Flexible line drive control method and system for cambered surface precision measurement

Abstract

The invention provides a flexible line drive control method and system for cambered surface precision measurement, and relates to the technical field of data processing, and the method comprises the steps: executing a magnetic flux-position conversion algorithm according to an MCU, and generating absolute position data; packaging the absolute position data according to a synchronous serial communication protocol to obtain a data frame, and transmitting the data frame through a differential signal link to generate a data frame to be analyzed; the motion controller receives the data frame to be analyzed and analyzes the data frame, and when the analyzed state flag indicates an effective position, a synchronous driving instruction of the multi-rotor sliding block is generated based on the analyzed position data; when the analyzed state flag indicates abnormity, the driving instruction is interrupted, magnetic railing ruler damage diagnosis is activated, and finally a damage diagnosis report is generated. According to the invention, the measurement efficiency is improved, and the requirement of continuous and rapid measurement of the cambered surface is met.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a flexible line drive control method and system for precision measurement of curved surfaces. Background Technology

[0002] Precision measurement technology for curved surfaces has significant application value in fields such as high-end equipment manufacturing, for example, in the quality inspection of complex curved surface components such as turbine blades and optical lenses. Traditional measurement methods, such as contact probes and non-contact optical measurements, have limitations: contact measurements are easily affected by probe radius correction errors, which may introduce deviations of several micrometers in the measurement of curved surfaces with small radii of curvature; optical measurements are sensitive to environmental vibrations and light interference, and lack real-time performance in dynamic scenarios.

[0003] While some traditional techniques offer high measurement accuracy, probe radius correction relies on linear fitting between adjacent measurement points. When measuring dense point clouds, mechanical errors amplify signal noise, leading to increased error rates in regions of abrupt curvature changes. Furthermore, some traditional methods lack dynamic correction capabilities for the measurement path, making them unsuitable for real-time measurement of complex surfaces. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a flexible line drive control method and system for precision measurement of curved surfaces, which improves the measurement efficiency and meets the needs of continuous and rapid measurement of curved surfaces.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A first aspect is a flexible line-driven control method for precision measurement of curved surfaces, the method comprising:

[0007] Step 1: Deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source.

[0008] Step 2: Based on the magnetic signal source, the magnetic flux signal is captured by the absolute encoder integrated into the multi-moving sub-slider, and a nonlinear spatial mapping algorithm is fused to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale to obtain the original magnetic flux signal.

[0009] Step 3: The original magnetic flux signal is analyzed by an AMR sensor to generate incremental code track speed parameters; the original magnetic flux signal is analyzed by a Hall sensor array to generate absolute code track signals. After harmonic interference is eliminated by an interpolation circuit, the cylindrical surface projection unfolding algorithm is fused, and the magnetic flux-position conversion algorithm is executed by the MCU to generate absolute position data.

[0010] Step 4: Encapsulate the absolute position data according to the synchronous serial communication protocol to obtain a data frame, and transmit it through the differential signal link to generate the data frame to be parsed;

[0011] Step 5: The motion controller receives the data frame to be parsed and performs parsing processing. When the parsed status flag indicates a valid position, a synchronous drive command for the multi-movement sub-slider is generated based on the parsed position data. When the parsed status flag indicates an abnormality, the drive command is interrupted and the magnetic scale damage diagnosis is activated, and finally a damage diagnosis report is generated.

[0012] Furthermore, a dual-track magnetic grating ruler is deployed on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source, including:

[0013] Step 11: Calculate the minimum magnetization period length and the total number of magnetization units of the incremental code track using the radius of curvature of the arc track and the preset measurement resolution; Based on the period length and the total number of magnetization units, perform continuous and equally spaced alternating magnetization operations of N and S poles on the surface of the arc track along the motion path direction to finally generate the periodic incremental magnetized code track and the magnetization period length parameters.

[0014] Step 12: Based on the magnetization period length parameter, calculate the physical length of a single symbol in the absolute code channel, and determine the maximum number of binary symbols that can be accommodated according to the total length of the arc track; with the maximum number of symbols as the upper limit, generate a pseudo-random binary sequence through a linear feedback shift register algorithm of a predetermined order;

[0015] Step 13: Based on the pseudo-random binary sequence and the physical length of a single code element, perform absolute code channel magnetization coding mapping to generate an aperiodic magnetization coding distribution, and spatially align and deploy it in parallel with the periodic incremental magnetization code channel to generate a magnetic signal source.

[0016] Furthermore, step 2 includes:

[0017] Step 21: Based on the magnetic signal source, the spatial pose parameters of the multi-moving sub-slider are acquired in real time, and the instantaneous spatial position offset vector, attitude rotation matrix and local normal vector between the sensor array and the magnetic grating ruler are calculated using the arc surface track geometric model.

[0018] Step 22: Based on the instantaneous attitude rotation matrix and local normal vector, derive the actual pointing vector of the sensor detection axis through the nonlinear spatial mapping direction mapping function, and calculate the direction deviation vector with respect to the local normal vector to obtain the distortion direction; calculate the normal distance between the sensor and the magnetic grating based on the instantaneous spatial position offset vector, and derive the effective deviation angle according to the direction deviation vector to calculate the distortion level;

[0019] Step 23: Generate dynamic compensation instructions based on the expected distortion direction and distortion level, and fine-tune the physical pointing of the sensor array in real time to establish the compensated acquisition pointing state; under the established compensated acquisition pointing state, the original spatial magnetic flux distribution signal of the dual-track magnetic grating ruler is synchronously captured by the sensor array to obtain the magnetic flux signal.

[0020] Step 24: The magnetic flux signal and pose parameters are fused to obtain the original magnetic flux signal.

[0021] Furthermore, step 3 includes:

[0022] Step 31: Extract the incremental code track component from the original magnetic flux signal and input it into the AMR sensor to analyze the phase change and generate the velocity parameter; and extract the absolute code track component and analyze it into the original digital symbol sequence through the Hall sensor array, and generate the purified sequence through harmonic filtering by the interpolation circuit.

[0023] Step 32: The purified sequence is input into the MCU for matching with the pre-stored pseudo-random sequence to determine the unique position interval and obtain the coarse-grained absolute position estimate.

[0024] Step 33: Based on the velocity parameter and the coarse-grained absolute position estimate, and according to the curvature parameter of the arc surface, the phase of the incremental code track near the coarse position is mapped from the arc space to the linear unfolded space; the phase change of the incremental code track is calculated in the linear unfolded space to generate the fine-tuning displacement.

[0025] Step 34: The MCU fuses the coarse-grained position estimate with the fine-grained displacement to obtain the fused position value in linear space, and applies the flux-position conversion algorithm to reverse map the fused position to the actual arc space coordinates based on the pre-stored arc curvature parameters, finally generating absolute position data.

[0026] Furthermore, the absolute position data is encapsulated into a data frame according to a synchronous serial communication protocol and transmitted through a differential signal link to generate a data frame to be parsed, including:

[0027] Step 41: The communication protocol processing unit receives the absolute position data result and generates status flag information based on the real-time monitored signal integrity index; the communication protocol processing unit merges the absolute position data result and the status flag information result to generate the original data packet to be encapsulated.

[0028] Step 42: After receiving the original data packet to be encapsulated, the communication protocol processing unit sequentially adds the start synchronization header, the calculated check code, and the end flag as specified in the protocol to generate a complete protocol data frame that conforms to the synchronous serial communication protocol frame format.

[0029] Step 43: The differential signal driving circuit receives the complete protocol data frame, converts the contained serial data stream into differential voltage signal pairs, and sends them to the motion controller side through the differential signal transmission link.

[0030] Step 44: The signal receiving and conditioning circuit receives the differential voltage signal pair, amplifies and shapes it to restore it to a single-ended digital signal stream, and extracts the complete bit stream of the protocol frame structure through sampling, and finally generates the data frame to be parsed.

[0031] Furthermore, the motion controller receives and processes the data frame to be parsed. When the parsed status flag indicates a valid position, it generates a synchronous drive command for the multi-movement sub-slider based on the parsed position data. When the parsed status flag indicates an abnormality, it interrupts the drive command and activates the magnetic scale damage diagnosis, ultimately generating a damage diagnosis report, including:

[0032] Step 51: The communication protocol stack processing unit of the motion controller decodes the data frame to be parsed, extracts the absolute position data, status flag information and check code result, and transmits the check code and the original data part to the data verification and status determination unit for validity determination.

[0033] Step 52: If the data is deemed valid, the multi-axis synchronous control unit performs trajectory planning based on the absolute position data to generate target instructions, calculates synchronous drive instructions, and finally outputs them to the motor driver.

