle, the processor can directly output the adjustment amount, and adjust the corresponding studs accordingly for rapid correction, realizing a closed-loop process of "decoding, adjustment, and verification." Simultaneously, the decoding of the mapping function in this invention is automatically executed during each test, enabling real-time recording of angle data and out-of-tolerance alarms in conjunction with the processor. When the device tilts due to vibration or temperature changes, the system can issue a timely warning, avoiding testing under out-of-tolerance conditions.

[0045] 4. This invention is not limited to a specific inspection machine model or a single inspection object, but is applicable to all micron-level geometric measurement scenarios based on the principle of laser diffraction. As long as the inspection system includes a laser source, a diffraction aperture, and a detector, a mapping function can be constructed during the calibration stage, and angle decoding and measurement compensation can be achieved during the usage stage. Specific application scenarios include, but are not limited to: precision inspection of edge gaps of parts, online measurement of thermal deformation of components, warpage detection of disk-type parts, deformation assessment during processing, and measurement of shaft-hole fit offset, etc. In the above scenarios, the vertical incident accuracy of the laser beam directly affects the measurement accuracy, and the method provided by this invention can effectively solve the angle measurement and compensation problem. Attached Figure Description

[0046] Figure 1 This is an exploded view of the structure of Embodiment 1 of the present invention.

[0047] Figure 2 This is a front view of the calibration stage device in Embodiment 1 of the present invention.

[0048] Figure 3 This is a front view of the detection stage device in Embodiment 1 of the present invention.

[0049] Figure 4 This is a top view of the arrangement of adjusting studs in Embodiment 1 of the present invention.

[0050] Figure 5 This is a schematic diagram illustrating the principle of adjusting the stud in any direction of tilt in Embodiment 1 of the present invention.

[0051] Figure 6 This is a flowchart illustrating the overall operation of Embodiment 2 of the present invention.

[0052] Figure 7 This is a schematic diagram of the azimuth scanning data acquisition principle in Embodiment 2 of the present invention.

[0053] Figure 8 This is a flowchart of the mapping function construction and angle decoding method in Embodiment 2 of the present invention.

[0054] Figure 9This is a geometric diagram illustrating the decoupling of laser angle offset and translation in Embodiment 2 of the present invention.

[0055] Reference numerals in the attached diagram: 1. Base; 2. Optical axis adjustment seat; 3. Adjustment stud; 4. Laser; 5. Diffraction aperture plate; 6. Calibration support plate; 7. Angle measuring stage; 8. Level; 9. Detector. Detailed Implementation

[0056] The present invention will be further described below with reference to the accompanying drawings.

[0057] Example 1

[0058] like Figure 1 , Figure 2 , Figure 3 As shown, a laser optical axis calibration device based on the inversion of the incident angle using diffraction image features includes a calibration component and a calibration execution component. The calibration component is used to change the tilt angle and azimuth angle of the laser beam relative to the diffraction aperture to assist in acquiring calibration data. The calibration execution component is used to generate and acquire diffraction images and adjust the attitude and position of the laser based on the calibration results. The calibration component includes a calibration support plate 6 and a goniometer stage 7 mounted on the calibration support plate 6.

[0059] The calibration execution assembly includes a base 1 and, mounted on the base 1, an attitude adjustment mechanism, an optical axis adjustment seat 2, a position adjustment mechanism, a laser 4, a diffraction plate 5, a level 8, a detector 9, and a processor. The attitude adjustment mechanism is installed between the base 1 and the optical axis adjustment seat 2 and is used to adjust the attitude of the optical axis adjustment seat 2. The attitude adjustment mechanism includes four matching adjusting studs 3 and locking nuts.

[0060] The base 1 is a square aluminum alloy plate with a Φ20mm through hole in the center for bolts to pass through and fix it to the worktable or measuring platform 7. Each of the four corners of the base has an M6 threaded hole for installing adjusting studs 3. The lower surface of the base 1 is flush with the upper surface of the worktable or measuring platform 7. The base is made of 6061-T6 aluminum alloy and has undergone aging treatment to eliminate internal stress.

