Search

CN-121999014-A - Linear motion module accurate control method and system based on AI vision

CN121999014ACN 121999014 ACN121999014 ACN 121999014ACN-121999014-A

Abstract

The invention relates to the technical field of image recognition control, in particular to an AI vision-based linear motion module accurate control method and system, comprising the following steps: and acquiring an interference fringe image, extracting a gray path, analyzing the center point displacement to generate a coordinate sequence, repairing the interference characteristic position, aligning the image and the control time sequence to generate a synchronous time table, and forming a control output configuration set by the response of the mapping structure. In the invention, a peak value path is constructed through interference fringe gray information in an image frame, the pixel displacement of a central point is extracted, a coordinate migration sequence is constructed, fine granularity capturing and tracking of an execution state are realized, an external interference area is identified according to the gradient change of an image, vector repairing is carried out, the stability and the characteristic reliability of image data are ensured, the response of the repaired image is aligned with the time sequence of a control instruction, an impulse response lag statistical mechanism is constructed, the accurate mapping adjustment of control frequency is realized, and the control precision and the response consistency in the execution of high-frequency displacement are improved.

Inventors

  • ZHENG ZHIFENG
  • ZHANG XIN

Assignees

  • 深圳市美蓓亚斯科技有限公司

Dates

Publication Date
20260508
Application Date
20260204

Claims (10)