[0034] Step 53: If the data is determined to be abnormal, the safety control unit interrupts the drive command and sends an emergency stop command to the motor driver, activating the magnetic scale damage diagnosis task to generate a diagnosis start command; the diagnosis unit controls the slider to scan at a constant speed based on the diagnosis start command, collects position and magnetic flux signals in real time and calculates them, and finally generates a damage diagnosis report.

[0035] Furthermore, the MCU fuses the coarse-grained position estimate with the fine-tuned displacement to obtain a fused position value in linear space. Then, it applies a flux-position transformation algorithm to inversely map the fused position to the actual arc surface space coordinates based on pre-stored arc surface curvature parameters, ultimately generating absolute position data, including:

[0036] Step 61: Based on the coarse-grained absolute position estimation result and the fine-grained displacement result, the MCU performs a vector superposition operation, algebraically adding the fine-grained displacement result and the coarse-grained absolute position estimation result in the linear expansion space to calculate and generate the linear spatial fused position value.

[0037] Step 62: Divide the linear space fused position value by the radius of curvature to obtain the corresponding center angle value. Calculate the tangential, radial, and vertical coordinates of the arc surface based on the center angle value to generate the preliminary absolute position. Separate the real-time magnetic field strength of the absolute code track from the original magnetic flux signal, compare the deviation ratio between the real-time magnetic field strength and the pre-stored ideal value, and generate the compensated absolute position.

[0038] Step 63: Based on the compensated absolute position, perform the inverse calculation operation of the cumulative arc length of the motion path, bind and encapsulate the arc length value with the spatiotemporal identifier, and finally generate the absolute position data.

[0039] Secondly, a flexible line-driven control system for precision measurement of curved surfaces includes:

[0040] The deployment module is used to deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source.

[0041] The correction module is used to capture magnetic flux signals based on a magnetic signal source by an absolute encoder integrated into a multi-moving sub-slider, and to integrate a nonlinear spatial mapping algorithm to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale, so as to obtain the original magnetic flux signal.

[0042] The calculation module is used to analyze the original magnetic flux signal using an AMR sensor to generate incremental code tracks and speed parameters; it analyzes the original magnetic flux signal using a Hall sensor array to generate absolute code tracks, eliminates harmonic interference through an interpolation circuit, fuses the cylindrical projection unfolding algorithm, and generates absolute position data based on the magnetic flux-position conversion algorithm executed by the MCU.

[0043] The transmission module is used to encapsulate absolute position data into data frames according to a synchronous serial communication protocol, and transmit them through a differential signal link to generate data frames to be parsed.

[0044] The diagnostic module is used by the motion controller to receive and parse the data frames to be parsed. When the parsed status flag indicates a valid position, it generates a synchronous drive command for the multi-movement sub-slider based on the parsed position data. When the parsed status flag indicates an abnormality, it interrupts the drive command and activates the magnetic scale damage diagnosis, and finally generates a damage diagnosis report.

[0045] Thirdly, a computing device, comprising:

[0046] One or more processors;

[0047] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0048] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0049] The above-described solution of the present invention has at least the following beneficial effects:

[0050] By deploying a dual-track magnetic grating ruler, the periodic magnetization structure of the incremental code track can quickly generate velocity parameters, while the pseudo-random binary sequence magnetization encoding of the absolute code track can provide absolute position information. The integration of a nonlinear spatial mapping algorithm corrects the signal acquisition path in real time, effectively eliminating signal deviations caused by changes in the spatial pose of the arc surface, improving the accuracy of the original magnetic flux signal, and thus improving the accuracy of the final position measurement.

[0051] By employing a Hall sensor array to analyze the absolute code channel signal and using an interpolation circuit to eliminate harmonic interference, the influence of external electromagnetic factors on the signal is reduced. Furthermore, by encapsulating data frames using a synchronous serial communication protocol and transmitting them via a differential signal link, the anti-interference performance during data transmission is improved. Attached Figure Description

[0052] Figure 1 This is a schematic flowchart of a flexible line drive control method for precision measurement of curved surfaces provided in an embodiment of the present invention.

[0053] Figure 2 This is a schematic diagram of a flexible line drive control system for precision measurement of curved surfaces provided in an embodiment of the present invention. Detailed Implementation

[0054] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0055] like Figure 1 As shown, embodiments of the present invention propose a flexible line-driven control method for precision measurement of curved surfaces, the method comprising the following steps:

[0056] Step 1: Deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source.

[0057] Step 2: Based on the magnetic signal source, the magnetic flux signal is captured by the absolute encoder integrated into the multi-moving sub-slider, and a nonlinear spatial mapping algorithm is fused to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale to obtain the original magnetic flux signal.

[0058] Step 3: The original magnetic flux signal is analyzed by an AMR sensor to generate incremental code track speed parameters; the original magnetic flux signal is analyzed by a Hall sensor array to generate absolute code track signals. After harmonic interference is eliminated by an interpolation circuit, the cylindrical surface projection unfolding algorithm is fused, and the magnetic flux-position conversion algorithm is executed by the MCU to generate absolute position data.

[0059] Step 4: Encapsulate the absolute position data according to the synchronous serial communication protocol to obtain a data frame, and transmit it through the differential signal link to generate the data frame to be parsed;

[0060] Step 5: The motion controller receives the data frame to be parsed and performs parsing processing. When the parsed status flag indicates a valid position, a synchronous drive command for the multi-movement sub-slider is generated based on the parsed position data. When the parsed status flag indicates an abnormality, the drive command is interrupted and the magnetic scale damage diagnosis is activated, and finally a damage diagnosis report is generated.

[0061] In this embodiment of the invention, by deploying a dual-track magnetic grating ruler, the periodic magnetization structure of the incremental code track can quickly generate velocity parameters, and the pseudo-random binary sequence magnetization encoding of the absolute code track can provide absolute position information. The nonlinear spatial mapping algorithm is integrated to correct the signal acquisition path in real time, effectively eliminating signal deviations caused by changes in the spatial pose of the arc surface, improving the accuracy of the original magnetic flux signal, and thus improving the accuracy of the final position measurement.

[0062] By employing a Hall sensor array to analyze the absolute code channel signal and using an interpolation circuit to eliminate harmonic interference, the influence of external electromagnetic factors on the signal is reduced. Furthermore, by encapsulating data frames using a synchronous serial communication protocol and transmitting them via a differential signal link, the anti-interference performance during data transmission is improved.

[0063] In a preferred embodiment of the present invention, a dual-track magnetic grating ruler is deployed on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source, including:

[0064] Step 11: Calculate the minimum magnetization period length and the total number of magnetization units of the incremental code track using the radius of curvature of the arc track and the preset measurement resolution; Based on the period length and the total number of magnetization units, perform continuous and equally spaced alternating magnetization operations of N and S poles on the surface of the arc track along the motion path direction to finally generate the periodic incremental magnetized code track and the magnetization period length parameters.

[0065] Step 12: Based on the magnetization period length parameter, calculate the physical length of a single symbol in the absolute code channel, and determine the maximum number of binary symbols that can be accommodated according to the total length of the arc track; with the maximum number of symbols as the upper limit, generate a pseudo-random binary sequence through a linear feedback shift register algorithm of a predetermined order;

[0066] Step 13: Based on the pseudo-random binary sequence and the physical length of a single code element, perform absolute code channel magnetization coding mapping to generate an aperiodic magnetization coding distribution, and spatially align and deploy it in parallel with the periodic incremental magnetization code channel to generate a magnetic signal source.

[0067] In this embodiment of the invention, the radius of curvature of the arc track is set according to the requirements of flexible production line or circular production line scenario, and the value range is 50mm to 5000mm (suitable for the measurement of arc surfaces of small and medium-sized precision components); the preset measurement resolution is 1μm (compliant with the HCLA series encoder resolution standard of up to 1μm).

[0068] Based on the 1μm resolution and the feasibility of actual magnetization technology, the minimum magnetization period length of the incremental code track is 5mm (i.e., alternating magnetization of N and S poles with a 5mm interval along the direction of motion). The period length can balance the signal recognition stability and high resolution requirements, which is consistent with the magnetization design of the HCLA series incremental code tracks.

[0069] First, calculate the total unfolded length L = R × θ (mm) using the radius of curvature R (mm) of the arc track and the central angle θ (radians, usually 0.5π to 2π). Divide the total length by the magnetization period length of 5 mm to obtain the total number of magnetization units (round to the nearest integer to ensure coverage of the entire arc track, and allow the last unit length to be no less than 2.5 mm, i.e. 1 / 2 of the period length).

[0070] Based on the 5mm magnetization period length of the incremental code channel, the physical length of a single symbol in the absolute code channel ranges from 5mm to 10mm (to avoid magnetic field crosstalk with the incremental code channel and ensure spatial compatibility between the two code channels). Combined with the HCLA series absolute track design, a length specification adapted to 8-bit pseudo-random binary sequences is adopted.