[0061] The optical axis adjusting seat 2 is a square aluminum alloy block. A Φ4mm center mounting hole is provided in the center of the optical axis adjusting seat 2 for setting the position adjustment mechanism and mounting the laser 4. Each of the four corners of the optical axis adjusting seat 2 has a Φ6.5mm through hole for the adjusting stud 3 to pass through. The upper surface of the optical axis adjusting seat 2 is a precision-machined plane with a flatness of 0.008mm and a surface roughness Ra of 1.6μm.

[0062] The position adjustment mechanism includes a laser mounting sleeve disposed in the central mounting hole, a first position adjustment component arranged in pairs along a first direction, a second position adjustment component arranged in pairs along a second direction, and a locking component; the first direction and the second direction are perpendicular to each other. The optical axis adjustment seat 2 has four position adjustment holes around the central mounting hole, each with a fine-tuning screw as a position adjustment component. The differential adjustment of the pairs of fine-tuning screws drives the laser mounting sleeve to move along the corresponding direction; after adjustment, the locking component secures the laser mounting sleeve.

[0063] Laser 4 is a 650nm wavelength semiconductor laser diode module with an operating voltage of 4V, an output power of approximately 5mW, and a housing diameter of Φ4mm. Laser 4 is fixed inside a laser mounting sleeve, allowing for two-dimensional positional fine-tuning while maintaining installation stability.

[0064] like Figure 4 and Figure 5 As shown, four studs 3 are M6×0.5 fine-pitch full-thread studs, which are respectively inserted into the four corner through holes of the optical axis adjusting seat 2. The lower end of each stud 3 is screwed into the M6 ​​threaded hole of the base 1, and the upper end is equipped with two M6 locking nuts. By rotating the stud 3 to change its screw depth into the base 1, the height of the corresponding angle of the optical axis adjusting seat 2 can be adjusted. The four studs 3 can be adjusted independently, so that the optical axis adjusting seat 2 can be tilted in any direction. After adjustment, the two locking nuts are tightened to prevent the studs 3 from loosening due to vibration.

[0065] The reason for choosing M6×0.5 fine thread is that the pitch of fine thread (0.5mm) is smaller than that of coarse thread (1mm), providing a smaller feed amount at the same rotation angle, thereby improving adjustment resolution. Using the diagonal length of the optical axis adjusting seat 2 as approximately 50mm as the lever arm, the feed amount per 1 / 10th of a revolution of the stud is 0.05mm, corresponding to an angle change of approximately... It can meet the needs of angle adjustment.

[0066] Both the diffraction aperture plate 5 and the calibration support plate 6 adopt a detachable installation structure. The diffraction aperture plate 5 is used to form a diffraction image; the calibration support plate 6 is used to connect the goniometer stage 7 and the worktable, so that the goniometer stage 7 can drive the base 1, the optical axis adjustment seat 2 and the laser 4 to rotate around the vertical axis to obtain multiple calibration azimuth angles.

[0067] The diffraction plate 5 is a 5mm thick aluminum alloy plate with a 0.3mm diffraction hole in the center, which is processed by precision drilling or laser drilling. The diffraction plate 5 is set on the light-emitting side of the laser 4. The laser beam passes through the diffraction hole to generate a Fraunhofer diffraction image, which is received by the detector 9 set above and transmitted to the processor for analysis.

[0068] During initial alignment, the position of the laser 4 is first adjusted using the position adjustment mechanism in the first and second directions, and the attitude adjustment mechanism is used to adjust the attitude of the optical axis adjustment seat 2 so that the laser beam can pass through the diffraction aperture on the diffraction aperture plate 5 and form a diffraction image on the detector 9; after the light transmission is completed, subsequent calibration and adjustment are performed.

[0069] The calibration assembly includes a calibration support plate 6 and a goniometer stage. The calibration support plate 6 is a 5mm thick aluminum alloy plate used to connect the goniometer stage 7 and the worktable, and is provided with mounting holes that mate with the goniometer stage 7 and the worktable.

