Intelligent calibration method and system for a contact displacement sensor

Abstract

The application discloses an intelligent calibration method and system of a contact type displacement sensor, and relates to the technical field of micro-positioning control.The method sets a posture stability threshold F1 in the preliminary posture evaluation stage, which is used for identifying whether to trigger a contact direction inversion compensation mechanism, and further calculates a calibration accuracy result index Delta z based on a vertical direction displacement value Zcorr after performing projection compensation, and introduces an error threshold F2 to perform secondary comparison and evaluation on the compensation result.If the calibration accuracy result index Delta z exceeds the error threshold F2, the automatic iteration process will be entered.The method introduces a double-threshold comparison and evaluation judgment mechanism, improves the error identification ability and system stability in the calibration process, ensures that the final output result has high consistency and reliability, and avoids the problems of insufficient one-time compensation accuracy and platform misadjustment in the traditional scheme.

Description

Technical Field

[0001] This invention relates to the field of micro-positioning control technology, specifically to an intelligent calibration method and system for a contact displacement sensor. Background Technology

[0002] Intelligent calibration of contact displacement sensors involves the fields of precision displacement measurement and micro-positioning control, and is a key component of intelligent manufacturing and opto-mechatronics systems. Furthermore, it falls under the category of displacement sensing and calibration technology for micro-nano optical focusing platforms. In this specific application, contact displacement sensors are widely integrated into miniature focusing platforms, microscopic assembly systems, and laser alignment systems for precise height sensing and attitude adjustment of target surfaces. Especially in high-precision scenarios, such as laser autofocus, MEMS packaging, and microlens bonding processes, the measurement accuracy of contact displacement sensors directly impacts the platform's focusing accuracy and product assembly quality.

[0003] Currently, most contact displacement sensors employ a single-axis signal output model, meaning they only acquire contact displacement signals in the vertical direction. However, during actual installation and use, minor errors in assembly posture, uneven contact surfaces, or slight platform offsets often result in the sensor probe not always contacting the target surface perpendicularly. This non-ideal contact angle causes an angle between the measurement path and the Z-axis, leading to a sensor output "displacement reading" that is higher than the actual vertical displacement value, resulting in a systematic error. Furthermore, inconsistent probe directions at different points during multi-point sampling on the platform exacerbate measurement differences between different sampling points, ultimately affecting the overall consistency and calibrability of the platform's focusing surface.

[0004] The aforementioned deviations primarily stem from a slight tilt angle exhibited by the probe during contact. Traditional algorithms, however, only assume the contact path is equivalent to the Z-axis displacement along the vertical axis, neglecting this directional deviation. While small, this angle is sufficient to introduce error in micrometer-level measurement systems. Without modeling and correction, it can lead to focus blurring or imaging deviations in autofocus systems; misalignment or interference in micro-assembly platforms; and misjudgment of conformity by calibration platforms, amplifying quality risks. Furthermore, inconsistent superposition of multiple probe directional errors can enhance sampling surface unevenness, further impacting control strategies due to platform attitude misjudgments. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent calibration method and system for contact displacement sensors, solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps:

[0007] S1. Set up a five-point sampling matrix on the micro-optical focusing platform, and set up acquisition devices in the five-point sampling matrix to acquire three-dimensional offset data in real time;

[0008] S2. Transmit the local offset vector geometrically to the synchronous control system, and preprocess the set of local offset vectors to obtain the reference normalized feature vector d.

[0009] S3. Based on the benchmark normalized feature vector d, calculate and output the preliminary attitude evaluation index Satt, perform a preliminary comparative evaluation, and trigger the contact direction inversion compensation mechanism based on the preliminary comparative evaluation results.

[0010] S4. After triggering the contact direction inversion compensation mechanism, define the actual contact direction unit vector for each point, and calculate and output the calibrated vertical displacement value Zcorr based on the contact direction unit vector to perform projection compensation.

[0011] S5. Based on the vertical displacement value Zcorr after calibration, calculate and output the calibration accuracy result index △z, and perform a secondary comparative evaluation and analysis to assess the reliability of the calibration.

[0012] Preferably, S1 includes S11 and S12;

[0013] S11. In the micro-optical focusing platform, a five-point sampling matrix is ​​set up, which includes sampling points P0, P1, P2, P3 and P4, to identify the attitude of the micro-optical focusing platform.

[0014] Wherein, P0 sampling point represents the center point of the micro-optical focusing platform, P1 sampling point represents the area below the center point of the micro-optical focusing platform, P2 sampling point represents the area above the center point of the micro-optical focusing platform, P3 sampling point represents the area to the right of the center point of the micro-optical focusing platform, and P4 sampling point represents the area to the left of the center point of the micro-optical focusing platform.

[0015] S12. Install a contact displacement sensor probe at each point in the five-point sampling matrix. The contact displacement sensor probe makes physical contact with the micro-optical focusing platform and collects the three-dimensional offset data of each point in the optical focusing platform in real time.

[0016] The three-dimensional offset data includes the horizontal axis offset X, the vertical axis offset Y, and the vertical axis offset Z.

[0017] Preferably, S2 includes S21 and S22;

[0018] S21. Construct a synchronous control system and set up a wireless communication network to wirelessly connect the communication module of the contact displacement sensor probe with the synchronous control system and transmit the real-time acquired three-dimensional offset data to the synchronous control system.

