Five-dimensional lens adjustment device for vertical laser interferometer and method of using the same

Abstract

This invention discloses a five-dimensional lens adjustment device and its usage method for a vertical laser interferometer, relating to the field of optical instrument technology. The device includes a base assembly and a laser interferometer, a three-dimensional translational large-span adjustment layer, a two-dimensional attitude micro-step adjustment layer, and a control module. The control module is configured to execute a two-way collaborative closed-loop adjustment strategy: preprocessing the interferogram frames acquired by the laser interferometer and extracting the central field-of-view feature parameters, evaluating the fringe contrast and zero-position offset; after the three-dimensional translational large-span adjustment is completed, initiating the pitch and yaw two-dimensional attitude micro-step adjustment, reusing the same feature calculation results from the first stage, and using zero-position offset feedback to compensate for the previous accumulated drift. The device of this invention adopts a hierarchical architecture of a three-dimensional translational large-span adjustment layer and a two-dimensional attitude micro-step adjustment layer. The functions of the two layers are decoupled and clearly divided, reducing the complexity of the control system and improving the efficiency and accuracy of lens adjustment.

Description

Technical Field

[0001] This invention relates to the field of optical instrument technology, and in particular to a five-dimensional lens adjustment device for a vertical laser interferometer and its usage method. Background Technology

[0002] In the field of optical interferometry precision measurement, vertical laser interferometers are widely used for detecting surface shape errors in optical components such as lenses and mirrors. During the detection process, the accuracy of the spatial pose adjustment of the lens under test relative to the wavefront emitted by the interferometer directly affects the quality of the interferogram and the accuracy of the surface shape detection results.

[0003] Existing five-dimensional lens adjustment devices on the market are bulky, heavy, and have small apertures. Adjusting pitch and yaw angles is inconvenient, and the two-layer adjustment mechanism is unsuitable for use with vertical laser interferometers. Furthermore, the adjustment mechanisms often rely on manual adjustment of each dimension using lead screws and spring resets, depending on operator experience, resulting in low efficiency and difficulty in quantifying the actual displacement and angle adjustment of each axis, thus hindering automated closed-loop control. Alternatively, existing electronic control repeatedly iterates and adjusts the X / Y translation and pitch / yaw attitude of the lens under test until the optical axis coincides with the turntable axis. While this method can achieve high-precision pose positioning, it relies on multiple rotation measurements and iterative calculations, making the adjustment process cumbersome and time-consuming. The lack of effective coordination and data reuse mechanisms in the adjustment stage means that pose deviations introduced by previous adjustments cannot be effectively compensated for in subsequent adjustments, affecting the simultaneous improvement of adjustment efficiency and accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a five-dimensional lens adjustment device for a vertical laser interferometer and its usage method, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a five-dimensional lens adjustment device for a vertical laser interferometer, comprising a base assembly and a laser interferometer, and further comprising: A three-dimensional translational large-span adjustment hierarchy consisting of X-axis translational units, Y-axis translational units, and Z-axis translational units; The pitch and yaw adjustment unit is fixedly connected to the Y-axis translation unit, and the two together constitute a two-dimensional attitude microstep adjustment hierarchy. And a control module, which is signal-connected to each translation unit and attitude adjustment unit, wherein the control module is configured to execute a two-way collaborative closed-loop control strategy: The first step involves performing XYZ three-dimensional translational large-span calibration, simultaneously calling the image preprocessing and central field-of-view feature calculation modules to preprocess the interferometric images acquired by the laser interferometer and extract the central field-of-view feature parameters, and evaluate the fringe contrast and zero-position offset. In the second stage, after the three-dimensional translational large-span adjustment is completed, the pitch and yaw two-dimensional attitude micro-step adjustment is initiated. The same feature calculation results calculated in the first stage are reused, and a lightweight pose re-verification mechanism for the inter-stage switching point is introduced to compensate for the previous cumulative drift using zero-position offset feedback. The lightweight pose recalibration mechanism incorporates the accumulated drift amount as the initial compensation value for the attitude microstepping adjustment stage through a closed-loop control algorithm, thereby achieving seamless drift connection and compensation between the two stages.

[0006] Preferably, the method of using zero-point bias feedback to compensate for preceding cumulative drift specifically includes: At the switching point between three-dimensional translational large-span adjustment and two-dimensional attitude micro-step adjustment, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated. The accumulated drift is decomposed into drift components in the pitch and yaw directions, which are then used as the initial compensation offset input to the drive control of the pitch and yaw adjustment unit during the attitude microstepping adjustment stage. During the attitude microstepping adjustment process, the zero-position offset signal is acquired in real time, and a closed-loop control algorithm is used to dynamically correct the compensation offset until the zero-position offset converges to the allowable range.

[0007] Preferably, in the central field-of-view feature calculation module, the central field of view is a 512×512 pixel area, and the feature parameters include at least the stripe contrast value and the zero offset value; the frame preprocessing includes image denoising, contrast enhancement and edge detection.

[0008] Preferably, the pitch and yaw adjustment unit is used to achieve precise two-dimensional attitude adjustment of the lens in terms of pitch and yaw.

[0009] Preferably, the base assembly includes a flat marble plate and a column marble support plate vertically fixed on the flat marble plate, the laser interferometer is fixed on the flat marble plate, the laser interferometer is used to emit a measurement wavefront and receive the interference pattern formed after being returned by the lens under test, wherein the Z-axis translation unit is connected to the column marble support plate; The Z-axis translation unit is connected to the base assembly, the X-axis translation unit is connected to the Z-axis translation unit, and the Y-axis translation unit is connected to the X-axis translation unit. The X-axis translation unit, Y-axis translation unit, and Z-axis translation unit are all equipped with displacement sensing modules, and the pitch and yaw adjustment unit is equipped with an angle sensing module. Each sensing module is connected to the control module for signal feedback of the actual displacement and attitude adjustment of each axis in real time.