[0071] The value is determined by dividing the total length (mm) of the curved track by the physical length (mm) of a single code element. The value needs to be combined with the actual specifications of the magnetic scale. For example, when the maximum measuring stroke of the MSA50 series magnetic scale is 320mm, the maximum number of codes is about 64 (320mm ÷ 5mm); when the stroke is 1280mm, it is about 256 (1280mm ÷ 5mm). In actual calculation, a 5% to 10% redundancy should be reserved to cope with the machining error of the curved surface.

[0072] The order n is determined based on the maximum number of symbols, and must satisfy 2. n -1 ≥ maximum number of symbols; combined with the design of the HCLA series absolute orbit using an 8-bit pseudo-random binary sequence, the order n = 8 (2 8 -1 = 255), which can cover the requirement of a maximum of 256 code elements, ensuring the absolute uniqueness of the encoding at any position on the curved surface.

[0073] A pseudo-random binary sequence is generated using an 8th-order linear feedback shift register algorithm, where "0" and "1" correspond to the N and S pole magnetization states of the absolute code channel, respectively.

[0074] The periodic incremental code track and the non-periodic absolute code track are deployed in parallel along the width of the arc surface, with the spacing controlled between 0.5mm and 2mm (to accommodate a matching magnetic scale with a width of 20mm, avoiding magnetic field interference); the position alignment error along the direction of motion is ≤±0.1μm, ensuring that the encoder's Hall sensor array and AMR sensor can synchronously acquire signals from the two code tracks.

[0075] The magnetization parameters, designed with a 1μm resolution, are highly compatible with the accuracy and performance of the HCLA series encoders, ensuring high-resolution requirements for curved surface measurements. The dual-track spatial deployment parameters adapt to the complex operating conditions of flexible or circular production lines, supporting high-speed operation of multiple moving sub-sliders and improving system stability in flexible environments. The incremental track 5mm cycle and absolute track 8-bit sequence design are compatible with the magnetization processes of existing magnetic scales (such as the MSA50 series), reducing engineering implementation complexity.

[0076] In a preferred embodiment of the present invention, based on a magnetic signal source, an absolute encoder integrated into a multi-moving sub-slider captures magnetic flux signals, and a nonlinear spatial mapping algorithm is fused to real-time correct the signal acquisition path according to the spatial pose parameters of the encoder and the magnetic scale, thereby obtaining the original magnetic flux signal, including:

[0077] Step 21: Based on the magnetic signal source, the spatial pose parameters of the multi-moving sub-slider are acquired in real time, and the instantaneous spatial position offset vector, attitude rotation matrix and local normal vector between the sensor array and the magnetic grating ruler are calculated using the arc surface track geometric model.

[0078] Step 22: Based on the instantaneous attitude rotation matrix and local normal vector, derive the actual pointing vector of the sensor detection axis through the nonlinear spatial mapping direction mapping function, and calculate the direction deviation vector with respect to the local normal vector to obtain the distortion direction; calculate the normal distance between the sensor and the magnetic grating based on the instantaneous spatial position offset vector, and derive the effective deviation angle according to the direction deviation vector to calculate the distortion level;

[0079] Step 23: Generate dynamic compensation instructions based on the expected distortion direction and distortion level, and fine-tune the physical pointing of the sensor array in real time to establish the compensated acquisition pointing state; under the established compensated acquisition pointing state, the original spatial magnetic flux distribution signal of the dual-track magnetic grating ruler is synchronously captured by the sensor array to obtain the magnetic flux signal.

[0080] Step 24: The magnetic flux signal and pose parameters are fused to obtain the original magnetic flux signal.

[0081] In this embodiment of the invention, the spatial pose parameters include axial angle, radial angle, radial offset, pitch angle, and the gap between the sensor and the magnetic scale. The values ​​strictly adhere to installation tolerances: axial angle ≤ 1°, radial angle ≤ +0.5°, radial offset ≤ +0.30mm, pitch angle ≤ 0.5°; the recommended gap is 1.2mm (optimal value), with an allowable range of 0.9mm to 1.5mm (maximum not exceeding 1.6mm). The attitude sensors (such as miniature inclinometers or displacement sensors) integrated into the absolute encoder of the multi-moving sub-slider are used for real-time data acquisition, with the sampling frequency consistent with the encoder's data refresh rate (50KHz).

[0082] The geometric model of the curved track is constructed with the center of curvature as the origin, establishing a three-dimensional polar coordinate system (radius = track curvature radius, angle ∈ [starting angle, ending angle]), and mapping it to a Cartesian coordinate system (X, Y, Z). The Z-axis is the axis of the curved surface, and the XY plane is the radial section, adapting to the magnetic levitation loop structure. Based on the model, real-time pose parameters are substituted into the calculation:

[0083] The coordinate deviation (ΔX, ΔY, ΔZ) between the center of the sensor array and the surface of the magnetic scale, with a range of ≤0.30mm (radial) and ≤1.6mm (normal); describing the rotation state of the sensor array relative to the magnetic scale, with the rotation angle range matching the installation tolerance (axial ≤1°, radial ≤0.5°); the normal direction vector of the magnetic scale surface at the current position, derived based on the curvature parameter of the arc surface (radius adapted to flexible production line scenarios).

[0084] Construction of direction mapping function:

[0085] The nonlinear spatial mapping orientation mapping function takes "attitude rotation matrix + local normal vector" as input and "actual pointing vector of sensor detection axis" as output. The function form is a polynomial fitting (including first-order terms and second-order cross terms, such as the square of the rotation angle and the angle product term) to capture nonlinear pointing deviations under non-ideal attitudes.

[0086] After deriving the actual pointing vector of the sensor's detection axis, the vector difference between it and the local normal vector is used to obtain the direction deviation vector. Its direction is decomposed along the X / Y / Z axes, and the angular deviation range is ≤1° (axial) and ≤0.5° (radial), corresponding to the maximum offset of the installation tolerance.

[0087] The normal distance (i.e., actual gap, 0.9mm~1.5mm) between the sensor and the magnetic scale is calculated based on the instantaneous spatial position offset vector; the effective deviation angle is derived by combining the directional deviation vector (converted from the angle deviation and distance, with a range of ≤50μm / m); the final distortion level is the product of the deviation angle and the normal distance, with a range of ≤2μm (matching encoder repeatability <2μm).

[0088] Based on the distortion direction and magnitude, fine-tuning instructions are generated: axial angle deviation → drive rotation mechanism for compensation (range 0°~1°), radial offset → drive translation mechanism for compensation (0~0.30mm), normal distance deviation → fine-tune sensor height (0.9mm~1.5mm), ensuring that after compensation, the deviation between the sensor detection axis and the local normal vector is ≤0.1°, and the normal distance is stable at 1.2±0.3mm.

[0089] In the compensated directional state, the dual magnetic flux signals are synchronously captured by the encoder's Hall sensor array (absolute code track) and AMR sensor (incremental code track) at a sampling frequency of 50KHz to ensure signal time alignment; the range of the captured magnetic flux signals matches the magnetization intensity of the magnetic grating (avoiding damage to the magnetic grating by exceeding 50mT magnetic field density).

[0090] The magnetic flux signal is aligned with the real-time pose parameters according to the timestamp, and a weighted fusion algorithm is used (the larger the pose deviation, the higher the corresponding parameter weight, with a weight range of 0.1 to 0.5) to correct and compensate for signal fluctuations caused by residual errors, and finally generate the original magnetic flux signal with a signal error ≤2μm (meeting the repeatability accuracy requirements).

[0091] The process of constructing, training, and implementing a nonlinear spatial mapping algorithm model:

[0092] Input layer: 4 key pose parameters (axial angle, radial angle, radial offset, normal distance); Output layer: distortion direction vector (3 components) and distortion order (1 scalar); Model structure: a quadratic polynomial fitting model is used, through first-order terms (e.g., axial angle) and second-order terms (e.g., axial angle)... 2 Intersection terms (such as axial angle × radial angle) capture nonlinear relationships to ensure coverage of all poses within the installation tolerance range.