[0019] S22. In the synchronous control system, receive three-dimensional offset data in real time, preprocess the three-dimensional offset data, and obtain the reference normalized feature vector d at the j-th contact of the i-th sampling point. ij ;

[0020] The reference normalized feature vector d at the j-th contact of the i-th sampling point ij Including the horizontal axis offset X at the j-th contact of the i-th sampling point. ij The vertical axis offset Y at the j-th contact of the i-th sampling point ij Vertical axis offset Z at the j-th contact with the i-th sampling point ij ;

[0021] The preprocessing includes outlier identification and removal, noise smoothing, and coordinate normalization and benchmark alignment.

[0022] The outlier identification and removal process calculates the mean and standard deviation of all directions in each sample of 3D offset data, and calculates the difference between the direction and mean of the real-time acquired 3D offset data. If the output result is more than 3 times the standard deviation, it is identified as an outlier and removed.

[0023] The noise smoothing is achieved by using a Savitzky-Golay filter to fit a polynomial within a sliding window to smooth the 3D offset data after outlier identification and removal.

[0024] The coordinate normalization and benchmark alignment are achieved by setting the mean of the three-dimensional offset data of the acquisition point P0 to a zero vector and performing differential compensation on all three-dimensional offset data.

[0025] Preferably, S3 includes S31 and S32;

[0026] S31, Based on the benchmark normalized feature vector d ij Construct a spatial perturbation covariance tensor to identify the baseline normalized eigenvectors d in each of the five-point sampling matrices. ij The geometric deviation trend is obtained by taking the maximum eigenvalue of the spatial perturbation covariance tensor, obtaining the preliminary attitude evaluation index Satt, and identifying the distribution intensity in the strongest perturbation direction in the entire contact displacement sensor probe perturbation vector field.

[0027] Preferably, in step S32, a preliminary comparative evaluation is performed based on the acquired preliminary attitude evaluation index Satt. The preliminary comparative evaluation is performed by sampling multiple sets of similar micro-optical focusing platforms, measuring the preliminary attitude evaluation index Satt under normal operating conditions, and calculating the mean and three times the standard deviation to obtain the attitude stability threshold F1. The attitude stability threshold is then compared with the real-time acquired attitude evaluation index Satt to analyze the attitude stability of the current contact displacement sensor probe, and the contact direction inversion compensation mechanism is triggered based on the evaluation results. The specific evaluation content is as follows.

[0028] When the initial attitude evaluation index Satt ≤ attitude stability threshold F1, it indicates that the attitude is stable under normal disturbance, and the calibration is completed.

[0029] When the initial attitude assessment index Satt > attitude stability threshold F1, it indicates that the vertical axis offset Z is tilted, and the contact direction inversion compensation mechanism is triggered.

[0030] Preferably, S4 includes S41 and S42;

[0031] S41. After the initial comparative evaluation triggers the contact direction inversion compensation mechanism, the actual contact direction unit vector of all sampling points is defined, and then the total displacement length Zraw of the i-th sampling point is collected by the contact displacement sensor probe for the original vertical axis offset Z reading of each contact point. i .

[0032] Preferably, S42, based on the actual contact direction unit vector and the total displacement length Zraw of the i-th sampling point. i The system performs projection compensation for the vertical axis offset Z, outputs the calibrated vertical displacement value Zcorr after projection compensation, and uses the synchronous control system to deduce the actual direction vector of the touch displacement sensor probe based on the calibrated vertical displacement value Zcorr.

[0033] Preferably, S5 includes S51 and S52;

[0034] S51. After performing projection compensation, extract the calibrated vertical displacement value Zcorr and compare it with the calibrated vertical displacement average value Zcorr. avg The calibration accuracy result index △z is calculated and output to measure the consistency of the vertical axis offset Z residual between the sampling points of the five-point sampling matrix of the low-light focusing platform.

[0035] Preferably, in step S52, after outputting the calibration accuracy result index △z, a second comparison evaluation is performed. The second comparison evaluation is performed by setting an error threshold F2 based on the surface flatness error allowed by the micro-light focusing platform. The real-time acquired calibration accuracy result index △z is compared with the error threshold F2 to determine the error status of the micro-light focusing platform after projection compensation. An iterative mechanism is then executed based on the second comparison evaluation result. The specific evaluation content is as follows.

[0036] When the calibration accuracy result Δz ≤ error threshold F2, it indicates that the results are highly consistent after projection compensation, and the calibration is successful.

[0037] When the calibration accuracy result index Δz > error threshold F2, it indicates that there is an error in the height after projection compensation. At this time, the iteration mechanism is activated, and the contact direction inversion compensation mechanism is triggered again based on the current projection compensation.

[0038] An intelligent calibration system for a contact displacement sensor includes a sampling point sampling module, a data transmission processing module, an attitude vector offset analysis module, a projection compensation module, and a calibration accuracy result analysis module.

[0039] The sampling point sampling module sets up a five-point sampling matrix on the micro-optical focusing platform and sets up acquisition devices in the five-point sampling matrix to acquire three-dimensional offset data in real time.

[0040] The data transmission processing module transmits the local offset vector set to the synchronization control system and preprocesses the local offset vector set to obtain the reference normalized feature vector d.

[0041] The attitude vector offset analysis module calculates and outputs the preliminary attitude evaluation index Satt based on the benchmark normalized feature vector d, performs a preliminary comparative evaluation, and triggers the contact direction inversion compensation mechanism based on the preliminary comparative evaluation results.