[0010] Preferably, in the dual-path collaborative closed-loop adjustment, the translational large-span adjustment of the first path and the attitude micro-step adjustment of the second path use the same set of feature calculation results as the evaluation basis to avoid systematic deviations introduced by the difference in feature extraction between the two paths.

[0011] A method for five-dimensional lens adjustment of a vertical laser interferometer, comprising the aforementioned five-dimensional lens adjustment device for a vertical laser interferometer, specifically including the following steps: Step 1: Place the lens to be tested on the movable support plate of the pitch and yaw adjustment unit, start the laser interferometer, and acquire the initial interference pattern frame; Step 2: The control module performs the first adjustment: Based on the fringe contrast and zero offset information output by the frame preprocessing and central field-of-view feature calculation module after the initial interferogram frame is processed, the X-axis translation unit, Y-axis translation unit and Z-axis translation unit are driven to perform three-dimensional translation large-span adjustment until the fringe contrast reaches the preset threshold. Step 3: The control module performs the second stage of adjustment: At the switching point between stages, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated; the same feature calculation result in step 2 is reused to decompose the cumulative drift into drift components in the pitch and yaw directions, which are used as the initial compensation offset input to the drive control of the attitude adjustment unit to start the pitch-yaw two-dimensional attitude microstep adjustment; Step 4: During the attitude microstep adjustment process, the zero-position offset signal is acquired in real time, and the compensation offset is dynamically corrected using a closed-loop control algorithm until the zero-position offset converges to the allowable range, thus completing the five-dimensional pose adjustment of the lens.

[0012] Preferably, the closed-loop control algorithm is a PID control algorithm or an adaptive control algorithm, which uses the zero-position bias value as the control input and outputs an attitude adjustment drive signal.

[0013] Preferably, the calculation of the cumulative drift includes: decomposing the difference between the current actual pose and the theoretical target pose of each axis into two degrees of freedom, pitch and yaw, through coordinate transformation, and obtaining the drift component value in each direction.

[0014] Preferably, it also includes an iterative optimization step: after completing the attitude microstep adjustment, if the zero-position offset does not meet the preset accuracy requirements, then repeat steps two to four until the quality of the interferogram meets the detection requirements.

[0015] The technical effects and advantages of this invention are as follows: (1) The device of the present invention adopts a layered architecture of three-dimensional translational large-span adjustment level and two-dimensional attitude micro-step adjustment level. The large-span adjustment level is responsible for the rapid positioning of the lens in the three-axis direction in space, and the micro-step adjustment level is responsible for the precise adjustment of the pitch and yaw attitude of the lens. The functions of the two levels are decoupled and the division of labor is clear, avoiding the contradiction between stroke and accuracy of a single adjustment mechanism, so that large-range movement and high-precision positioning can be taken into account. The dual-stroke collaborative closed-loop adjustment strategy decomposes the five-dimensional adjustment into two logically independent steps, forming a step-by-step implementation of coarse and fine adjustment. Compared with the traditional five-dimensional simultaneous adjustment method, it reduces the complexity of the control system, improves the convergence speed and operability of adjustment, and improves the lens adjustment efficiency and accuracy.

[0016] (2) The present invention uses the same set of feature calculation results as the evaluation basis for the first translational calibration and the second attitude calibration, which is different from the traditional approach of extracting features independently in different adjustment stages. This fundamentally avoids the systematic deviation caused by differences in feature extraction algorithms or different image acquisition times, and ensures the consistency and continuity of the evaluation criteria between the two stages. (3) At the switching node between coarse and fine adjustment, the present invention reads the actual pose data of each axis in real time through the displacement sensing module and the angle sensing module, compares it with the theoretical target pose and calculates the cumulative drift amount, directly quantifies the cumulative deviation caused by mechanical clearance, thermal deformation or control error in the previous adjustment, and injects it into the second stage control in the form of initial compensation bias amount, thus realizing the seamless connection of drift. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of a partial explosion of the device of the present invention; Figure 2 This is a schematic diagram of the front structure of the device of the present invention; Figure 3 This is a flowchart of the method steps of the present invention.

[0018] In the attached diagram: 1. X-axis translation unit; 2. Y-axis translation unit; 3. Z-axis translation unit; 4. Pitch and yaw adjustment unit; 5. Laser interferometer; 6. Flat marble plate; 7. Column marble support plate; 8. Linear bearing; 9. Linear guide rail. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This invention provides, for example Figures 1-3 The image shows a five-dimensional lens adjustment device for a vertical laser interferometer.