[0093] Based on the input pose parameters and the output distortion parameters, the specific form of the quadratic polynomial fitting model is as follows (variable definitions: α is the axial angle, β is the radial angle, d is the radial offset, h is the normal distance; Δx, Δy, Δz are the distortion direction vector components, s is the distortion order; k is the polynomial coefficient):

[0094] X-direction distortion Δx:

[0095] Δx=k0+k1α+k2β+k3d+k4h+k5α 2 +k6β 2 +k7d 2 +k8h 2 +k9αβ+k 10 αd+k 11 αh+k 12 βd+k 13 βh+k14 dh;

[0096] Y-direction distortion Δy:

[0097] Δy=m0+m1α+m2β+m3d+m4h+m5α 2 +m6β 2 +m7d 2 +m8h 2 +m9αβ+m 10 αd+m 11 αh+m 12 βd+m 13 βh+m 14 dh;

[0098] Z-direction distortion Δz:

[0099] Δz=n0+n1α+n2β+n3d+n4h+n5α 2 +n6β 2 +n7d 2 +n8h 2 +n9αβ+n 10 αd+n 11 αh+n 12 βd+n 13 βh+n 14 dh;

[0100] Distortion level s:

[0101] s=p0+p1α+p2β+p3d+p4h+p5α 2 +p6β 2 +p7d 2 +p8h 2 +p9αβ+p 10 αd+p 11 αh+p 12 βd+p 13 βh+p 14 dh;

[0102] Δx (X-direction distortion) represents the deviation of the magnetic flux signal acquisition path in the X direction (tangent direction of the arc surface), in μm, with a value range of ≤2μm (matching encoder repeatability <2μm), mainly caused by the combined effect of axial angle α and radial offset d.

[0103] Δy (Y-direction distortion) represents the deviation of the magnetic flux signal acquisition path in the Y-direction (radial direction), in μm, with a value range of ≤2μm. It is significantly affected by the radial angle β and the radial offset d. As the angle or offset increases, the distortion also increases.

[0104] Δz (Z-direction distortion) represents the deviation of the magnetic flux signal acquisition path in the Z-direction (normal direction), in μm, with a value range of ≤2μm. It is mainly caused by the fluctuation of the normal distance h and the tilt of the radial angle β. The more unstable the distance, the more obvious the distortion.

[0105] s (distortion level) represents the overall deviation of the acquisition path, in μm, with a value range of ≤2μm. It is the result of the synthesis of Δx, Δy, and Δz, and is used to quantify the overall impact of non-ideal pose on the magnetic flux signal, providing a magnitude basis for dynamic compensation.

[0106] k0 represents a constant term used to compensate for the inherent zero-position distortion of the system (such as mechanical deviations in sensor installation and initial circuit offsets). When all pose parameters are zero (α=β=d=h=0), Δx=k0, and its value is usually ≤0.5μm to ensure that the initial distortion in the X direction is minimized when there is no pose deviation.

[0107] k1α represents the first-order term of the axial angle α (0°~1°), where k1 is a coefficient that reflects the linear effect of α on the distortion in the X direction. When α increases, if k1 is positive, Δx increases linearly with α (e.g., when α=1°, this term contributes k1×1° of distortion).

[0108] k2β represents the first-order term of the radial angle β (0°~+0.5°), where k2 is a coefficient that quantifies the linear effect of β on the distortion in the X direction. The larger the tilt angle of β, the more significant the effect of this term on Δx (because the range of β is small, the absolute value of k2 is usually less than k1).

[0109] k3d represents the first-order term of the radial offset d (0~+0.30mm), where k3 is a coefficient reflecting the linear effect of d on the distortion in the X direction; the larger d is (the farther the radial offset), if k3 is positive, Δx increases linearly with d.

[0110] k4h represents the first-order term of the normal distance h (0.9mm~1.5mm), where k4 is a coefficient that quantifies the indirect effect of h on distortion in the X direction (h mainly affects the Z direction, so the absolute value of k4 is usually small).

[0111] k5α 2 The quadratic term of the axial angle α is represented by k5, which is a coefficient that captures the nonlinear cumulative effect of α (e.g., the distortion at α=1° is twice as large as that at α=0.5°). When k5 is positive, the aggravating effect of increasing α on Δx is more obvious.

[0112] k6β 2 The quadratic term of the radial angle β is represented by k6, which is a coefficient that reflects the nonlinear effect of the square of the tilt angle β on the distortion in the X direction (because the range of β is small, the absolute value of k6 is usually less than k5).

[0113] k7d 2The quadratic term representing the radial offset d, with k7 as the coefficient, quantifies the cumulative effect of the square of d on the distortion in the X direction (when d=0.3mm, this term has a more significant effect than the linear term).

[0114] k8h 2 The quadratic term of the normal distance h is represented by k8, which is a coefficient that reflects the indirect nonlinear effect of the square of h on the distortion in the X direction (h mainly affects the Z direction, so the absolute value of k8 is relatively small).

[0115] k9αβ represents the intersection of axial angle α and radial angle β, where k9 is a coefficient that captures the coupled effect of the two on the distortion in the X direction (e.g., when α=0.5° and β=0.3°, this term contributes k9×0.5°×0.3° of distortion, which is not equal to the sum of the effects of α or β alone).

[0116] k 10 αd represents the intersection of axial angle α and radial offset d, k 10 The coefficient is used to quantify the combined effect of α rotation and d offset on X-direction distortion (when the two work together, the distortion may be more pronounced than when they act alone).

[0117] k 11 αh represents the intersection of the axial angle α and the normal distance h, k 11 The coefficient represents the indirect effect of the α rotation and h gap on the X-direction distortion (since h mainly affects the Z-direction, this coefficient is relatively small).

[0118] k 12 βd represents the intersection of radial angle β and radial offset d, k 12 The coefficient is used to capture the combined effect of β tilt and d offset on X-direction distortion (both are radial parameters, and their coupling effect is significant).

[0119] k 13 βh represents the intersection of the radial angle β and the normal distance h, k 13 The coefficient represents the combined effect of β tilt and h gap on X-direction distortion (since both have a more significant impact on Y / Z directions, this coefficient is relatively small).

[0120] k 14 dh represents the intersection of radial offset d and normal distance h, k 14 The coefficient represents the indirect effect of the combined effect of the d offset and h gap on the X-direction distortion (the coefficient is usually small).

[0121] m0 represents a constant term that compensates for zero-position distortion in the Y direction. Its value is ≤0.5μm, ensuring that the initial distortion is minimized when there is no pose deviation.

[0122] m1α represents the first-order term of the axial angle α, where m1 is a coefficient that quantifies the linear effect of α on the distortion in the Y direction (the direct effect of α rotation on the Y direction).

[0123] m2β represents the first-order term of the radial angle β, where m2 is a coefficient and is a key term affecting the distortion in the Y direction (β is the radial tilt, which directly causes the path offset in the Y direction, so the absolute value of m2 is usually large).

[0124] m3d represents the first-order term of the radial offset d, where m3 is a coefficient that reflects the linear effect of d on the distortion in the Y direction (d is the radial offset, which is closely related to the Y direction).

[0125] m4h represents the first-order term of the normal distance h, where m4 is a coefficient that quantifies the indirect effect of h on the distortion in the Y direction (the coefficient is relatively small).

[0126] m5α 2 The term represents the quadratic term of the axial angle α, and m5 is a coefficient that captures the nonlinear cumulative effect of α on the Y direction.

[0127] m6β 2 The term represents the quadratic term of the radial angle β, with m6 as the coefficient. Since β has a significant effect on the Y direction, the absolute value of this coefficient is usually large.

[0128] m7d 2 The term represents the quadratic term of the radial offset d, and m7 is a coefficient that reflects the cumulative effect of the square of d on the distortion in the Y direction.

[0129] m8h 2 The quadratic term representing the normal distance h is m8, which is a coefficient with a relatively small impact.

[0130] m9αβ represents the interaction term between α and β, where m9 is a coefficient that quantifies the synergistic effect of the two on the distortion in the Y direction.

[0131] m 10 αd represents the intersection of α and d, m 10 The coefficient represents the combined effect of α rotation and d offset on the Y direction.

[0132] m 11 αh represents the intersection of α and h, m 11 As a coefficient, its impact is relatively small.

[0133] m 12 βd represents the interaction term between β and d, m 12 Since β and d are both radial parameters, the coupling effect is significant.

[0134] m 13 βh represents the intersection of β and h, m 13 The coefficient is used to capture the combined effect of β tilt and h gap on the Y direction.

[0135] m 14 dh represents the intersection of d and h, m 14 As a coefficient, its impact is relatively small.

[0136] n0 represents a constant term that compensates for zero-position distortion in the Z direction, with a value ≤ 0.5 μm.

[0137] n1α represents the first-order term of the axial angle α, where n1 is a coefficient that quantifies the linear effect of α on the distortion in the Z direction (the effect is relatively small).

[0138] n2β represents the first-order term of the radial angle β, where n2 is a coefficient that reflects the linear effect of β tilt on the distortion in the Z direction (β tilt causes fluctuations in the normal distance, so the coefficient is more obvious).

[0139] n3d represents the first-order term of the radial offset d, where n3 is a coefficient with a relatively small impact (d is a radial parameter with a weak correlation to the Z direction).

[0140] n4h represents the first-order term of the normal distance h, where n4 is a coefficient and a key term affecting the distortion in the Z direction (h is the vertical gap, which directly determines the path deviation in the Z direction, so the absolute value of n4 is the largest).