[0042] The projection compensation module defines the actual contact direction unit vector for each point after triggering the contact direction inversion compensation mechanism, and calculates and outputs the calibrated vertical displacement value Zcorr based on the contact direction unit vector to perform projection compensation.

[0043] The calibration accuracy result analysis module calculates and outputs the calibration accuracy result index Δz based on the vertical displacement value Zcorr after calibration, and performs a secondary comparative evaluation to analyze the reliability of the calibration.

[0044] This invention provides an intelligent calibration method and system for contact displacement sensors. It offers the following advantages:

[0045] (1) This method constructs a five-point sampling matrix in a micro-optical focusing platform to collect three-dimensional offset data of each point in real time. Then, the offset vector set is preprocessed in the synchronous control system to obtain the reference normalized feature vector. Based on this, a spatial perturbation covariance tensor is constructed to output the preliminary attitude evaluation index Satt and identify the attitude perturbation intensity in the probe direction. When the attitude is determined to be unstable, the system automatically triggers the contact direction inversion compensation mechanism, constructs the contact direction unit vector of each sampling point, and performs direction projection compensation on the original Z-direction path reading to output a more realistic Zcorr. The above process can effectively identify and compensate for the systematic deviation caused by the non-perpendicular contact of the probe, improve the vertical direction measurement accuracy of the final calibration result, and effectively enhance the sensor's adaptability to contact attitude perturbations on a multi-degree-of-freedom mounting platform.

[0046] (2) This method not only sets an attitude stability threshold F1 in the initial attitude assessment stage to identify whether the contact direction inversion compensation mechanism is triggered, but also, after performing projection compensation, further calculates the calibration accuracy result index Δz based on the calibrated vertical displacement value Zcorr, and introduces an error threshold F2 to perform a secondary comparative evaluation of the compensation result. If the calibration accuracy result index Δz exceeds the error threshold F2, it will enter the automatic iteration process, re-triggering the direction unit vector estimation and Z-direction projection compensation, thereby constructing a calibration evaluation closed loop with adaptive iteration capability. By introducing a dual threshold comparison judgment mechanism, this method improves the error identification capability and system stability during the calibration process, ensuring that the final output result has high consistency and credibility, and avoiding problems such as insufficient one-time compensation accuracy and platform misadjustment in traditional schemes.

[0047] (3) The intelligent calibration method of this approach integrates outlier identification and removal, Savitzky-Golay filtering and smoothing algorithm, and coordinate normalization and compensation mechanism based on the center sampling point P0 during the three-dimensional offset data processing stage. This effectively eliminates abnormal fluctuations and random errors in the sensor data, maintaining the trend continuity and spatial consistency of the data. Simultaneously, by uniformly remapping the five-point offset vectors to a relative disturbance space referenced by the center point, attitude normalization and standardization across multiple platforms can be achieved, improving the universality and comparability between systems. This method realizes an integrated calibration control closed loop from data acquisition to processing, and from discrimination to feedback, effectively enhancing the flexibility of platform deployment and the adaptability of the algorithm to different platform environments. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the intelligent calibration method for a contact displacement sensor according to the present invention.

[0049] Figure 2 This is a schematic diagram of the steps of an intelligent calibration system for a contact displacement sensor according to the present invention;

[0050] Figure 3 This is a schematic diagram of the evaluation process for an intelligent calibration method for a contact displacement sensor according to the present invention. Detailed Implementation

[0051] 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.

[0052] Example 1

[0053] Please see Figure 1 and Figure 3 This invention provides an intelligent calibration method for a contact displacement sensor. To achieve the above objectives, this invention employs the following technical solution, comprising the following steps:

[0054] S1. Set up a five-point sampling matrix on the micro-optical focusing platform, and set up acquisition devices in the five-point sampling matrix to acquire three-dimensional offset data in real time;

[0055] S2. Transmit the local offset vector geometrically to the synchronous control system, and preprocess the set of local offset vectors to obtain the reference normalized feature vector d.

[0056] S3. Based on the benchmark normalized feature vector d, calculate and output the preliminary attitude evaluation index Satt, perform a preliminary comparative evaluation, and trigger the contact direction inversion compensation mechanism based on the preliminary comparative evaluation results.

[0057] S4. After triggering the contact direction inversion compensation mechanism, define the actual contact direction unit vector for each point, and calculate and output the calibrated vertical displacement value Zcorr based on the contact direction unit vector to perform projection compensation.

[0058] S5. Based on the vertical displacement value Zcorr after calibration, calculate and output the calibration accuracy result index △z, and perform a secondary comparative evaluation and analysis to assess the reliability of the calibration.

[0059] In this embodiment, the method sets up a five-point sampling matrix on a micro-optical focusing platform to collect three-dimensional offset data of each point in real time. The collected local offset vectors are geometrically transmitted to the synchronous control system, where data preprocessing operations, including outlier identification, noise smoothing, and coordinate normalization, are performed to obtain a benchmark normalized feature vector set. Based on this, a spatial perturbation covariance tensor is constructed, and a preliminary attitude evaluation index Satt is output. This index is compared with an attitude stability threshold to determine whether the probe has directional tilt. If it does, a contact direction inversion compensation mechanism is automatically triggered. The actual contact direction of the probe is derived through unit vector reconstruction, and the original Z-axis measurement value is corrected based on a triangular projection model to output the true calibrated vertical displacement value Zcorr. Finally, based on the calibrated vertical displacement value Zcorr of multiple sampling points, the calibration accuracy result index Δz is calculated to determine the consistency and reliability of the compensation result. If the result does not meet the standard, iterative compensation is performed again. This method comprehensively constructs the entire process from 3D offset perception and attitude trend judgment to orientation projection compensation and residual closed-loop verification. It not only achieves intelligent identification and adaptive compensation for probe attitude disturbances but also enhances the self-calibration capability of the micro-platform under multi-degree-of-freedom pose conditions. It effectively eliminates measurement errors caused by contact angle deviations, improving the platform's focusing accuracy, consistency, and the stability and robustness of the sensor system. It is particularly suitable for deployment in compact, high-precision optical platforms or micro-assembly platforms sensitive to multi-pose errors.