[0021] Example 1: Includes a base assembly, a laser interferometer 5, an X-axis translation unit 1, a Y-axis translation unit 2, a Z-axis translation unit 3, a pitch and yaw adjustment unit 4, and a control module. The Z-axis translation unit 3 is connected to the base assembly. The base assembly includes a flat marble plate 6 and a vertically fixed marble support plate 7 on the flat marble plate 6. The laser interferometer 5 is fixed on the flat marble plate 6. The laser interferometer 5 uses a frequency-stabilized He-Ne laser with a wavelength stability of ±0.1ppm. The output beam aperture is determined according to the aperture of the lens under test. The laser interferometer 5 is used to emit a measurement wavefront and receive the interference pattern formed after returning from the lens under test. Among them, the Z-axis translation unit 3 is connected to the column marble support plate 7, the X-axis translation unit 1 is connected to the Z-axis translation unit 3, the Y-axis translation unit 2 is connected to the X-axis translation unit 1, and the pitch and yaw adjustment unit 4 is fixedly connected to the Y-axis translation unit 2. The three-dimensional translation large-span adjustment level is composed of the X-axis translation unit 1, the Y-axis translation unit 2 and the Z-axis translation unit 3. The pitch and yaw adjustment unit 4 and the Y-axis translation unit 2 constitute a two-dimensional attitude micro-step adjustment level. This facilitates the three-dimensional translation large-span adjustment level to achieve coarse adjustment of the lens XYZ axis, and the two-dimensional attitude micro-step adjustment level to achieve fine adjustment of the lens pitch and yaw direction, realizing adjustment in five dimensions and adjusting the various pose states of the lens. In a preferred embodiment, the Z-axis translation unit 3 is connected to the column marble support plate 7 via a linear bearing 8, and the X-axis translation unit 1 is connected to the Z-axis translation unit 3 via a linear guide rail 9. The X-axis translation unit 1, Y-axis translation unit 2, and Z-axis translation unit 3 are all composed of existing straight rail sliding translation structures. The pitch and yaw adjustment unit 4 is composed of an existing movable support plate and a pitch structure and a left and right yaw structure, realizing precise two-dimensional attitude adjustment of the lens pitch and yaw. By adjusting the X-axis / Y-axis / Z-axis translation and angle, it is convenient to adjust the position and orientation of the laser interferometer lens. The structure is compact, stable, accurate, displays the position in real time, and has high operating efficiency.

[0022] X-axis translation unit 1, Y-axis translation unit 2, and Z-axis translation unit 3 are all equipped with displacement sensing modules, while pitch and yaw adjustment unit 4 is equipped with an angle sensing module. Each sensing module is signal-connected to the control module for real-time feedback of the actual displacement and attitude adjustment of each axis. The displacement sensing modules utilize Renishaw TONiC series incremental grating rulers in conjunction with RESM20 circular grating encoders. The angle sensing module is integrated into the bottom of the movable support plate of pitch and yaw adjustment unit 4, employing a two-axis MEMS capacitive tilt sensor that is sensitive to the pitch and yaw angles of the movable support plate relative to the horizontal plane. This sensor integrates a temperature compensation circuit, maintaining stable measurement accuracy within the ambient temperature range. Each sensing module is connected to the digital signal input interface of the control module via shielded cables to suppress random noise interference. The signals from the displacement and angle sensing modules are synchronously acquired within the control module using a unified clock source, ensuring strict temporal alignment of the multi-axis pose data.

[0023] The motion control card of the control module performs a zero-position reset operation on the grating ruler signal: During each power-on initialization, the control module returns each translation unit to the mechanical limit switch mounted at the end of the guide rail, using this as the absolute zero reference point. Then, based on the theoretical target pose coordinates, it drives each axis to the predetermined position. During movement, the control module calculates the deviation between the commanded position and the sensor feedback position in real time. When the deviation exceeds ±2μm, a position correction action is triggered, and a speed compensation amount is output to the servo driver via an incremental PID controller, ensuring that the following error of each axis's motion trajectory is always controlled within ±1μm. The angle sensing module synchronously feeds back the real-time changes in pitch and yaw angles. During the angle fine-tuning stage, the control module captures minute angle changes with a resolution of 0.0001°, supporting the precise execution of subsequent closed-loop control algorithms.

[0024] The control module is signal-connected to each translation unit and attitude adjustment unit. Its hardware architecture consists of existing modules such as an embedded controller, motion control module, image acquisition module, and analog input module. The embedded controller, equipped with a processor, memory, and solid-state drive, ensures deterministic scheduling of the control loop through system operation. The motion control module provides an 8-axis stepper / servo motor control interface, with each axis equipped with an independent position comparison trigger output function, which can trigger the synchronous acquisition of interferometric frames when reaching a designated position. The image acquisition module connects to the CCD camera inside the laser interferometer 5 via an interface, utilizing bandwidth to achieve lossless transmission of pixel images. The analog input module receives differential voltage signals from the angle sensing module.

[0025] The control module is configured to execute a two-stage collaborative closed-loop control strategy. The software implementation of the two-stage collaborative closed-loop control strategy within the control module is based on a state machine architecture, specifically including the following five states: idle state, first stage coarse adjustment state, inter-stage switching state, second stage fine adjustment state, and completion state.

[0026] The specific strategy is as follows: The first step involves performing XYZ three-dimensional translational large-span calibration, simultaneously calling the image preprocessing and central field-of-view feature calculation modules to preprocess the interferometric images acquired by the laser interferometer 5 and extract the central field-of-view feature parameters, and evaluate the fringe contrast and zero offset. In the second stage, after the three-dimensional translational large-span adjustment is completed, the pitch and yaw two-dimensional attitude micro-step adjustment is initiated. The same feature calculation results calculated in the first stage are reused, and a lightweight pose re-verification mechanism for the inter-stage switching point is introduced to compensate for the previous cumulative drift using zero-position offset feedback. Among them, the lightweight pose recalibration mechanism incorporates the accumulated drift amount as the initial compensation value in the attitude microstepping adjustment stage through a closed-loop control algorithm, thereby achieving seamless connection and compensation of drift between the two stages.