[0070] During the calibration phase, the calibration support plate 6 is installed between the goniometer stage 7 and the worktable, with the diffraction aperture plate 5 positioned on the light-emitting side of the laser 4. The laser beam is adjusted by the stud 3 to form a known calibration tilt angle θ0 relative to the diffraction aperture. The goniometer stage 7 then rotates the base 1, the optical axis adjustment seat 2, and the laser 4 around the vertical axis to obtain multiple calibration azimuth angles. After calibration, the calibration components are removed, and the laser optical axis calibration device enters the testing phase.

[0071] It should be noted that during the calibration stage, the laser beam forms a calibration diffraction image through the diffraction aperture, and the detector 9 acquires the calibration diffraction image and transmits it to the processor. As a result, the training input of the mapping function and the current diffraction image in the calibration stage have the same imaging mechanism, avoiding the introduction of system deviation due to the difference between the calibration image type and the image type used.

[0072] In some embodiments, the adjusting stud is a differential thread stud to further improve the adjustment resolution. The differential thread stud includes an upper thread and a lower thread: the upper thread is an M5×0.25 ultra-fine thread, which mates with the lock nut; the lower thread is an M6×0.5 fine thread, which mates with the threaded hole in the base 1. Both threads have the same direction of rotation. When the stud is rotated one revolution, the upper thread raises the optical axis adjusting seat 2 by 0.25 mm, while the lower thread screws the stud into the base 1 by 0.5 mm. The difference between the two is the effective feed rate: 0.5 − 0.25 = 0.25 mm / revolution. Compared to the 0.5 mm / revolution of a standard M6×0.5 stud, the adjustment resolution is improved by two times. The manually resolvable feed rate of 1 / 20th of a revolution is 0.0125 mm, corresponding to an angle change of approximately arctan(0.0125 / 50) ≈ 0.014° over a 50 mm lever arm, which is significantly better than a standard stud.

[0073] In some embodiments, a reflector is mounted above an optical axis adjustment base, and calibration is performed using the characteristics of the reflected light after the laser beam illuminates the reflector. Since the reflector is mounted above the optical axis adjustment base, its mirror surface is perpendicular to the upper surface of the base (i.e., perpendicular to the laser beam direction), and the reflected light will return to the laser along its original path. Therefore, the calibration function of the reflector needs to be used in conjunction with external optical instruments (such as an autocollimator), or the following method can be employed: a transparent glass plate is placed at a 45° angle in front of the reflector as a beam splitter, directing a portion of the reflected light onto a side screen for observation.

[0074] Example 2

[0075] like Figure 6 As shown, a laser optical axis calibration method based on inverting the incident angle using diffraction image features includes the following steps:

[0076] Step S1, Initialization

[0077] During the calibration phase, the calibration components are assembled with the calibration execution components, and the optical axis adjustment seat 2 is adjusted to the initial reference posture by adjusting the stud 3 and the level 8. The laser 4 is turned on to confirm that the laser beam can pass through the diffraction aperture on the diffraction plate 5 and form a diffraction image on the detector 9.

[0078] In this embodiment, the wavelength λ of the laser beam is 650nm.

[0079] Step S2, Calibration Data Acquisition

[0080] like Figure 7 As shown, by rotating the stud 3, the optical axis adjusting seat 2 is tilted relative to the platform of the angle measuring stage 7 to the set position. Each calibration tilt angle Then, rotate the goniometer 7 around the vertical axis, and during the rotation, set each calibrated azimuth angle. The first direction of incident tilt angle component Second direction incident tilt angle component Based on the tilt angle component of the laser beam and tilt angle components Extracting the calibration azimuth angle Feature vector of the diffraction image It is represented as follows:

[0081]

[0082] in, This represents the offset (measured offset) of the center of the diffraction image relative to the calibration zero point (the center of the diffraction image when the tilt angle is 0). The central peak light intensity of the diffraction image corresponding to the i-th calibrated azimuth angle is... ; The theoretical maximum central peak intensity of the diffraction image; The thickness of the diffraction aperture plate 5; The radius of the diffraction aperture is 0.3 mm in this embodiment; Let be the asymmetry of the diffraction pattern corresponding to the i-th azimuth angle. ; and These are the total light intensities (obtained by accumulating the gray levels of detector pixels) of the diffraction image corresponding to the i-th azimuth angle within the windows on the left and right sides of the central axis.