[0060] Example 2

[0061] Please see Figure 1 Specifically: S1 includes S11 and S12;

[0062] S11. In the micro-optical focusing platform, a five-point sampling matrix is ​​set up, which includes sampling points P0, P1, P2, P3 and P4, to identify the attitude of the micro-optical focusing platform.

[0063] Wherein, P0 sampling point represents the center point of the micro-optical focusing platform, P1 sampling point represents the area below the center point of the micro-optical focusing platform, P2 sampling point represents the area above the center point of the micro-optical focusing platform, P3 sampling point represents the area to the right of the center point of the micro-optical focusing platform, and P4 sampling point represents the area to the left of the center point of the micro-optical focusing platform.

[0064] S12. Install a contact displacement sensor probe at each point in the five-point sampling matrix. The contact displacement sensor probe makes physical contact with the micro-optical focusing platform and collects the three-dimensional offset data of each point in the optical focusing platform in real time.

[0065] The 3D offset data includes the horizontal offset X, the vertical offset Y, and the vertical offset Z.

[0066] In this embodiment, the method, through step S1, first sets up a five-point sampling matrix on the micro-optical focusing platform, including the center point P0, upper and lower points P1 / P2, and left and right points P3 / P4, so that the sampling points can cover the key geometric areas of the platform, thereby realizing spatial perception of the overall attitude state of the platform. This five-point sampling matrix arrangement can effectively identify whether the platform has local warping, tilting, or nonlinear deformation trends. Subsequently, in step S12, contact displacement sensor probes are arranged at the five sampling point positions, so that each point has physical contact with the micro-optical focusing platform, enabling real-time acquisition of minute offsets in the horizontal axis (X), vertical axis (Y), and vertical axis (Z) directions at the corresponding positions under actual working conditions. This implementation not only establishes a symmetrical measurement architecture with the center point as a reference but also constructs a preliminary three-dimensional attitude information sampling framework, providing a high-quality data foundation for subsequent attitude modeling, orientation compensation, and error assessment. Therefore, this step enhances the integrity of displacement data and the ability to identify spatial distribution features, while ensuring the accuracy and global adaptability of subsequent calibration and compensation algorithms. This, in turn, improves the overall attitude recognition accuracy, focusing stability, robustness, and scalability of the displacement measurement system.

[0067] Example 3

[0068] Please see Figure 1 Specifically: S2 includes S21 and S22;

[0069] S21. Construct a synchronous control system and set up a wireless communication network to wirelessly connect the communication module of the contact displacement sensor probe with the synchronous control system and transmit the real-time acquired three-dimensional offset data to the synchronous control system.

[0070] S22. In the synchronous control system, receive three-dimensional offset data in real time, preprocess the three-dimensional offset data, and obtain the reference normalized feature vector d at the j-th contact of the i-th sampling point. ij ;

[0071] The reference normalized feature vector d at the j-th contact point of the i-th sampling point ij Including the horizontal axis offset X at the j-th contact of the i-th sampling point. ij The vertical axis offset Y at the j-th contact of the i-th sampling point ij Vertical axis offset Z at the j-th contact with the i-th sampling point ij ;

[0072] Preprocessing includes outlier identification and removal, noise smoothing, and coordinate normalization and baseline alignment;

[0073] Outlier identification and removal involves calculating the mean and standard deviation of all directions in each sample of 3D offset data, and then calculating the difference between the real-time acquired 3D offset data and the mean. If the output value is more than 3 times the standard deviation, it is identified as an outlier and removed.

[0074] Noise smoothing uses a Savitzky-Golay filter to smooth the 3D offset data after outlier identification and removal, and fits a polynomial within a sliding window. This method is suitable for maintaining data trends under small perturbations.

[0075] Coordinate normalization and benchmark alignment are achieved by setting the mean of the three-dimensional offset data of the acquisition point P0 to a zero vector and performing differential compensation on all three-dimensional offset data. This repositions the offset vectors of all points to a relative perturbation space with the center point as the reference.