[0027] In summary, the transition conditions between states are triggered by the real-time values ​​of the fringe contrast V and the zero-position offset. In the idle state, the control module continuously monitors the image acquisition status of the laser interferometer 5. Once a valid interferogram frame is detected, it automatically jumps to the first coarse adjustment state. In the first coarse adjustment state, the control module drives the three-dimensional translation unit movement using a variable step-size hill-climbing algorithm. The search direction is along the direction of the fastest descent of the fringe contrast gradient. The initial search step size is set to 2 mm, and the step size is gradually reduced to 0.1 mm as the V value increases to avoid crossing the optimal peak. When the increment of the V value is less than 0.01 in three consecutive searches, it is determined that a local optimum has been reached, and then it jumps to the inter-process switching state. In the inter-process switching state, the control module performs lightweight pose recalibration, completes the calculation of the cumulative drift and the allocation of the initial compensation offset, and the entire process takes no more than 50 ms. After transitioning to the second fine-tuning state, the control module drives the three sets of threaded pairs of the pitch and yaw adjustment unit 4 in adjustment steps of 0.001°, while simultaneously updating the angle feedback data at a rate of 1kHz, forming a high-bandwidth closed loop. When the V value reaches 0.85 or higher and is maintained continuously for more than 100 frames, it transitions to the completion state and issues a calibration completion signal, while saving the final pose data to the local log file.

[0028] In a preferred embodiment, compensating for prior accumulated drift using zero-position bias feedback specifically includes: At the switching point between three-dimensional translational large-span adjustment and two-dimensional attitude micro-step adjustment, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated. The accumulated drift is decomposed into drift components in the pitch and yaw directions, which are then used as the initial compensation offset input to the drive control of the pitch and yaw adjustment unit 4 during the attitude microstepping adjustment stage. During the attitude microstepping adjustment process, the zero-position offset signal is acquired in real time, and a closed-loop control algorithm is used to dynamically correct the compensation offset until the zero-position offset converges to the allowable range.

[0029] In summary, the physical causes of cumulative drift include: thermal elongation due to temperature rise in the ball screw after prolonged operation; creep displacement of the rolling elements caused by the reciprocating motion of the crossed roller guides; cumulative step loss error during open-loop drive of the stepper motor; and minute bending deformation of the marble base under ambient temperature fluctuations. The combined effect of these factors causes the actual pose of each axis to deviate from the theoretical target pose, resulting in a five-dimensional cumulative drift D=[dx,dy,dz,dα,dβ]. T .

[0030] At the inter-process switching point, the control module performs precise drift calculation and compensation through the following steps: First, the motion control card of the control module reads the position count values ​​of the three displacement sensing modules. After resolution conversion by the grating ruler and temperature compensation correction, the actual translation P of the current X, Y, and Z axes is obtained. act =[x act ,y act ,z act ] T .

[0031] At the same time, the angle sensing module outputs the current pitch angle α. act and yaw angle β act .

[0032] Secondly, the control module retrieves the theoretical target pose vector P from the built-in memory. target =[x target ,y target ,z target ,α target ,β target ] T The theoretical target pose is precisely calibrated by a standard plane mirror before the device leaves the factory. The calibration method is as follows: a λ / 10 standard plane mirror with known surface accuracy is installed on the support plate, and the interference fringes are adjusted manually to reach the optimal state. The values ​​of each sensing module at this time are read and solidified as the theoretical target pose.

[0033] Subsequently, the difference vector ΔP=P is calculated. act -P target =[Δx,Δy,Δz,Δα,Δβ] T, where Δx, Δy, and Δz are the cumulative drift in each translation direction, and Δα and Δβ are the cumulative drift in the angular direction.

[0034] The control module decomposes the cumulative drift Δα and Δβ in the angular direction into the initial compensation bias of three sets of control sub-units using the Jacobian matrix J. The Jacobian matrix J is a 3×2 matrix, representing the mapping relationship between the positions and attitude angles of the three adjustment sub-units of the pitch and yaw adjustment unit 4. The elements of this Jacobian matrix consist of the circumferential distribution radius R of the three adjustment sub-units on the movable support plate and the included angle. , , (In this embodiment) , , ) Determined, the specific expression is: , ; , ; , In this embodiment, with R=65mm, a precise mapping relationship can be obtained from the angular drift to the axial compensation of the threaded pair, ensuring that the initial offset of the three sets of threaded pairs can synchronously and coordinately eliminate the accumulated angular drift from the previous stage.

[0035] During the attitude fine-stepping adjustment process, the control module continuously calculates the zero-position offset value at a period of 0.5 seconds. The physical meaning of the zero-position offset value is the deviation between the pixel coordinates of the center of the zero-order interference fringe within a 512×512 pixel ROI and the ROI center coordinates (256, 256). The deviation is expressed in pixels and converted using a calibration coefficient K to obtain the actual translational deviation. For example, in this embodiment, the calibration coefficient K = 2.5 μm / pixel. This zero-position offset signal reflects the residual deviation between the current lens optical axis and the interferometer optical axis.