[0141] n5α 2 Let n5 represent the quadratic term of α, where n5 is a coefficient with a relatively small impact.

[0142] n6β 2 The term represents the quadratic term of β, and n6 is the coefficient, which captures the nonlinear cumulative effect of the β tilt on the Z direction (this coefficient is more significant because β affects the normal distance).

[0143] n7d 2 Let n represent the quadratic term of d, where n7 is a coefficient with a relatively small impact.

[0144] n8h 2 This represents the quadratic term of h, with n8 being a coefficient that reflects the nonlinear cumulative effect of the square of h on the distortion in the Z direction (h is the core parameter, and this coefficient is quite significant).

[0145] n9αβ represents the interaction term between α and β, where n9 is a coefficient with a relatively small impact.

[0146] n 10 αd represents the intersection of α and d, n 10 As a coefficient, its impact is relatively small.

[0147] n 11 αh represents the intersection of α and h, n 11 The coefficient is used to quantify the combined effect of α rotation and h gap on the Z direction.

[0148] n 12βd represents the intersection of β and d, n 12 As a coefficient, its impact is relatively small.

[0149] n 13 βh represents the intersection of β and h, n 13 As a coefficient, since both β and h affect the normal, the coupling effect is quite significant.

[0150] n 14 dh represents the intersection of d and h, n 14 As a coefficient, its impact is relatively small.

[0151] p0 represents a constant term that compensates for the zero-point deviation of the total distortion, with a value ≤0.5μm.

[0152] p1α represents the first-order term of α, where p1 is the coefficient, quantifying the linear contribution of α to the total distortion.

[0153] p2β represents the first-order term of β, where p2 is the coefficient, reflecting the linear effect of β on the total distortion.

[0154] p3d represents the first-order term of d, where p3 is the coefficient, quantifying the linear effect of d on the total distortion.

[0155] p4h represents the first-order term of h, where p4 is the coefficient and is the key contributor to the total distortion (h directly affects the normal deviation).

[0156] p5α 2 Let p5 represent the quadratic term of α, and p5 be the coefficient, which captures the effect of the nonlinear accumulation of α on the total distortion.

[0157] p6β 2 p6 represents the quadratic term of β, and p6 is the coefficient, reflecting the effect of the nonlinear accumulation of β on the total distortion.

[0158] p7d 2 p7 represents the quadratic term of d, and p7 is the coefficient, which quantifies the effect of the nonlinear accumulation of d on the total distortion.

[0159] p8h 2 Let p8 be the quadratic term of h, and p8 be the coefficient, which is an important nonlinear contribution to the total distortion.

[0160] p9αβ represents the interaction term of α and β, where p9 is the coefficient, capturing the synergistic coupling effect of the two on the total distortion.

[0161] p 10 αd represents the intersection of α and d, p 10 The coefficient is used to quantify the combined effect of the two on the total distortion.

[0162] p 11 αh represents the intersection of α and h, p 11The coefficient represents the combined effect of the two factors on the total distortion.

[0163] p 12 βd represents the interaction term between β and d, p 12 The coefficient is used to capture the synergistic coupling effect of the two on the total distortion.

[0164] p 13 βh represents the interaction term between β and h, p 13 The coefficient is denoted as , as both affect the normal direction and contribute significantly to the total distortion.

[0165] p 14 dh represents the intersection of d and h, p 14 The coefficient is used to quantify the combined effect of the two on the total distortion.

[0166] Model training:

[0167] Within the installation tolerance range, simulate typical poses (e.g., axial angle 0°→1°, interval 0.1°; normal distance 0.9mm→1.5mm, interval 0.1mm), and simultaneously record pose parameters and actual distortions (measured by laser interferometer, accuracy ≤1μm) to generate 1000 sets of samples (80% training, 20% validation).

[0168] Training process: The least squares method is used to optimize the polynomial coefficients so that the error between the model's predicted distortion and the actual distortion is ≤2μm; the validation set error must be ≤2μm to ensure generalization; physical consistency verification (such as the distortion increasing with increasing distance).

[0169] Training data collection:

[0170] Input parameter selection: Axial angle: 0°~1° (0.2° interval, 6 points in total); Radial angle: 0°~+0.5° (0.1° interval, 6 points in total); Normal distance: 0.9mm~1.5mm (0.2mm interval, 4 points in total).

[0171] Output data measurement: For each sampling point, the actual magnetic flux distortion value is measured using a laser interferometer (accuracy ≤ 1μm), generating 6×6×4=144 sets of samples (80% for training and 20% for verification) to ensure coverage of all typical installation deviation scenarios.

[0172] A simplified polynomial consisting of "first-order terms + key quadratic terms" is adopted (only terms with the greatest impact on distortion, such as axial angle × radial angle and normal distance, are retained). 2 This reduces the number of coefficients and avoids complex calculations.

[0173] With the core objective of "the deviation between predicted distortion and actual distortion ≤ 2μm" (matching encoder repeatability < 2μm), we do not define a complex error function separately, but directly use the absolute value of the deviation as the basis for optimization.

[0174] Based on physical laws, preset coefficients are used (e.g., when the normal distance increases, the corresponding distortion coefficient is positive) to ensure that the initial prediction direction is correct.

[0175] Iterative adjustment: Each iteration only adjusts the 3 to 5 coefficients with the greatest impact (such as the first term of the normal distance and the second term of the axial angle). The adjustment range is dynamically set according to the size of the deviation (adjusting more when the deviation is large and adjusting less when the deviation is small, with a range of 0.01 to 0.1), avoiding the tediousness of fixed step size.

[0176] When the prediction bias of all samples in the validation set is ≤2μm, the iteration stops. There is no need to calculate the sum of squared errors; the actual deviation reaching the target is used as the convergence criterion.

[0177] The prediction performance of the optimized coefficients was checked using validation set samples, with the following requirements: all sample deviations ≤ 2μm (meeting the repeatability accuracy of <2μm); the prediction deviation of key parameters (such as normal distance of 1.2mm) ≤ 1μm (matching 1μm resolution); only coefficients that meet the above conditions were retained and solidified into the encoder MCU to support real-time operation.

[0178] Model Implementation:

[0179] The trained model is embedded into the encoder's internal MCU. After receiving the pose parameters in real time, the distortion result is output within 1ms, supporting the real-time performance of dynamic compensation (matching a 50KHz data refresh rate).

[0180] By employing a nonlinear spatial mapping algorithm, the magnetic flux signal acquisition path error is improved, avoiding measurement deviations caused by non-ideal poses. It is compatible with pose fluctuations within installation tolerances, meeting the complex operating conditions of flexible or circular production lines. The algorithm's stability is unaffected by temperature drift, ensuring consistent signal acquisition.

[0181] In a preferred embodiment of the present invention, step 3 includes:

[0182] Step 31: Extract the incremental code track component from the original magnetic flux signal and input it into the AMR sensor to analyze the phase change and generate the velocity parameter; and extract the absolute code track component and analyze it into the original digital symbol sequence through the Hall sensor array, and generate the purified sequence through harmonic filtering by the interpolation circuit.

[0183] Step 32: The purified sequence is input into the MCU for matching with the pre-stored pseudo-random sequence to determine the unique position interval and obtain the coarse-grained absolute position estimate.

[0184] Step 33: Based on the velocity parameter and the coarse-grained absolute position estimate, and according to the curvature parameter of the arc surface, the phase of the incremental code track near the coarse position is mapped from the arc space to the linear unfolded space; the phase change of the incremental code track is calculated in the linear unfolded space to generate the fine-tuning displacement.

[0185] Step 34: The MCU fuses the coarse-grained position estimate with the fine-grained displacement to obtain the fused position value in linear space, and applies the flux-position conversion algorithm to reverse map the fused position to the actual arc space coordinates based on the pre-stored arc curvature parameters, finally generating absolute position data.

[0186] In this embodiment of the invention, the incremental code track component (corresponding to a 5mm long S / N alternating magnetization structure, i.e., the incremental code track period is 5mm) is separated from the original magnetic flux signal; the AMR sensor samples this component, and the sampling frequency is consistent with the encoder data refresh rate (50KHz) to ensure that the magnetic field change corresponding to every 1μm displacement can still be captured at the maximum speed of 5m / s.

[0187] The real-time velocity is calculated by detecting the phase change of the magnetic field generated by the S / N alternation and combining it with the incremental code track period (5mm); the velocity parameter range covers 0 to 5m / s (including the maximum crossing velocity), the calculation accuracy is adapted to 1μm resolution, and the velocity error is ≤0.1% (based on sensor linearity).

[0188] Separate the absolute code channel component (corresponding to an 8-bit pseudo-random binary sequence magnetization code) from the original magnetic flux signal.