[0076] In this embodiment, the method achieves efficient transmission and high-quality processing of 3D offset data from the acquisition end to the analysis end through the coordinated implementation of steps S2 and S22 in step S2. In S21, by constructing a synchronous control system and setting up a wireless communication network, the contact displacement sensor probes installed at each position of the five-point sampling matrix are wirelessly connected to the control system, ensuring that the 3D offset data of each sampling point can be transmitted to the central processing module in real time and stably. This design simplifies the system wiring structure and improves the platform deployment flexibility. Next, in S22, a series of preprocessing operations are performed on the acquired 3D offset data, including outlier identification and removal, perturbation smoothing implemented by the Savitzky-Golay filter, and coordinate normalization and alignment based on the center sampling point P0. Each set of preprocessed data is structured into a reference normalized feature vector d at the j-th contact of the i-th sampling point. ij This method fully preserves the micro-displacement characteristics of the data in the three directions: horizontal offset (X), vertical offset (Y), and vertical offset (Z). Through centering compensation, each offset vector is mapped to a relative perturbation coordinate system referenced to the center point. This implementation significantly improves the reliability, consistency, and comparability of the data, laying a high-precision computational foundation for subsequent attitude modeling, perturbation analysis, and compensation mechanisms. Overall, this step achieves intelligent acquisition and standardized data processing, effectively improving the system's data robustness, attitude discrimination accuracy, and the platform's scalable communication capabilities.

[0077] Example 4

[0078] Please see Figure 1 and Figure 3 Specifically: S3 includes S31 and S32;

[0079] S31, Based on the benchmark normalized feature vector d ijConstruct a spatial perturbation covariance tensor to identify the baseline normalized eigenvectors d in each of the five-point sampling matrices. ij The geometric deviation trend is obtained by taking the maximum eigenvalue of the spatial perturbation covariance tensor, obtaining the preliminary attitude evaluation index Satt, and identifying the distribution intensity in the strongest perturbation direction in the entire contact displacement sensor probe perturbation vector field.

[0080] The initial attitude assessment index Satt is calculated and output using the following algorithm formula;

[0081]

[0082] In the formula, λmax represents the principal characteristic of the tensor, represents the intensity of the direction of maximum perturbation, N represents the total number of sampling points in the five-point sampling matrix, and M represents the number of times each sampling point is repeated. Let represent the mean of the normalized eigenvectors of the baseline, and T represent the vector transpose function, used to transform a 3×1 column vector into a 1×3 row vector, which is then used for outer product operations to construct a 3×3 matrix.

[0083] The spatial perturbation covariance tensor is used to normalize the eigenvector d of each reference. ij Calculate its offset relative to the mean vector, then perform tensor external accumulation and normalization. This is the classic method for constructing the sample covariance matrix. In a three-dimensional scene, this is the perturbation tensor, which describes the distribution trend, directionality, and concentration of multiple contact offset vectors in space. Essentially, it is a statistical feature of the "distribution structure" of a three-dimensional perturbation field. It can be understood as the distribution of offset energy in three-dimensional space in the X, Y, and Z directions, as well as whether the correlation between each axis is linked to the offset.

[0084] The ratio is the relative offset vector of the j-th sample, representing the degree to which the sample deviates from the overall trend.

[0085] S32. Based on the obtained preliminary attitude evaluation index Satt, a preliminary comparative evaluation is conducted. The preliminary comparative evaluation is carried out by sampling multiple sets of similar micro-optical focusing platforms, measuring the preliminary attitude evaluation index Satt under normal operating conditions, and calculating the mean and three times the standard deviation to obtain the attitude stability threshold F1. The attitude stability threshold is then compared with the real-time acquired attitude evaluation index Satt to analyze the attitude stability of the current contact displacement sensor probe, and the contact direction inversion compensation mechanism is triggered based on the evaluation results. The specific evaluation content is as follows.

[0086] When the initial attitude evaluation index Satt ≤ attitude stability threshold F1, it indicates that the attitude is stable under normal disturbance, and the calibration is completed.

[0087] When the initial attitude assessment index Satt > attitude stability threshold F1, it indicates that the vertical axis offset Z is tilted, that is, the abnormal attitude is tilted due to disturbance. At this time, the contact direction inversion compensation mechanism is triggered.

[0088] In this embodiment, the method constructs an attitude evaluation and compensation determination mechanism based on perturbation feature modeling through the collaborative processing of S31 and S32. In S31, the reference normalized feature vector d at the j-th contact of the i-th sampling point obtained in the aforementioned steps is used. ij A three-dimensional spatial perturbation covariance tensor is constructed. By calculating the outer product of the deviations of each offset vector relative to the overall mean vector, a 3×3 symmetric tensor is generated, and its principal eigenvalues ​​are extracted to output the preliminary attitude evaluation index Satt. This index quantifies the geometric concentration of the strongest perturbation direction of the probe throughout the contact process, reflecting the strength and directionality of the overall attitude perturbation trend, and possessing high spatial distribution sensitivity and direction recognition capability. Subsequently, in S32, an evaluation benchmark is established based on the preliminary attitude evaluation index Satt sample data under historical normal working conditions. Its statistical mean and three times the standard deviation are extracted to construct the attitude stability threshold F1, and the current preliminary attitude evaluation index Satt is compared with F1 in real time. When Satt exceeds the attitude stability threshold F1, it is determined that there is an abnormal directional deviation in the current probe attitude, and the contact direction inversion compensation mechanism is automatically triggered. This implementation method offers significant advantages: it not only achieves quantifiable modeling and feature extraction of multi-point perturbation modes, but also constructs a highly adaptive stability discrimination logic through standardized evaluation methods. This effectively identifies potential attitude tilt states, avoids compensation inaccuracies or calibration failures due to error accumulation, and significantly enhances the system's perception depth of attitude perturbations, intelligent response capabilities, and error feedforward control level.

[0089] Example 5

[0090] Please see Figure 1 Specifically: S4 includes S41 and S42;

[0091] S41. After the initial comparative evaluation triggers the contact direction inversion compensation mechanism, the actual contact direction unit vector of all sampling points is defined, and then the total displacement length Zraw of the i-th sampling point is collected by the contact displacement sensor probe for the original vertical axis offset Z reading of each contact point. i .