[0083] The first direction and the second direction correspond to the X and Y directions, which are perpendicular to each other within the detector's receiving surface, respectively.

[0084] In this embodiment, if the tilt angle is calibrated If the angle is less than 0.3°, then the offset is... and It can independently support high-precision fitting without measuring feature vectors. The central peak light intensity and asymmetry This directly increases the central peak light intensity and asymmetry It is zero.

[0085] Repeat the feature vector extraction process described above to extract features at different calibration tilt angles. Different azimuth angles Corresponding feature vector This allows for coverage of the required two-dimensional angular range.

[0086] Step S3: Construct the mapping function

[0087] like Figure 8 As shown, a dataset is constructed based on the calibration data extracted in step S2. . Using feature vectors As input, the tilt angle component of the laser beam and For the output, a multi-layer feedforward neural network (such as a fully connected network with 12 hidden layers and 832 nodes per layer) is constructed. Data from the dataset is used as training samples, and a supervised learning approach is employed to train the multi-layer feedforward neural network—that is, each set of data... As an "input-label" pair, the mean square error between the network output and the true angle is minimized using the backpropagation algorithm. After training, the neural network automatically establishes a mapping relationship from diffraction image features to tilt angles, without the need for manual derivation or selection of function forms. During use, simply input the real-time extracted feature vectors into the trained multilayer feedforward neural network to directly output the tilt angle components. The advantages of this method are: strong adaptability to nonlinear mapping relationships, no need to pre-assume function forms; the richer the calibration data, the higher the network fitting accuracy; and the trained network can be directly reused on the same device without retraining each time.

[0088] In some embodiments, the mapping relationship between the feature vector and the tilt angle component is constructed by building a linear model, a polynomial model, or by table lookup interpolation.

[0089] In this embodiment, the mapping function is updated at preset time intervals (e.g., every 6 months or after the device has undergone maintenance or relocation) to ensure that the mapping function matches the current state of the device. By comparing the parameter changes of the old and new mapping functions, the stability of the device can be assessed, providing a basis for maintenance decisions. The necessity of updating the mapping function lies in the fact that although tightening the lock nut can prevent the stud from loosening, temperature changes during long-term use (aluminum alloy has a thermal expansion coefficient of 23 × 10⁻⁻⁻⁴) can still contribute to the stability of the device. 6 / °C, 30mm height varies by about 0.7μm per 1°C) and mechanical vibration can still cause slight offsets, causing the mapping function to deviate from the actual state.

[0090] Step S4: Preparation of the detection device

[0091] During the testing phase, the calibration execution component is adjusted to the calibration state, the base 1 is placed on the worktable, the diffraction plate 5 is kept on the light-emitting side of the laser 4, and the detector 9 is placed at a distance of L=200mm above the diffraction plate 5. The detector 9 is connected to the processor.

[0092] Step S5: Diffraction Image Acquisition and Angle Decoding

[0093] Laser 4 is activated, and the laser beam generates a diffraction image through the diffraction aperture. Detector 9 acquires the diffraction image and transmits it to the processor. The processor extracts feature vectors from the image. Substituting the feature vector into the mapping function constructed in step S3, the tilt angle component of the laser beam is obtained. and Based on tilt angle components and Obtain the total tilt angle .

[0094] In this embodiment, if the total tilt angle greater than the preset angle threshold If the angle threshold is too high, the processor will issue an alarm signal (such as a screen prompt or audible alarm) to prompt the operator to adjust stud 3 and reacquire and decode the diffraction image. This prevents detection from being performed under excessive tilt and avoids affecting detection accuracy due to excessive compensation residual. The preferred value is 0.1°.

[0095] Taking a typical working condition as an example: if the measured offset in the first direction is 0.035mm and the measured offset in the second direction is 0.023mm (corresponding to tilts of approximately 0.01° and 0.007°), the central peak light intensity and asymmetry are close to the calibrated zero value. The mapping function decodes the incident tilt angle component in the first direction as 0.01° and the incident tilt angle component in the second direction as 0.007°, with a total incident tilt angle of approximately 0.012°.