[0189] Hall sensor array analysis: The Hall sensor array performs multi-node synchronous sampling of the absolute code channel components (the sampling interval matches the physical length of the symbol, the symbol length is 5-10mm, compatible with the 5mm incremental period), converting the magnetic field signal into the original digital symbol sequence ("0" / "1" corresponds to the N pole / S pole).

[0190] Interpolation electronic components eliminate harmonic interference in the original sequence and filter out electromagnetic noise (such as 50Hz power frequency interference) in the industrial environment through smoothing processing to generate a purified 8-bit pseudo-random symbol sequence, ensuring that the symbol recognition error rate is ≤0.1%.

[0191] Noise sources and processing objectives: The noise in the original digital symbol sequence mainly comes from electromagnetic interference in the industrial environment (such as 50Hz power frequency interference) and sensor sampling noise, which manifests as glitches or jumps in the symbol signal (deviation range ±3~5μm); the processing objective is to attenuate the noise by more than 20dB and ensure that the symbol recognition error rate of the cleaned sequence is ≤0.1%.

[0192] A sliding window of 3 to 5 consecutive symbols (the window size matches the physical length of the symbol; a 5 to 10 mm symbol corresponds to a window length of 15 to 50 mm) is used to cover a local continuous area of ​​the absolute code channel.

[0193] The mean value of the original symbol sequence within the window is taken. If the deviation of a symbol value from the mean value of the window exceeds ±3σ (σ is the standard deviation of the signal within the window, about 1 to 2 μm), it is determined to be noise and replaced with the mean value to eliminate isolated transitions.

[0194] For incomplete windows (length < 3 symbols) at the beginning and end of the sequence, the mean of one-sided windows is used (e.g., the mean of the last 3 symbols is used to correct the beginning) to avoid edge distortion; the symbol transition amplitude of the processed sequence is ≤ 1μm, which is compatible with the encoder's 1μm resolution. The effectiveness of the filtering is verified by comparing the sequence matching success rate before and after processing (improved to over 99.9%).

[0195] The MCU calls the pre-stored 8-bit pseudo-random binary sequence (which is completely consistent with the absolute code channel magnetization encoding), performs sliding window matching between the cleaned sequence and the pre-stored sequence, and locates the unique index of the current code element in the sequence.

[0196] Based on the physical length of the symbol corresponding to the index (5-10mm), determine the coarse-grained absolute position estimate, with an error range of ≤10mm (i.e., the length of a single symbol).

[0197] Model building:

[0198] Coarse-grained absolute position estimation, arc curvature parameters (curvature radius R, adapted to magnetic levitation loop scene), incremental code track phase change; phase mapping value in linear unfolded space; based on the geometric characteristics of the arc surface, construct the mapping relationship; project the incremental code track phase (corresponding arc length s = Rθ, θ is the central angle) of the arc space to the linear space (length L = s), realize the 1:1 mapping from arc to linear, and ensure that the projection deviation is ≤1μm.

[0199] Model training:

[0200] Training data acquisition: On an arc track with a known radius of curvature, the arc coordinates (R, θ) of multiple feature points and their corresponding linear unfolded lengths L are measured using a laser interferometer, generating 500 sets of samples (covering the entire arc path); Training process: The least squares method is used to optimize the mapping parameters so that the error between the projected linear length and the actual measured value is ≤1μm (matching resolution); 20% of the samples are reserved for validation to ensure generalization across the curvature range (adapting to different arc surfaces of flexible production lines); Convergence criterion: The average error between the training set and the validation set is ≤1μm.

[0201] The specific process of optimizing mapping parameters using the least squares method:

[0202] The model mapping parameter is the transformation coefficient k (dimensionless) from arc space to linear space. The goal is to make the error between the calculated linear unfolded length and the actual measured length ≤ 1μm (matching resolution). Fifty feature points are uniformly selected throughout the entire arc track (e.g., 320mm or 1280mm, corresponding to the maximum measurement stroke in the document), covering different curvature positions (central angle θ from 0 to the maximum angle). For each feature point, the actual linear unfolded length L corresponding to the arc coordinates is measured using a laser interferometer (accuracy ±0.5μm), and the arc space parameters (radius of curvature R, central angle θ) are recorded simultaneously.

[0203] Initial parameter settings are based on geometric relationships (linear length ≈ R × θ), with initial conversion coefficients set to w0 = 1.0 ± 0.1 (allowing for initial deviation). For each feature point, the calculated linear length L' = w × R × θ under the current parameter w is calculated, and the error between this calculated value and the actual value L is eᵢ = L' - L. The total sum of squared errors S = Σeᵢ 2 .

[0204] Minimize S by adjusting w, correcting w in each iteration (step size 0.001), repeating 50–100 times until S no longer decreases significantly (change < 1 μm). 2 The final optimized w must satisfy the error eᵢ of all feature points ≤ 1μm and the average error ≤ 0.5μm; select another 10 feature points that did not participate in the training (covering different curvatures) to verify that their error is also ≤ 1μm, to ensure that the parameters are effective in the full arc range.

[0205] Model Implementation: The trained mapping relationship is solidified into a lookup table and stored in the MCU. After receiving coarse-grained position and curvature parameters in real time, the projection calculation from arc to linear is completed within 1ms, supporting the real-time generation of fine-tuned displacement.

[0206] Within the linear expansion space, based on the velocity parameters (reflecting phase changes per unit time) resolved by the AMR sensor, the displacement corresponding to the incremental code track phase change near the coarse-grained position is calculated with an accuracy ≤1μm (based on 1μm resolution).

[0207] The MCU fuses the coarse-grained position estimate (mm level) with the fine-grained displacement (μm level) to obtain the fused position value in linear space (error ≤ 2μm, meeting the repeatability accuracy requirements); then, through the flux-position conversion algorithm, the linear position is inversely mapped to the actual arc surface coordinates (R, θ) based on the pre-stored arc surface curvature parameters, and finally the absolute position data is generated.

[0208] High-precision positioning: By fusing coarse-grained estimation with μm-level fine-tuning, it meets the position requirements of precision motion control and is compatible with the dynamic control of multi-moving sub-slider. Interpolation circuit filtering and pseudo-random sequence matching reduce the impact of industrial noise on signal analysis and improve stability under complex working conditions. The cylindrical surface projection unfolding algorithm is built based on the geometric characteristics of curved surfaces, accurately realizing the mapping between curved and linear spaces, and is suitable for the curved track scenarios of magnetic levitation loop lines.

[0209] In a preferred embodiment of the present invention, absolute position data is encapsulated into a data frame according to a synchronous serial communication protocol and transmitted through a differential signal link to generate a data frame to be parsed, including:

[0210] Step 41: The communication protocol processing unit receives the absolute position data result and generates status flag information based on the real-time monitored signal integrity index; the communication protocol processing unit merges the absolute position data result and the status flag information result to generate the original data packet to be encapsulated.

[0211] Step 42: After receiving the original data packet to be encapsulated, the communication protocol processing unit sequentially adds the start synchronization header, the calculated check code, and the end flag as specified in the protocol to generate a complete protocol data frame that conforms to the synchronous serial communication protocol frame format.

[0212] Step 43: The differential signal driving circuit receives the complete protocol data frame, converts the contained serial data stream into differential voltage signal pairs, and sends them to the motion controller side through the differential signal transmission link.

[0213] Step 44: The signal receiving and conditioning circuit receives the differential voltage signal pair, amplifies and shapes it to restore it to a single-ended digital signal stream, and extracts the complete bit stream of the protocol frame structure through sampling, and finally generates the data frame to be parsed.

[0214] In this embodiment of the invention, the absolute position data is 26 bits (corresponding to encoder position information, right-aligned, MSB priority, unused low bits are set to 0); the status flag bit is 2 bits (corresponding to general conditions: b1 and b0, where "L" indicates the vehicle is offline and "H" indicates the vehicle is online, and it is also associated with the indicator light status, such as a solid blue light corresponding to a valid position, a purple / red light corresponding to an invalid position, etc.); the signal integrity indicators include data validity (based on the position parsing result) and internal temperature (when >80℃, additional information of the status bit is triggered). The communication protocol processing unit merges the 26-bit position data with the 2-bit status flag bit to generate a 28-bit original data packet to be encapsulated.

[0215] Start Synchronization Header: Added according to the selected synchronous serial communication protocol (such as SSI, BiSS-C, MODBUS-RTU, etc.), with a length of 1 to 2 bytes (for example, the SSI protocol synchronization header is a specific clock pulse sequence, and BiSS-C is a start flag); Checksum: Uses parity check or CRC check (industry standard method), with a length of 1 byte, and the calculation range covers 28 bits of the original data packet to ensure data transmission integrity; End Flag: 1 byte (such as a specific binary sequence "11111111"), indicating the end of frame transmission; Total length of complete data frame: 28 bits (original data) + 8 to 16 bits (synchronization header) + 8 bits (checksum) + 8 bits (end flag) = 52 to 60 bits, adapting to the frame format specified by the protocol.