[0092] S42, Based on the actual contact direction unit vector and the total displacement length Zraw of the i-th sampling point. iThe projection compensation of the vertical axis offset Z is performed, and the calibrated vertical displacement value Zcorr after projection compensation is output. The actual direction vector of the touch displacement sensor probe is deduced in reverse by the synchronous control system based on the calibrated vertical displacement value Zcorr.

[0093] The vertical displacement value Zcorr after calibration is calculated and output using the following algorithm formula;

[0094] Zcorr i =Zraw i ·(n i ·e Z );

[0095] In the formula, Zcoor i Represents the vertical displacement value of the i-th sampling point after calibration, n i Let e ​​represent the actual contact direction unit vector at the i-th sampling point. Z Z represents the ideal vertical downward unit vector of the vertical axis offset Z, taking the value [0,0,1].

[0096] (n i ·e Z ) represents the cosine of the angle between the probe direction of the tactile displacement sensor and the offset Z of the vertical axis, which is the projection coefficient of this direction on the Z-axis;

[0097] The physical meaning of the formula: The probe is not always strictly perpendicular to the vertical axis by the offset Z; if the probe is slightly tilted, the total displacement Zraw of the original reading at the i-th sampling point... i The path length along the probe direction will overestimate the actual vertical displacement. What we need to know is how much the actual vertical axis offset Z has moved. For example, if you press a button at an angle with a pen, the pen moves 5mm, but the vertical direction only moves 4.9mm. We need to use projection to transform the oblique path into a vertical path.

[0098] In this embodiment, step S4 of the method is implemented through the cooperation of S41 and S42, constructing a high-precision calibration compensation mechanism based on orientation reconstruction and geometric projection. Specifically, in S41, when the preliminary attitude evaluation result Satt exceeds the stability threshold F1, the contact orientation inversion compensation mechanism is automatically triggered. By normalizing the three-dimensional offset vector of each sampling point in multiple contacts, its actual contact orientation unit vector is calculated and defined, thereby accurately describing the orientation and attitude of the probe during the actual contact process. Simultaneously, the total displacement length Zraw of the i-th sampling point during the contact process is collected. i This constitutes a complete vector model of the probe's actual contact path. Then, in S42, based on the principle of vector projection, the actual contact direction unit vector and the total displacement length Zraw of the i-th sampling point are combined.i The ideal vertical downward unit vector e with offset Z from the vertical axis. Z That is, by performing dot product projection on [0,0,1], the calibrated vertical displacement value Zcorr of the i-th sampling point is obtained. i This accurately reconstructs the true indentation of the probe along the vertical offset Z direction. This projection compensation method is essentially equivalent to converting the tilted path into its orthogonal projection along the vertical offset Z direction, effectively eliminating the systematic overestimation error caused by probe angle deviation. Furthermore, the calibrated vertical displacement value Zcorr at the i-th sampling point... i Not only does it possess high reliability, but it can also assist in obtaining the probe attitude offset trend through reverse calculation, further enriching the system's ability to perceive platform attitude changes. This implementation significantly improves the sensor's geometric adaptability and compensation accuracy to attitude disturbances, and is particularly suitable for micro-optical focusing platforms with compact structures, uncontrollable installation angles, or extreme sensitivity to micron-level errors, enabling high-precision adaptive calibration without changing the hardware structure.

[0099] Example 6

[0100] Please see Figure 1 and Figure 3 Specifically: S5 includes S51 and S52;

[0101] S51. After performing projection compensation, extract the calibrated vertical displacement value Zcorr and compare it with the calibrated vertical displacement average value Zcorr. avg The calibration accuracy result index △z is calculated and output to measure the consistency of the vertical axis offset Z residual between the sampling points of the five-point sampling matrix of the low-light focusing platform.

[0102] The calibration accuracy result index Δz is calculated and output using the following algorithm formula;

[0103]

[0104] The calculation logic of the formula is as follows: the degree of consistency of error among the five points is calculated by the mean absolute deviation.

[0105] S52. After outputting the calibration accuracy result index △z, a second comparison evaluation is performed. The second comparison evaluation sets an error threshold F2 based on the surface flatness error allowed by the micro-light focusing platform. The real-time acquired calibration accuracy result index △z is compared with the error threshold F2 to determine the error of the micro-light focusing platform after projection compensation. An iterative mechanism is executed based on the second comparison evaluation result. The specific evaluation content is as follows.

[0106] When the calibration accuracy result Δz ≤ error threshold F2, it indicates that the results are highly consistent after projection compensation, and the calibration is successful.

[0107] When the calibration accuracy result index Δz > error threshold F2, it indicates that there is an error in the height after projection compensation. At this time, the iteration mechanism is activated, and the contact direction inversion compensation mechanism is triggered again based on the current projection compensation.