[0036] The control module uses an incremental PID control algorithm to dynamically correct the initial compensation bias: in each control cycle, the current zero-point bias value is multiplied by the proportional coefficient K. p (For example, in this embodiment, 0.6) is used as the proportional control value, and the integral of the zero-position offset value over time is multiplied by the integral coefficient K. i (For example, in this embodiment, 0.02) is used as the integral control quantity, and the derivative of the zero-position offset value with respect to time is multiplied by the differential coefficient K. d(For example, 0.1 is used in this embodiment) as the differential control quantity. The sum of the three terms is the additional adjustment quantity of each adjustment subunit within the control cycle. This additional adjustment quantity, combined with the initial compensation bias quantity, drives the three threaded pairs together, forming a two-layer control structure that combines coarse compensation and fine correction. The judgment condition for the zero-position bias to converge to the allowable range is: within 50 consecutive control cycles, the absolute value of the zero-position bias value is always less than 1 pixel. In this embodiment, the allowable range is ±1 pixel, i.e., ±2.5μm. Once the convergence condition is met, the control module locks the current position of each axis to prevent pose fluctuations caused by subsequent environmental interference.

[0037] In the central field-of-view feature calculation module, the central field of view is a 512×512 pixel area. The feature parameters include at least the fringe contrast value and the zero offset value. The image processing pipeline of the central field-of-view feature calculation module runs in a multi-threaded parallel mode on the embedded controller of the control module. The image acquisition thread is responsible for acquiring the latest frame of the interference image from the circular buffer of the image acquisition module. The preprocessing thread is responsible for performing denoising, enhancement and edge detection operations. The feature extraction thread is responsible for calculating the contrast and zero offset. The three threads exchange data through shared memory. The coordination mechanism adopts a producer-consumer model to ensure that the processing frame rate is not less than 25fps. Image preprocessing includes image denoising, contrast enhancement, and edge detection. These are all existing image preprocessing methods that can effectively remove noise while preserving the sharp edge details of the interference fringes, avoiding the fringe contrast decay caused by excessive smoothing, effectively suppressing the excessive amplification of background noise, improving the accuracy of subsequent feature extraction, and making the output binary edge image clearly outline the contour lines of the interference fringes, providing a basis for subsequent center localization.

[0038] In the dual-process collaborative closed-loop adjustment, the translational large-span adjustment in the first process and the attitude micro-step adjustment in the second process use the same set of feature calculation results as the evaluation basis. This avoids systematic deviations introduced by differences in feature extraction between the two processes. When the first process adjustment is completed and the process jumps to the second process fine adjustment, the control module does not re-acquire images and perform calculations. Instead, it extracts the latest set of valid feature calculation results from the cache queue as the initial evaluation benchmark for the second process. Through this reuse strategy, the differences in image grayscale caused by ambient light fluctuations or laser power drift between two independent image acquisitions are avoided; the inconsistency in stripe edge positioning caused by slight fluctuations in filter parameters during the two image preprocessing processes is avoided; and the frequency locking deviation caused by different initial values ​​of the FFT spectrum peak search algorithm in the two feature calculations is avoided. The reuse of the same feature calculation results also simplifies the software architecture of the control module in terms of program implementation. It eliminates the need to reinitialize the image acquisition hardware and reload algorithm parameters during inter-program switching, reducing state switching overhead and lowering the risk of system software instability. During the reuse process, if the interval between the data timestamp in the buffer queue and the current time exceeds a threshold (i.e., the pause time is too long), the control module will automatically discard the original buffered data and re-acquire and calculate a new interferogram frame to ensure that the evaluation benchmark matches the current lens pose state.

[0039] Example 2: Based on Example 1, a five-dimensional lens adjustment method for a vertical laser interferometer is also provided, specifically including the following steps: Step 1: Place the lens under test on the movable support plate of the pitch and yaw adjustment unit 4. The lens under test is clamped using vacuum adsorption to assist in positioning. Start the laser interferometer 5 to acquire the initial interference frame. The CCD camera of the laser interferometer 5 starts to continuously acquire interference frames at a frame rate of 30fps. The image acquisition module of the control module synchronously receives each frame of image through an external trigger mode. The trigger signal is provided by the periodic position comparison output of the motion control card to ensure that the image acquisition and the movement of each axis are strictly synchronized in time. Step 2: The control module performs the first stage of adjustment: Based on the fringe contrast and zero-position offset information output by the initial interferogram frame after frame preprocessing and central field-of-view feature calculation module, the X-axis translation unit 1, Y-axis translation unit 2 and Z-axis translation unit 3 are driven to perform three-dimensional translational large-span adjustment until the fringe contrast reaches the preset threshold. The three-dimensional translational large-span adjustment is performed using a variable step size hill climbing algorithm combined with a three-dimensional spiral search strategy.

[0040] Specifically, the control module first acquires the fringe contrast V0 of the initial interferogram frame. V0 is typically between 0.1 and 0.4, depending on the randomness of the initial lens mounting pose. If V0 < 0.1, it indicates a large lens pose deviation. In this case, the control module drives three translation units to perform a large-step three-dimensional spiral scan with an amplitude of 5mm: centered on the current position, the X and Y axes gradually increase the spiral radius within ±15mm, while the Z axis simultaneously performs a small-amplitude oscillation search within ±10mm, achieving rapid coverage of the maximum possible pose range in three-dimensional space. During the spiral scan, the control module calculates the fringe contrast of each search point at a period of 0.5 seconds and records the coordinates of the search point with the largest V value up to the current time. After completing one round of spiral scan, the control module directly drives each axis to move to the coordinate position with the largest V value, completing coarse positioning.

[0041] If V0 ≥ 0.1, then directly enter the fine climbing search stage: that is, the control module sequentially performs disturbance, observation and adjustment operations on the X, Y and Z axes respectively.