[0216] The differential signal drive circuit receives complete protocol data frames (serial data streams) and converts them into differential voltage signal pairs (if conforming to the RS485 standard, the voltage range is ±2V to ±6V, transmitted through the 485-A / 485-B pins); the transmission link uses TPU high-flexibility cable (double shielded, twisted pair structure, shielding rate 90%, outer diameter 4.4±0.15mm), with a length range of 0.5m (default) to 10m (maximum allowable length).

[0217] Match the maximum clock frequency of the protocol, such as 1MHz for SSI and 2.5MHz for BiSS-C, to ensure a data refresh rate of 50KHz.

[0218] The signal receiving and conditioning circuit receives differential voltage signal pairs, amplifies and shapes them (adjusting the gain to compensate for cable loss, ensuring the signal amplitude is ≥0.5V), and then restores them to a single-ended digital signal stream (high and low levels correspond to logic "1" and "0" respectively, with a level range of 0-5V). The bit stream is extracted according to the sampling frequency specified in the protocol (matching the transmission rate, such as 1MHz clock corresponding to 1μs sampling interval), and the start synchronization header, check code (after verification), and end flag are identified. Finally, the 28-bit original data packet is parsed to generate the data frame to be parsed.

[0219] The differential signal link, combined with double-shielded twisted-pair cable, enhances electromagnetic interference resistance and adapts to complex industrial environments; the checksum mechanism further ensures data integrity. It supports multiple synchronous serial communication protocols such as SSI, BiSS-C, and MODBUS-RTU, and can be adapted to different motion controller interfaces, flexibly meeting the equipment integration needs of flexible or circular production lines. The transmission rate matches a 50kHz data refresh rate, balancing high-speed transmission with long-distance deployment. A 2-bit status flag is associated with the vehicle's online status and data validity; combined with indicator light information, it facilitates real-time monitoring of transmission status and rapid location of anomalies (such as offline, invalid position, etc.).

[0220] In a preferred embodiment of the present invention, the motion controller receives and parses the data frame to be parsed. When the parsed status flag indicates a valid position, a synchronous drive command for the multi-movement sub-slider is generated based on the parsed position data. When the parsed status flag indicates an abnormality, the drive command is interrupted and the magnetic scale damage diagnosis is activated, ultimately generating a damage diagnosis report, including:

[0221] Step 51: The communication protocol stack processing unit of the motion controller decodes the data frame to be parsed, extracts the absolute position data, status flag information and check code result, and transmits the check code and the original data part to the data verification and status determination unit for validity determination.

[0222] Step 52: If the data is deemed valid, the multi-axis synchronous control unit performs trajectory planning based on the absolute position data to generate target instructions, calculates synchronous drive instructions, and finally outputs them to the motor driver.

[0223] Step 53: If the data is determined to be abnormal, the safety control unit interrupts the drive command and sends an emergency stop command to the motor driver, activating the magnetic scale damage diagnosis task to generate a diagnosis start command; the diagnosis unit controls the slider to scan at a constant speed based on the diagnosis start command, collects position and magnetic flux signals in real time and calculates them, and finally generates a damage diagnosis report.

[0224] In this embodiment of the invention, the communication protocol stack processing unit of the motion controller decodes the data frame to be parsed according to the selected synchronous serial communication protocol (such as SSI, BiSS-C, MODBUS-RTU, etc.), extracts 26 bits of absolute position data (right-aligned, MSB priority, unused low bits are 0), 2 bits of status flag bits (b1 and b0, "H" indicates that the vehicle is online, "L" indicates that the vehicle is offline) and 1 byte of check code.

[0225] The data verification and status determination unit compares the extracted check code with the recalculated check value of the original data (26-bit position data + 2-bit status flag). If they match, the verification passes. The validity of the data is determined by combining the status flag and the indicator light status (e.g., a solid blue light corresponds to a valid position, and a purple / red light corresponds to an invalid position). The determination time is ≤10us (SSI protocol) or 20us (BiSS-C protocol), matching the protocol timeout requirements.

[0226] The multi-axis synchronous control unit performs trajectory planning based on effective absolute position data and a preset trajectory (such as an arc surface measurement path), generating target position and speed commands (speed range 0-5m / s, matching the maximum operating speed).

[0227] The calculated synchronous drive command (including position deviation compensation and speed adjustment parameters) is output through the motor driver interface to ensure that the synchronization error of the multi-moving sub-slider is ≤2μm (matching encoder repeatability) and the response delay is ≤350ms (after power-on initialization).

[0228] The safety control unit immediately interrupts the original drive command and sends an emergency stop command to the motor driver to ensure that the slider decelerates to a stop within ≤50ms, thus avoiding mechanical damage under abnormal conditions.

[0229] Based on the diagnostic start command, the diagnostic unit controls the slider to scan at a low speed (e.g., 1 m / s, below the maximum speed) along the arc track at a constant speed, and collects position data and magnetic flux signals in real time (sampling frequency 50 kHz).

[0230] Compare the real-time magnetic flux signal with the pre-stored standard magnetic flux distribution (when there is no damage), calculate the deviation value (a deviation ≥ 5μm is judged as a potential damage point); combine with magnetic graphics card auxiliary detection to locate the damage location (accuracy ≤ 1mm, refer to installation tolerance).

[0231] The report includes the coordinates of the damage point (spatial location on the arc surface), the damage type (such as abnormal magnetization or physical scratches), the abnormal magnetic flux deviation value, and the recommended maintenance plan. The generation time is ≤10s (covering a maximum measurement stroke of 1280mm).

[0232] By employing dual verification using checksums and status flags, combined with the anti-interference capabilities of differential signal transmission, the reliability of data transmission and analysis is improved, making it suitable for complex industrial environments and meeting the collaborative requirements of precision measurement of curved surfaces. In case of anomalies, the drive is quickly interrupted and an emergency stop is initiated to prevent equipment damage; damage diagnosis can accurately locate the fault point, reducing maintenance time. It supports multiple protocols such as SSI and BiSS-C, flexibly adapting to different motion controllers and enhancing the ease of equipment integration.

[0233] like Figure 2 As shown, embodiments of the present invention also provide a flexible line-driven control system for precision measurement of curved surfaces, comprising:

[0234] The deployment module is used to deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source.

[0235] The correction module is used to capture magnetic flux signals based on a magnetic signal source by an absolute encoder integrated into a multi-moving sub-slider, and to integrate a nonlinear spatial mapping algorithm to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale, so as to obtain the original magnetic flux signal.

[0236] The calculation module is used to analyze the original magnetic flux signal using an AMR sensor to generate incremental code tracks and speed parameters; it analyzes the original magnetic flux signal using a Hall sensor array to generate absolute code tracks, eliminates harmonic interference through an interpolation circuit, fuses the cylindrical projection unfolding algorithm, and generates absolute position data based on the magnetic flux-position conversion algorithm executed by the MCU.

[0237] The transmission module is used to encapsulate absolute position data into data frames according to a synchronous serial communication protocol, and transmit them through a differential signal link to generate data frames to be parsed.

[0238] The diagnostic module is used by the motion controller to receive and parse the data frames to be parsed. When the parsed status flag indicates a valid position, it generates a synchronous drive command for the multi-movement sub-slider based on the parsed position data. When the parsed status flag indicates an abnormality, it interrupts the drive command and activates the magnetic scale damage diagnosis, and finally generates a damage diagnosis report.

[0239] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0240] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0241] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0242] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A flexible linear drive control method for precision measurement of curved surfaces, characterized in that, The method includes: Step 1: Deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source. Step 2: Based on the magnetic signal source, the magnetic flux signal is captured by the absolute encoder integrated into the multi-moving sub-slider, and a nonlinear spatial mapping algorithm is fused to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale to obtain the original magnetic flux signal. Step 3: The original magnetic flux signal is analyzed by an AMR sensor to generate incremental code track speed parameters; the original magnetic flux signal is analyzed by a Hall sensor array to generate absolute code track signals. After harmonic interference is eliminated by an interpolation circuit, the cylindrical surface projection unfolding algorithm is fused, and the magnetic flux-position conversion algorithm is executed by the MCU to generate absolute position data. Step 4: Encapsulate the absolute position data according to the synchronous serial communication protocol to obtain a data frame, and transmit it through the differential signal link to generate the data frame to be parsed; Step 5: The motion controller receives the data frame to be parsed and performs parsing processing. When the parsed status flag indicates a valid position, a synchronous drive command for the multi-movement sub-slider is generated based on the parsed position data. When the parsed status flag indicates an abnormality, the drive command is interrupted and the magnetic scale damage diagnosis is activated, and finally a damage diagnosis report is generated.