[0108] In this embodiment, step S5 of the method, through the implementation of steps S51 and S52, constructs a calibration result reliability evaluation and iterative optimization mechanism based on post-compensation residual analysis. Specifically, after completing the contact direction projection compensation for each sampling point in step S51, the system extracts the calibrated vertical displacement value Zcorr and calculates its overall calibrated vertical displacement average value Zcorr. avg Subsequently, an evaluation formula based on the mean absolute deviation is constructed to output the calibration accuracy result index Δz, which measures the consistency of the Z-direction residuals after compensation at each point in the five-point sampling matrix. This index quantifies the spatial flatness performance after calibration, reflecting the overall stability and consistency of the platform's attitude calibration. In S52, the calibration accuracy result index Δz calculated in real time is compared and evaluated with the pre-set error threshold F2. If the calibration accuracy result index Δz ≤ the error threshold F2, the calibration is considered to have passed, indicating that the platform has achieved high consistency under the current attitude. If the calibration accuracy result index Δz > the error threshold F2, it indicates that there is still residual attitude deviation or insufficient direction compensation. The system will automatically start the iteration mechanism to trigger the contact direction reconstruction and projection compensation process again, forming an intelligent closed-loop calibration chain. This stage effectively avoids the risk of outputting erroneous calibration results by introducing "multi-point residual consistency evaluation and dynamic threshold trigger control," improves the reliability of calibration output results and the platform's focusing consistency, and enhances the adaptive optimization capability and calibration convergence stability under complex installation attitudes, non-ideal contact states, or batch measurement scenarios.

[0109] Example 7

[0110] Please see Figure 1 and Figure 2 A smart calibration system for a contact displacement sensor includes a sampling point sampling module, a data transmission processing module, an attitude vector offset analysis module, a projection compensation module, and a calibration accuracy result analysis module.

[0111] The sampling point sampling module sets up a five-point sampling matrix on the micro-optical focusing platform and sets up acquisition devices in the five-point sampling matrix to acquire three-dimensional offset data in real time.

[0112] The data transmission processing module transmits the local offset vector geometrically to the synchronous control system and preprocesses the set of local offset vectors to obtain the reference normalized feature vector d.

[0113] The attitude vector offset analysis module calculates and outputs the preliminary attitude evaluation index Satt based on the benchmark normalized feature vector d, performs a preliminary comparative evaluation, and triggers the contact direction inversion compensation mechanism based on the preliminary comparative evaluation results.

[0114] The projection compensation module defines the actual contact direction unit vector for each point after triggering the contact direction inversion compensation mechanism, and calculates and outputs the calibrated vertical displacement value Zcorr based on the contact direction unit vector to perform projection compensation.

[0115] The calibration accuracy result analysis module calculates and outputs the calibration accuracy result index Δz based on the vertical displacement value Zcorr after calibration, and performs a secondary comparative evaluation to analyze the reliability of the calibration.

[0116] 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.

Claims

1. An intelligent calibration method for a contact displacement sensor, characterized by: The method comprises the following steps: S1, setting a five-point sampling matrix on a micro-optical focusing platform, and setting an acquisition device in the five-point sampling matrix to collect three-dimensional offset data in real time; S2, geometrically transmitting the local offset vector to a synchronous control system, pre-processing the local offset vector set, and obtaining a reference normalized feature vector d; S3, based on the reference normalized feature vector d, calculating an initial attitude evaluation index Satt, performing an initial comparative evaluation, and triggering a contact direction inversion compensation mechanism based on the initial comparative evaluation result; The S3 includes S31 and S32; S31, based on the reference normalized eigenvector Construct a spatial disturbance covariance tensor, identify the geometric deviation trend of the respective reference normalized eigenvector in the five-point sampling matrix, and obtain the preliminary attitude evaluation index Satt by taking the maximum eigenvalue of the spatial disturbance covariance tensor. The distribution intensity in the strongest disturbance direction in the entire contact displacement sensor probe disturbance vector field is identified. S32, based on the obtained initial attitude evaluation index Satt, performing an initial comparative evaluation, measuring the initial attitude evaluation index Satt under the normal running state by sampling multiple groups of the same micro-optical focusing platform, and obtaining a posture stability threshold F1 by calculating the mean value and 3 times the standard deviation. The posture stability threshold is compared with the real-time obtained attitude evaluation index Satt to analyze the posture stability of the current contact displacement sensor probe, and the contact direction inversion compensation mechanism is triggered based on the evaluation result. The specific evaluation content is as follows: When the initial attitude evaluation index Satt is less than or equal to the posture stability threshold F1, it indicates that the normal posture stability is disturbed, and the calibration is completed at this time; When the initial attitude evaluation index Satt is greater than the posture stability threshold F1, it indicates that the vertical axis direction offset Z exists in the tilt, and the contact direction inversion compensation mechanism is triggered at this time; S4, after triggering the contact direction inversion compensation mechanism, defining the actual contact direction unit vector of each point, and based on the contact direction unit vector, calculating and outputting the calibrated vertical direction displacement value Zcorr to perform projection compensation; S5, based on the calibrated vertical direction displacement value Zcorr, calculating and outputting the calibration accuracy result index△z, and performing secondary comparative evaluation to analyze the reliability of the calibration.

2. The method of intelligent calibration of a contact displacement sensor according to claim 1, wherein: The S1 includes S11 and S12; S11, in the micro-optical focusing platform, a five-point sampling matrix is set, which includes P0 sampling point, P1 sampling point, P2 sampling point, P3 sampling point and P4 sampling point, and the attitude of the micro-optical focusing platform is identified; Wherein, P0 sampling point represents the center point of the micro-optical focusing platform, P1 sampling point represents the lower side of the center point of the micro-optical focusing platform, P2 sampling point represents the upper side of the center point of the micro-optical focusing platform, P3 sampling point represents the right side of the center point of the micro-optical focusing platform, and P4 sampling point represents the left side of the center point of the micro-optical focusing platform; S12, a contact displacement sensor probe is installed at each point in the five-point sampling matrix, which is in physical contact with the micro-optical focusing platform, and three-dimensional offset data of each point in the optical focusing platform is collected in real time; The three-dimensional offset data includes horizontal axis direction offset X, vertical axis direction offset Y and vertical axis direction offset Z.