[0042] Taking the X-axis as an example, the control module first applies a step size δ in the positive direction, initially δ=0.5mm, and observes the change in the V value ΔV. plus After reverting to the origin, apply a synchronous phase length δ in the negative direction and observe ΔV. minus Compare ΔV plus and ΔV minus If both are positive, move δ in the direction with the larger increment of V; if both are negative, keep the position unchanged and reduce the step size δ = δ × 0.7; if one is positive and the other is negative, move δ × sign(ΔV) in the positive direction. plus -ΔV minus ).

[0043] After completing the single-axis adjustment of the X-axis, the Y-axis and Z-axis are processed sequentially, and this process is repeated iteratively. After each round of full traversal of all three axes, it is checked whether the current V value has reached the preset threshold V. threshold =0.6, if this is reached, the first stage of adjustment is terminated and the process transitions to the inter-stage switching point; if not, the next iteration continues, but the step size δ decreases exponentially with each iteration: δ k =δ0×exp(-k / 10), where k is the iteration round number. This ensures fast search speed while avoiding crossing the peak value of V, keeping the deviation between the lens optical axis and the interferometer optical axis within the range corresponding to ensuring the fringe contrast reaches the preset threshold. During the first adjustment process, the control module continuously records the motion trajectory of each axis and saves the V value and position coordinates of each step to local storage, providing complete kinematic history data for the calculation of the cumulative drift in the subsequent third step.

[0044] Step 3: The control module performs the second stage of adjustment: At the switching point between stages, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated; the same feature calculation result in step 2 is reused to decompose the cumulative drift into drift components in the pitch and yaw directions, which are used as the initial compensation offset input to the drive control of the attitude adjustment unit to start the pitch-yaw two-dimensional attitude microstep adjustment; The trigger condition for the inter-process switching point is that the stripe contrast V has three consecutive sampled values ​​all greater than 0.6 and the fluctuation range between the maximum and minimum values ​​is less than 0.05. After the switching is triggered, the control module executes the following sub-steps: Read the actual pose of each axis, and after temperature correction and zero offset compensation, obtain the actual translation vector P. trans_act Simultaneously, the digital output of the angle sensor is read, and the actual angle vector P is obtained after digital filtering. ang_act .

[0045] The control module reads the theoretical target pose vector P from the non-volatile flash memory. targetThis vector is updated during each maintenance and calibration of the device. The difference between the actual and target vectors is calculated, where the first three components are the cumulative translational drift and the last two components are the cumulative angular drift. Considering that in actual operation, the cumulative translational drift mainly comes from the thermal expansion of the Z-axis lead screw and the creep of the Y-axis guide rail, and the cumulative angular drift mainly comes from the relaxation of the tension spring stress of the pitch and yaw adjustment unit 4 and the change in the preload of the threaded pair.

[0046] The cumulative angular drifts Δα and Δβ are mapped to the axial adjustment amounts of the three adjustment subunits using the generalized inverse of the Jacobian matrix J, and δ is calculated. base Then, the control module multiplies each value by the transmission ratio of the threaded pair, converts it into the corresponding number of stepper motor drive pulses, and temporarily stores it in the command buffer of the motion control card as the initial drive command for the second stage of fine-tuning.

[0047] The control module simultaneously sends initial pulse sequences to the three stepper motor drivers of the pitch and yaw adjustment unit 4. Each pulse sequence is planned according to an S-shaped acceleration and deceleration curve to ensure that the three sets of threaded pairs start synchronously and smoothly reach their positions, avoiding transient shock disturbances caused by asynchronous start-up. After reaching the position, the control module immediately enters the fine-tuning closed-loop control stage in step four.

[0048] Step 4: During the attitude micro-stepping adjustment process, the zero-position offset signal is acquired in real time. A closed-loop control algorithm is used to dynamically correct the compensation offset until the zero-position offset converges to the allowable range, completing the five-dimensional pose adjustment of the lens. The attitude micro-stepping adjustment adopts a dual-loop cascade control structure, where the inner loop is an angle position closed loop and the outer loop is an image feature closed loop based on the zero-position offset. The inner loop control cycle is 1ms. In each control cycle, the control module reads the current actual pitch angle and yaw angle values ​​of the pitch and yaw adjustment unit 4 through the angle sensing module, compares them with the inner loop target angle value, and calculates the angle deviation. This angle deviation is calculated by the PI controller and outputs a speed command to the driver of each stepper motor. To eliminate the risk of noise amplification, the inner loop does not introduce a differential term and uses PI control. The stepper motor drives the threaded pair to rotate, pushing the movable support plate to move in the direction of reducing the angle deviation, forming a rapid angular position lock. The function of the inner loop is to quickly suppress the interference of external vibration and mechanical disturbance on the attitude of the movable support plate and ensure the short-term stability of the angle attitude.

[0049] The outer loop control cycle is 500ms. Within each outer loop cycle, the control module calculates the center field-of-view characteristics of the current interferogram frame to obtain the zero-offset value. This zero-offset value reflects the residual deviation between the lens optical axis and the interferometer optical axis. The outer loop controller uses the zero-offset value as feedback, compares it with the target value to obtain an error signal, and outputs the angle target correction value after calculation by the PID controller. This correction value is transmitted to the inner loop as the new target angle value for the inner loop, thus realizing a complete control link of "zero-offset deviation → angle target correction → angular position closed-loop tracking". The advantage of the dual-loop cascade structure is that the outer loop responds to changes in image features at a slower sampling frequency, ensuring control stability; the inner loop quickly suppresses angle disturbances with a high sampling rate, and the two work together to achieve fast and stable convergence.