2. The flexible line drive control method for precision measurement of curved surfaces according to claim 1, characterized in that, A dual-track magnetic grating ruler is deployed on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source, including: Step 11: Calculate the minimum magnetization period length and the total number of magnetization units of the incremental code track using the radius of curvature of the arc track and the preset measurement resolution; Based on the period length and the total number of magnetization units, perform continuous and equally spaced alternating magnetization operations of N and S poles on the surface of the arc track along the motion path direction to finally generate the periodic incremental magnetized code track and the magnetization period length parameters. Step 12: Based on the magnetization period length parameter, calculate the physical length of a single symbol in the absolute code channel, and determine the maximum number of binary symbols that can be accommodated according to the total length of the arc track; with the maximum number of symbols as the upper limit, generate a pseudo-random binary sequence through a linear feedback shift register algorithm of a predetermined order; Step 13: Based on the pseudo-random binary sequence and the physical length of a single code element, perform absolute code channel magnetization coding mapping to generate an aperiodic magnetization coding distribution, and spatially align and deploy it in parallel with the periodic incremental magnetization code channel to generate a magnetic signal source.

3. The flexible line drive control method for precision measurement of curved surfaces according to claim 2, characterized in that, Step 2 includes: Step 21: Based on the magnetic signal source, the spatial pose parameters of the multi-moving sub-slider are acquired in real time, and the instantaneous spatial position offset vector, attitude rotation matrix and local normal vector between the sensor array and the magnetic grating ruler are calculated using the arc surface track geometric model. Step 22: Based on the instantaneous attitude rotation matrix and local normal vector, derive the actual pointing vector of the sensor detection axis through the nonlinear spatial mapping direction mapping function, and calculate the direction deviation vector with respect to the local normal vector to obtain the distortion direction; calculate the normal distance between the sensor and the magnetic grating based on the instantaneous spatial position offset vector, and derive the effective deviation angle according to the direction deviation vector to calculate the distortion level; Step 23: Generate dynamic compensation instructions based on the expected distortion direction and distortion level, and fine-tune the physical pointing of the sensor array in real time to establish the compensated acquisition pointing state; under the established compensated acquisition pointing state, the original spatial magnetic flux distribution signal of the dual-track magnetic grating ruler is synchronously captured by the sensor array to obtain the magnetic flux signal. Step 24: The magnetic flux signal and pose parameters are fused to obtain the original magnetic flux signal.

4. The flexible line drive control method for precision measurement of curved surfaces according to claim 3, characterized in that, Step 3 includes: Step 31: Extract the incremental code track component from the original magnetic flux signal and input it into the AMR sensor to analyze the phase change and generate the velocity parameter; and extract the absolute code track component and analyze it into the original digital symbol sequence through the Hall sensor array, and generate the purified sequence through harmonic filtering by the interpolation circuit. Step 32: The purified sequence is input into the MCU for matching with the pre-stored pseudo-random sequence to determine the unique position interval and obtain the coarse-grained absolute position estimate. Step 33: Based on the velocity parameter and the coarse-grained absolute position estimate, and according to the curvature parameter of the arc surface, the phase of the incremental code track near the coarse position is mapped from the arc space to the linear unfolded space; the phase change of the incremental code track is calculated in the linear unfolded space to generate the fine-tuning displacement. Step 34: The MCU fuses the coarse-grained position estimate with the fine-grained displacement to obtain the fused position value in linear space, and applies the flux-position conversion algorithm to reverse map the fused position to the actual arc space coordinates based on the pre-stored arc curvature parameters, finally generating absolute position data.

5. The flexible line drive control method for precision measurement of curved surfaces according to claim 4, characterized in that, The absolute position data is encapsulated into a data frame according to the synchronous serial communication protocol and transmitted through a differential signal link to generate the data frame to be parsed, including: Step 41: The communication protocol processing unit receives the absolute position data result and generates status flag information based on the real-time monitored signal integrity index; the communication protocol processing unit merges the absolute position data result and the status flag information result to generate the original data packet to be encapsulated. Step 42: After receiving the original data packet to be encapsulated, the communication protocol processing unit sequentially adds the start synchronization header, the calculated check code, and the end flag as specified in the protocol to generate a complete protocol data frame that conforms to the synchronous serial communication protocol frame format. Step 43: The differential signal driving circuit receives the complete protocol data frame, converts the contained serial data stream into differential voltage signal pairs, and sends them to the motion controller side through the differential signal transmission link. Step 44: The signal receiving and conditioning circuit receives the differential voltage signal pair, amplifies and shapes it to restore it to a single-ended digital signal stream, and extracts the complete bit stream of the protocol frame structure through sampling, and finally generates the data frame to be parsed.

6. The flexible line drive control method for precision measurement of curved surfaces according to claim 5, characterized in that, The motion controller receives and parses the data frame to be parsed. When the parsed status flag indicates a valid position, it generates a synchronous drive command for the multi-movement sub-slider based on the parsed position data. When the parsed status flag indicates an anomaly, the drive instruction is interrupted and the magnetic scale damage diagnosis is activated, ultimately generating a damage diagnosis report, including: Step 51: The communication protocol stack processing unit of the motion controller decodes the data frame to be parsed, extracts the absolute position data, status flag information and check code result, and transmits the check code and the original data part to the data verification and status determination unit for validity determination. Step 52: If the data is deemed valid, the multi-axis synchronous control unit performs trajectory planning based on the absolute position data to generate target instructions, calculates synchronous drive instructions, and finally outputs them to the motor driver. Step 53: If the data is determined to be abnormal, the safety control unit interrupts the drive command and sends an emergency stop command to the motor driver, activating the magnetic scale damage diagnosis task to generate a diagnosis start command; the diagnosis unit controls the slider to scan at a constant speed based on the diagnosis start command, collects position and magnetic flux signals in real time and calculates them, and finally generates a damage diagnosis report.

7. The flexible line drive control method for precision measurement of curved surfaces according to claim 6, characterized in that, The MCU fuses the coarse-grained position estimate with the fine-grained displacement to obtain a fused position value in linear space. Then, it applies a flux-position transformation algorithm to inversely map the fused position to the actual arc surface space coordinates based on pre-stored arc surface curvature parameters, ultimately generating absolute position data, including: Step 61: Based on the coarse-grained absolute position estimation result and the fine-grained displacement result, the MCU performs a vector superposition operation, algebraically adding the fine-grained displacement result and the coarse-grained absolute position estimation result in the linear expansion space to calculate and generate the linear spatial fused position value. Step 62: Divide the linear space fused position value by the radius of curvature to obtain the corresponding center angle value. Calculate the tangential, radial, and vertical coordinates of the arc surface based on the center angle value to generate the preliminary absolute position. Separate the real-time magnetic field strength of the absolute code track from the original magnetic flux signal, compare the deviation ratio between the real-time magnetic field strength and the pre-stored ideal value, and generate the compensated absolute position. Step 63: Based on the compensated absolute position, perform the inverse calculation operation of the cumulative arc length of the motion path, bind and encapsulate the arc length value with the spatiotemporal identifier, and finally generate the absolute position data.

8. A flexible line drive control system for precision measurement of curved surfaces, the system implementing the method as described in any one of claims 1 to 7, characterized in that, include: The deployment module is used to deploy a dual-track magnetic grating ruler on the surface of the curved track. The incremental code track adopts a periodic magnetization structure, and the absolute code track adopts pseudo-random binary sequence magnetization encoding to generate a magnetic signal source. The correction module is used to capture magnetic flux signals based on a magnetic signal source by an absolute encoder integrated into a multi-moving sub-slider, and to integrate a nonlinear spatial mapping algorithm to correct the signal acquisition path in real time according to the spatial pose parameters of the encoder and the magnetic scale, so as to obtain the original magnetic flux signal. The calculation module is used to analyze the original magnetic flux signal using an AMR sensor to generate incremental code track speed parameters; The original magnetic flux signal is analyzed into an absolute code signal by a Hall sensor array. After harmonic interference is eliminated by an interpolation circuit, the cylindrical projection unfolding algorithm is fused, and the absolute position data is generated according to the magnetic flux-position conversion algorithm executed by the MCU. The transmission module is used to encapsulate absolute position data into data frames according to a synchronous serial communication protocol, and transmit them through a differential signal link to generate data frames to be parsed. The diagnostic module is used by the motion controller to receive and parse the data frames to be parsed. When the parsed status flag indicates a valid position, it generates a synchronous drive command for the multi-movement sub-slider based on the parsed position data. When the parsed status flag indicates an abnormality, it interrupts the drive command and activates the magnetic scale damage diagnosis, and finally generates a damage diagnosis report.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images