[0096] Step S6, Optical Axis Calibration

[0097] The attitude deviation of the optical axis adjustment seat 2 is determined based on the first direction incident tilt angle component and the second direction incident tilt angle component obtained by inversion in step S5, and the adjustment direction and adjustment amount of each adjustment stud 3 in the attitude adjustment mechanism are determined accordingly.

[0098] Based on the measured offsets of the current diffraction image in the first and second directions and the corresponding incident tilt angle components, the angular offset in the first direction is obtained. Second direction angular offset Its expression is:

[0099]

[0100]

[0101] like Figure 9 As shown, taking the first direction as an example, the original position O of the laser, the offset position O′, the original image center D of the detector, and the offset image center D′ constitute a geometric relationship, and DD′ corresponds to the measured offset. OO′ corresponds to the physical translation amount The angular offset in the first direction caused by the tilt of the laser beam Therefore, the physical translation can be obtained by subtracting the angular offset from the measured offset. That is, the physical translations of the laser beam in the first and second directions are respectively... = , .

[0102] Therefore, by first inverting the incident tilt angle and then decoupling the physical translation, the measured offset on detector 9 can be decomposed into angular offset components and positional offset components. Based on the physical translation... and Determine the adjustment direction and amount of the position adjustment mechanism, and drive the laser 4 to make fine adjustments in the opposite direction; after adjustment, reacquire the current diffraction image and repeat the angle inversion and displacement decoupling until the total incident tilt angle and the two-dimensional physical translation both meet the preset calibration conditions, and complete the laser optical axis calibration.

[0103] Through the above attitude calibration and position calibration, the angular deviation and misalignment of the laser beam can be eliminated respectively, so that the laser optical axis remains stable relative to the geometric reference of the diffraction aperture and detector 9, thereby providing consistent incident conditions for subsequent laser diffraction detection.

Claims

1. A laser optical axis calibration method based on inverting the incident angle using diffraction image features, characterized in that: The method includes: For multiple preset calibration tilt angles, the laser beam is positioned relative to the diffraction aperture to form the current preset calibration tilt angle, and the laser beam is positioned around the vertical axis to correspond to multiple calibration azimuth angles in sequence. At each calibration azimuth angle, a detector is used to acquire the calibration diffraction image formed by the laser beam passing through the diffraction aperture, and the first incident tilt angle component and the second incident tilt angle component of the calibration diffraction image are obtained according to the calibration tilt angle and the calibration azimuth angle. The corresponding feature vector is extracted from each calibration diffraction image. Based on the first direction incident tilt angle component, the second direction incident tilt angle component, and the feature vector corresponding to each calibrated diffraction image, a mapping function between the feature vector and the tilt angle component is constructed. The detector acquires the current diffraction image formed by the laser beam passing through the diffraction aperture under the state to be calibrated, and extracts the feature vector of the current diffraction image; the feature vector of the current diffraction image is input into the mapping function to obtain the first incident tilt angle component and the second incident tilt angle component of the laser beam under the state to be calibrated. Based on the incident tilt angle components in the first and second directions and the optical path distance, the angular offset of the current diffraction image in the first and second directions is obtained. The measured offsets of the current diffraction image in the first and second directions are obtained, and the differences between the measured offsets in the first and second directions and the corresponding angular offsets are calculated respectively to obtain the physical translation of the laser beam in the first and second directions. The laser's attitude and position are adjusted based on the incident tilt angle component and physical translation amount corresponding to the current diffraction pattern to complete the calibration of the laser optical axis.

2. The laser optical axis calibration method based on inversion of incident angle using diffraction image features according to claim 1, characterized in that: The feature vector includes the measured offset of the diffraction image in the first and second directions, as well as the central peak light intensity and image asymmetry; the measured offset is the offset of the centroid of the diffraction image relative to the calibration zero position.