[0050] The PI controller uses the angle deviation value as input. After calculation by the proportional and integral components, it outputs a speed command: Speed ​​command = proportional coefficient × angle deviation + integral coefficient × integral of angle deviation over time. The speed command is converted from digital to analog and output as an analog voltage to the stepper motor driver. The driver then generates a pulse signal of the corresponding frequency to drive the stepper motor to rotate. The proportional component is responsible for responding to the current deviation in real time, while the integral component is responsible for eliminating static errors. When the angle deviation is close to zero, the speed command is mainly maintained by the integral term, keeping the movable support plate in its current posture and achieving zero static error angular position locking. The convergence condition is: the zero-position offset value is consistently less than the preset allowable range for multiple consecutive outer loop cycles, and the change in the control output of each axis drive motor is less than a certain proportion of the rated value, indicating that the system has entered a steady state. After the convergence condition is met, the control module locks the current posture, completing the five-dimensional pose adjustment of the lens.

[0051] The closed-loop control algorithm is either a PID control algorithm or an adaptive control algorithm, using the zero-position bias value as the control input and outputting an attitude adjustment drive signal.

[0052] In this embodiment, the closed-loop control algorithm adopts a digital PID control algorithm, whose discretized form is as follows: In each sampling period, the controller reads the current zero-point bias value as a feedback quantity and compares it with the target value to obtain the deviation signal. The deviation signal is processed in parallel by a proportional, integral, and derivative element: the proportional element multiplies the deviation signal by a proportional coefficient to generate an instantaneous control quantity proportional to the magnitude of the deviation; the integral element accumulates the values ​​of the deviation signal in each sampling period and multiplies them by an integral coefficient to generate a control quantity used to eliminate static errors; the derivative element calculates the rate of change of the deviation signal at the current moment relative to the previous moment and multiplies it by a derivative coefficient to generate a predictive control quantity to suppress system oscillations. The output quantities of the three elements are added together to obtain the total control quantity, which is output as an attitude adjustment drive signal to each stepper motor driver. The output drive signal of the PID controller is expressed as: Control quantity = Proportional coefficient × Current deviation + Integral coefficient × Cumulative deviation + Derivative coefficient × Rate of change of deviation. The three coefficients are pre-tuned according to the system characteristics to achieve a balance between fast response and stable convergence.

[0053] The calculation of cumulative drift includes: decomposing the difference between the current actual pose and the theoretical target pose of each axis into two degrees of freedom, pitch and yaw, through coordinate transformation, and obtaining the drift component value in each direction.

[0054] At the switching point between programs, the control module reads the actual position data of the three translation units (X, Y, and Z) through the displacement sensing module, and reads the actual pitch and yaw angle data of the pitch and yaw adjustment unit 4 through the angle sensing module. Simultaneously, it retrieves the pre-calibrated theoretical target pose data from memory. The actual data is subtracted from the theoretical target pose to obtain an initial difference vector, which is expressed in both the three-dimensional translation coordinate system and the two-dimensional angle coordinate system.

[0055] Since the three adjustment subunits of the pitch and yaw adjustment unit 4 are circumferentially distributed on the movable support plate, there is a definite geometric mapping relationship between the axial displacement of a single adjustment subunit and the pitch or yaw angle change of the movable support plate. To obtain the precise compensation amount required for each adjustment subunit, the cumulative drift in the angular direction needs to be decomposed into the driving direction of each adjustment subunit. The coordinate transformation decomposition is based on the spatial layout geometry of the three adjustment subunits: first, a local coordinate system of the movable support plate is established, with the center of the support plate as the origin, rotating around the X-axis in the pitch direction and around the Y-axis in the yaw direction. Based on the installation position coordinates of each adjustment subunit on the support plate, the angular drift component is converted into the axial displacement at the position of each adjustment subunit. This displacement is numerically equal to the angular drift multiplied by the distance from the adjustment subunit to the corresponding rotation axis, and then multiplied by the sign factor of the rotation direction. After the above coordinate transformation decomposition, three independent compensation amounts corresponding to the three adjustment subunits are obtained from the overall angular drift. Each compensation amount is the axial adjustment displacement value that the adjustment subunit needs to perform to eliminate the previously accumulated angular drift. This decomposition process ensures that the coordinated motion of the three control subunits can accurately synthesize the required pitch and yaw compensation movements.

[0056] After completing the attitude microstepping adjustment, if the zero-position offset does not meet the preset accuracy requirements, steps two through four are repeated. After the attitude microstepping adjustment is completed, the control module performs a final quality judgment on the current interferogram, including whether the fringe contrast reaches the threshold and whether the zero-position offset converges to the allowable range. If any of the above indicators fails to meet the preset requirements, the control module automatically starts a new round of iterations until the interferogram quality meets the detection requirements.

[0057] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A five-dimensional lens adjustment device for a vertical laser interferometer, comprising a base assembly and a laser interferometer (5), characterized in that, Also includes: The three-dimensional translational large-span adjustment hierarchy is composed of the X-axis translation unit (1), the Y-axis translation unit (2) and the Z-axis translation unit (3); The pitch and yaw adjustment unit (4) is fixedly connected to the Y-axis translation unit (2), and the two constitute a two-dimensional attitude microstep adjustment hierarchy; And a control module, which is signal-connected to each translation unit and attitude adjustment unit, wherein the control module is configured to execute a two-way collaborative closed-loop control strategy: In the first stage, XYZ three-dimensional translational large-span adjustment is performed, and the image frame preprocessing and central field of view feature calculation module are called simultaneously to preprocess the interference image frame obtained by the laser interferometer (5) and extract the central field of view feature parameters to evaluate the fringe contrast and zero position offset. In the second stage, after the three-dimensional translational large-span adjustment is completed, the pitch and yaw two-dimensional attitude micro-step adjustment is initiated. The same feature calculation results calculated in the first stage are reused, and a lightweight pose re-verification mechanism for the inter-stage switching point is introduced to compensate for the previous cumulative drift using zero-position offset feedback. The lightweight pose recalibration mechanism incorporates the accumulated drift amount as the initial compensation value for the attitude microstepping adjustment stage through a closed-loop control algorithm, thereby achieving seamless drift connection and compensation between the two stages.