3. A method of intelligent calibration of a contact displacement sensor according to claim 2, characterized in that: The S2 includes S21 and S22; S21, construct a synchronous control system, set up a wireless communication network, and wirelessly connect the communication module of the contact displacement sensor probe to the synchronous control system, so as to transmit the real-time three-dimensional offset data to the synchronous control system; S22, real-time receiving three-dimensional migration data in the synchronous control system, and pre-processing the three-dimensional migration data to obtain a reference normalized feature vector at the i th sampling point and the j th contact ; the reference normalized feature vector at the i-th sampling point at the j-th contact including a horizontal axis direction offset amount at the i-th sampling point at the j-th contact , a vertical axis direction offset amount at the i-th sampling point at the j-th contact and a vertical axis direction offset amount at the i-th sampling point at the j-th contact ; The preprocessing includes abnormal value identification and elimination, noise smoothing, and coordinate normalization and reference alignment; The abnormal value identification and elimination calculates the mean and standard deviation of all directions in each three-dimensional offset data for each sample, and calculates the difference between the real-time three-dimensional offset data and the mean, and outputs the result greater than 3 times the standard deviation as an abnormal value and eliminates it. The noise smoothing uses a Savitzky-Golay filter to fit a polynomial in a sliding window to smooth the three-dimensional offset data after abnormal value identification and elimination. The coordinate normalization and reference alignment set the mean of the offset vector three-dimensional offset data of the collection point P0 as a 0 vector, and compensate for the difference between all three-dimensional offset data.

4. The method of intelligent calibration of a contact displacement sensor of claim 1, wherein: The S4 includes S41 and S42. S41、in the preliminary comparison and evaluation of the contact direction inversion compensation mechanism, by defining the actual contact direction unit vector of all sampling points, and then through the contact displacement sensor probe, the total length of the displacement of the i sampling point of the original vertical axis direction offset Z reading of each contact point is collected .

5. A method of intelligent calibration of a contact displacement sensor according to claim 4, characterized in that: S42、based on the actual contact direction unit vector and the total displacement length of the i-th sampling point The projection compensation of the vertical axis direction offset Z is performed, and the calibrated vertical direction displacement value Zcorr of the vertical axis direction offset Z after the projection compensation is output. The actual direction vector of the touch displacement sensor probe is inversely calculated according to the calibrated vertical direction displacement value Zcorr through the synchronous control system.

6. A method of intelligent calibration of a contact displacement sensor according to claim 5, characterized in that: The S5 includes S51 and S52. S51, after performing the projection compensation, extracting the calibrated vertical direction displacement value Zcorr and the calibrated vertical direction displacement average value of the calibrated vertical direction displacement value Zcorr The calculation output calibration precision result index Δz is performed, which measures the vertical axis direction deviation consistency between the five-point sampling matrix sampling points of the micro-light focusing platform.

7. A method of intelligent calibration of a contact displacement sensor according to claim 6, characterized in that: S52, output the calibration accuracy result index △z again for secondary comparison and evaluation, set an error threshold F2 according to the surface flatness error allowed by the micro-light focusing platform, and perform secondary comparison and evaluation on the real-time calibration accuracy result index △z and the error threshold F2, judge the error of the micro-light focusing platform after projection compensation, and execute the iteration mechanism based on the secondary comparison and evaluation result, and the specific evaluation content is as follows. When the calibration accuracy result index △z is less than or equal to the error threshold F2, it means that the height is consistent after projection compensation, and the calibration is successful at this time. When the calibration accuracy result index △z is greater than the error threshold F2, it means that there is an error in the height after projection compensation, and the iteration mechanism is started at this time, and the secondary contact direction inversion compensation mechanism is triggered based on the current projection compensation.

8. An intelligent calibration system for a contact displacement sensor, applied to the intelligent calibration method for a contact displacement sensor according to any one of claims 1-7, characterized in that: It includes a sampling point sampling module, a data transmission processing module, a posture vector offset analysis module, a projection compensation module, and a calibration accuracy result analysis module. The sampling point sampling module sets a five-point sampling matrix on the micro-light focusing platform, and sets a collection device in the five-point sampling matrix to collect three-dimensional offset data in real time. The data transmission processing module transmits the local offset vector set to the synchronous control system, and preprocesses the local offset vector set to obtain the reference normalized feature vector d. The posture vector offset analysis module calculates and outputs the preliminary attitude evaluation index Satt based on the reference normalized feature vector d, and performs preliminary comparison and evaluation, and triggers the contact direction inversion compensation mechanism based on the preliminary comparison and evaluation result. The projection compensation module defines the actual contact direction unit vector of each point after triggering the contact direction inversion compensation mechanism, and calculates and outputs the calibrated vertical direction displacement value Zcorr based on the contact direction unit vector to perform projection compensation. The calibration accuracy result analysis module calculates and outputs the calibration accuracy result index Δz based on the vertical direction displacement value Zcorr after calibration, and performs secondary comparison and evaluation to analyze the reliability of calibration.

Images