3. The laser optical axis calibration method based on the inversion of the incident angle using diffraction image features according to claim 1, characterized in that: The angular offset is obtained by multiplying the tangent of the corresponding incident tilt angle component by the optical path distance.

4. The laser optical axis calibration method based on the inversion of the incident angle using diffraction image features according to claim 1, characterized in that: The first incident tilt angle component is obtained by multiplying the cosine of the calibrated azimuth angle by the calibrated tilt angle; the second incident tilt angle component is obtained by multiplying the sine of the calibrated azimuth angle by the calibrated tilt angle.

5. The laser optical axis calibration method based on the inversion of the incident angle using diffraction image features according to claim 1, characterized in that: The mapping function is constructed as follows: Using the feature vectors corresponding to each calibrated diffraction image as input, and the corresponding first-direction incident tilt angle components and second-direction incident tilt angle components as supervision labels, a training sample set is constructed. The angle inversion model is trained using the training sample set, so that the angle inversion model establishes a mapping relationship from the feature vectors to the first-direction incident tilt angle components and second-direction incident tilt angle components.

6. The laser optical axis calibration method based on the inversion of the incident angle using diffraction image features according to claim 5, characterized in that: The angle inversion model is one of the following: neural network model, linear model, polynomial model, or lookup table interpolation model.

7. The laser optical axis calibration method based on the inversion of the incident angle using diffraction image features according to claim 1, characterized in that: After obtaining the incident tilt angle component of the laser beam under the state to be calibrated, if the total incident tilt angle of the laser beam is greater than the preset angle threshold, the attitude of the laser is initially adjusted according to the incident tilt angle component, and the current diffraction image is re-acquired after the attitude adjustment to perform incident tilt angle inversion.

8. A laser optical axis calibration device for retrieving the incident angle based on diffraction image features, characterized in that: The laser optical axis calibration device is used to perform the laser optical axis calibration method based on diffraction image feature inversion of the incident angle as described in claim 1; the laser optical axis calibration device includes a calibration component and a calibration execution component; The calibration assembly includes a calibration support plate (6) and a goniometer (7); the calibration support plate (6) is used to connect the goniometer (7) and the worktable; the goniometer (7) is used to drive the calibration execution assembly to rotate around the vertical axis, so that the laser beam corresponds to multiple calibration azimuth angles in sequence; The calibration execution component includes a base (1), an optical axis adjustment seat (2), an attitude adjustment mechanism, a position adjustment mechanism, a laser (4), a diffraction aperture plate (5), a detector (9), and a processor; The optical axis adjustment seat (2) is mounted on the base (1) via an attitude adjustment mechanism; the attitude adjustment mechanism is used to adjust the attitude of the optical axis adjustment seat (2); the laser (4) is mounted on the optical axis adjustment seat (2) via a position adjustment mechanism; the position adjustment mechanism is used to adjust the position of the laser (4) in the first direction and the second direction; the diffraction plate (5) is disposed on the light-emitting side of the laser (4) and is provided with a diffraction aperture for the laser beam to pass through; the detector (9) is disposed on the side of the diffraction plate (5) away from the laser (4) and is used to collect the diffraction image formed by the laser beam passing through the diffraction aperture; The processor is connected to the detector (9) and is used to obtain the incident tilt angle component and physical translation of the laser beam according to the diffraction image, and to determine the adjustment parameters of the attitude adjustment mechanism and the position adjustment mechanism according to the incident tilt angle component and the physical translation.

9. A laser optical axis calibration device based on inverting the incident angle using diffraction image features according to claim 8, characterized in that: The attitude adjustment mechanism includes multiple adjusting studs (3) and corresponding locking nuts; the adjusting studs (3) are spaced apart between the base (1) and the optical axis adjustment seat (2); by changing the support height of each adjusting stud (3) on the optical axis adjustment seat (2), the tilt attitude of the optical axis adjustment seat (2) in the first direction and the second direction is adjusted.

10. A laser optical axis calibration device based on inverting the incident angle using diffraction image features according to claim 8, characterized in that: The aperture of the diffraction aperture is 0.1 mm to 0.5 mm; the laser (4) is a semiconductor laser.

Images