2. The lens five-dimensional adjustment device for a vertical laser interferometer according to claim 1, characterized in that, The method of using zero-position bias feedback to compensate for preceding cumulative drift specifically includes: At the switching point between three-dimensional translational large-span adjustment and two-dimensional attitude micro-step adjustment, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated. The accumulated drift is decomposed into drift components in the pitch and yaw directions, and is used as the initial compensation offset for the attitude microstepping adjustment stage, which is then input into the drive control of the pitch and yaw adjustment unit (4). During the attitude microstepping adjustment process, the zero-position offset signal is acquired in real time, and a closed-loop control algorithm is used to dynamically correct the compensation offset until the zero-position offset converges to the allowable range.

3. The lens five-dimensional adjustment device for a vertical laser interferometer according to claim 1, characterized in that, In the central field-of-view feature calculation module, the central field of view is a 512×512 pixel area, and the feature parameters include at least the stripe contrast value and the zero offset value; the frame preprocessing includes image denoising, contrast enhancement and edge detection.

4. The lens five-dimensional adjustment device for a vertical laser interferometer according to claim 1, characterized in that, The pitch and yaw adjustment unit (4) is used to achieve precise two-dimensional attitude adjustment of the lens in pitch and yaw.

5. The lens five-dimensional adjustment device for a vertical laser interferometer according to claim 1, characterized in that, The base assembly includes a flat marble plate (6) and a column marble support plate (7) vertically fixed on the flat marble plate (6). The laser interferometer (5) is fixed on the flat marble plate (6). The laser interferometer (5) is used to emit a measurement wavefront and receive the interference pattern formed after returning through the lens under test. The Z-axis translation unit (3) is connected to the column marble support plate (7). The Z-axis translation unit (3) is connected to the base assembly, the X-axis translation unit (1) is connected to the Z-axis translation unit (3), and the Y-axis translation unit (2) is connected to the X-axis translation unit (1). The X-axis translation unit (1), Y-axis translation unit (2) and Z-axis translation unit (3) are all equipped with displacement sensing modules, and the pitch and yaw adjustment unit (4) is equipped with an angle sensing module. Each sensing module is connected to the control module for real-time feedback of the actual displacement and attitude adjustment of each axis.

6. The lens five-dimensional adjustment device for a vertical laser interferometer according to claim 1, characterized in that, In the dual-path collaborative closed-loop adjustment, the translational large-span adjustment of the first path and the attitude micro-step adjustment of the second path use the same set of feature calculation results as the evaluation basis to avoid systematic deviations introduced by the difference in feature extraction between the two paths.

7. A method for five-dimensional lens adjustment of a vertical laser interferometer, characterized in that, The five-dimensional lens adjustment method for the vertical laser interferometer includes the five-dimensional lens adjustment device for the vertical laser interferometer as described in any one of claims 1-6, specifically comprising the following steps: Step 1: Place the lens to be tested on the movable support plate of the pitch and yaw adjustment unit (4), start the laser interferometer (5), and obtain the initial interference frame; Step 2: The control module performs the first stage of adjustment: Based on the fringe contrast and zero offset information output by the frame preprocessing and central field-of-view feature calculation module after the initial interferogram frame is processed, the X-axis translation unit (1), Y-axis translation unit (2) and Z-axis translation unit (3) are driven to perform three-dimensional translation large-span adjustment until the fringe contrast reaches the preset threshold. Step 3: The control module performs the second stage of adjustment: At the switching point between stages, the actual pose data of each axis is read through the displacement sensing module and the angle sensing module, compared with the theoretical target pose, and the cumulative drift is calculated; the same feature calculation result in step 2 is reused to decompose the cumulative drift into drift components in the pitch and yaw directions, which are used as the initial compensation offset input to the drive control of the attitude adjustment unit to start the pitch-yaw two-dimensional attitude microstep adjustment; Step 4: During the attitude microstep adjustment process, the zero-position offset signal is acquired in real time, and the compensation offset is dynamically corrected using a closed-loop control algorithm until the zero-position offset converges to the allowable range, thus completing the five-dimensional pose adjustment of the lens.

8. The five-dimensional lens adjustment method for a vertical laser interferometer according to claim 7, characterized in that, The closed-loop control algorithm is a PID control algorithm or an adaptive control algorithm, which uses the zero-position bias value as the control input and outputs an attitude adjustment drive signal.

9. The five-dimensional lens adjustment method for a vertical laser interferometer according to claim 7, characterized in that, The calculation of the cumulative drift includes: decomposing the difference between the current actual pose and the theoretical target pose of each axis into two degrees of freedom, pitch and yaw, through coordinate transformation, and obtaining the drift component value in each direction.

10. The five-dimensional lens adjustment method for a vertical laser interferometer according to claim 7, characterized in that, It also includes iterative optimization steps: after completing the attitude microstep adjustment, if the zero-position offset does not meet the preset accuracy requirements, repeat steps two to four until the quality of the interferogram meets the detection requirements.

Images