  1. 1. The linear motion module accurate control method based on AI vision is characterized by comprising the following steps: S1, acquiring interference fringe images irradiated on the surface of an actuator of a high-speed sliding table, collecting continuous image frames in an image processing unit, extracting gray level change areas of interference fringes in the image frames, and identifying the central axial direction of the interference fringes in the image to obtain an interference peak gray level path set; S2, calling the interference peak gray level path set, extracting the horizontal and longitudinal pixel positions of the interference center point of the continuous frame under the image coordinate system, and judging the difference value according to the variation distance of the center point coordinate between the real-time frame and the previous frame in the horizontal direction and the sampling time interval to obtain a coordinate migration sequence executed by the sliding table; S3, executing a coordinate migration sequence according to the sliding table, identifying gray gradient direction values corresponding to the edges of the high-speed sliding table structure in the image frame and brightness distribution of the boundary area, and calculating pixel gray gradient direction consistency in continuous frames at the boundary to obtain a feature position set after interference repair; And S4, aligning the repaired position difference value with the control pulse signal time sequence through the interference repaired characteristic position set, and counting response lag change trend in adjacent control periods according to the time difference between the effective time of the instruction and the real-time image response delay to obtain a control pulse synchronization time table.
  2. 2. The AI vision-based linear motion module precise control method according to claim 1, wherein the interference peak gray scale path set comprises gray scale distribution peak positions, stripe path pixel tracks and inter-frame stripe morphology changes, the sliding table execution coordinate migration sequence comprises coordinate variation tracks, inter-frame displacement variation values and pixel point movement trends, the interference post-repair characteristic position set comprises edge vector offset values, characteristic lighting correction parameters and gradient direction correction indexes, and the control pulse synchronization schedule comprises control period identifiers, time delay variation curves and synchronization pairs Ji Chazhi.
  3. 3. The AI vision-based linear motion module accurate control method of claim 1, wherein the specific steps of S1 are as follows: S101, acquiring interference fringe images irradiated on the surface of a high-speed sliding table actuator, calling an image frame sequence continuously acquired in an image processing unit, performing linear traversal on gray values of pixel columns in the image frame, performing curve fitting on gray difference value sequences of adjacent pixel points, extracting position indexes of gray change areas, and generating an interference fringe gray mutation area set; S102, based on the interference fringe gray scale abrupt change region set, carrying out mean value aggregation on the coordinates of central points between adjacent abrupt change regions in each frame of image, constructing an interference center line position sequence in a two-dimensional matrix of the image frame, and carrying out linear fitting on longitudinal coordinate points of the center line in the image frame to generate an interference fringe center axial path sequence; S103, calling the interference fringe central axial path sequence, extracting gray peak points from each frame in continuous image frames along the central axial path coordinate, aggregating the gray peak points into a plurality of groups of homologous sequences according to the image frame acquisition time sequence index, and carrying out path numbering on each group of sequences to obtain an interference peak gray path set.
  4. 4. The AI vision-based linear motion module precise control method according to claim 3, wherein the specific steps of S2 are as follows: S201, invoking the interference peak gray level path set, extracting horizontal and longitudinal pixel indexes of gray level peak points corresponding to continuous frames on each path under an image coordinate system, performing time sequence arrangement on peak point coordinates of the continuous frames in the same path, and performing structural processing on the arranged coordinate sequences to generate an inter-frame interference center point pixel position sequence; S202, based on the inter-frame interference center point pixel position sequence, carrying out ratio operation on a distance change value of a center point in a transverse pixel coordinate direction in an adjacent frame and an image sampling time interval to obtain a pixel displacement increment in unit time, judging a threshold value as two pixel units according to the set transverse displacement, and carrying out index recording on points of which the displacement increment exceeds the threshold value in continuous frames to obtain a continuous transverse pixel mutation point sequence; S203, calling the continuous transverse pixel mutation point sequence, extracting the transverse coordinate values of the image corresponding to the multiple points, mapping the transverse coordinate values to an execution space coordinate system of the high-speed sliding table, and performing time sequence arrangement on the mapping coordinates to generate a sliding table execution coordinate migration sequence.
  5. 5. The AI-vision-based linear motion module precise control method of claim 4, wherein the specific step of S3 is: S301, invoking the sliding table to execute a coordinate migration sequence, extracting an image block of the edge region of the sliding table in an image frame corresponding to the execution coordinates, carrying out convolution processing on pixels in the image block of the edge region of the sliding table, obtaining component values of gray gradients in the transverse direction and the longitudinal direction, and carrying out pixel-by-pixel traversal on brightness values of the edge region to obtain gray gradient direction values and brightness distribution sets of the edge region; S302, calculating the angle difference value of the gray gradient direction of the boundary area in the continuous image frames at the same position based on the gray gradient direction value of the edge area and the brightness distribution set, counting the pixel proportion that the absolute value of the angle difference value is smaller than a preset angle consistency threshold value, and carrying out density statistics on the pixel points with gray change larger than a brightness mutation detection threshold value to obtain an edge interference characteristic point area index set; S303, calling the edge interference characteristic point region index set, performing vector translation operation on region image coordinates, performing coordinate correction according to the central position difference value of adjacent undisturbed regions, performing position update on the coordinate correction points, identifying continuously corrected position distribution, and generating an interference repaired characteristic position set.
  6. 6. The AI-vision-based linear motion module precise control method of claim 5, wherein the specific step of S4 is: s401, invoking the feature position set after interference repair, extracting position difference values of the same feature point in continuous image frames in a control period, arranging the position difference value sequences according to image acquisition time stamps, acquiring time sequence marks of control pulse signals issued by a controller, corresponding position variation after repair to a control pulse time axis, and generating a position response and control instruction alignment sequence; S402, based on the alignment sequence of the position response and the control command, extracting the command issuing time and the acquisition time of a corresponding position change response frame in a control period, calculating the difference value of the two time stamps as the image response lag time, and carrying out control period serialization processing on the lag time to generate a control period response lag change trend set; S403, calling the control period response lag change trend set, performing window smoothing on time difference values in the lag change sequence, establishing an equally-spaced pulse number index sequence by combining an original control instruction time axis, and matching and recording response delay of the control period and the pulse index to obtain a control pulse synchronization time table.
  7. 7. The AI vision-based linear motion module precise control method of claim 6, wherein the process of performing window smoothing on the time difference value in the hysteresis change sequence is specifically that average value calculation is performed on image response hysteresis time in three adjacent control periods in the control period response hysteresis change trend set, so as to obtain a time difference value sequence after corresponding smoothing; The process of establishing the equal interval pulse number index sequence by combining the original control instruction time axis specifically comprises the steps of taking a time stamp of a time sequence mark of a control pulse signal issued by a controller as an initial reference point, and numbering and indexing according to a fixed time interval; The process of matching and recording the response delay of the control period with the pulse index is specifically to record the time difference value after the smoothing processing of the time point at the time axis position corresponding to the numbered index.
  8. 8. The AI vision-based linear motion module precise control method according to claim 1, further comprising the step of S5: S5, calling time sequence configuration corresponding to a key control period of the high-speed sliding table in the control pulse synchronization time table, corresponding to a load peak change trend proceeding interval in an operation record of a real-time sliding table executor, measuring a structure response interval range in the corresponding control period by combining a deformation range and a limit threshold value of a main body material of the high-speed sliding table, and mapping the structure response interval to a controller output pulse frequency curve to obtain a sliding table control output configuration set; the sliding table control output configuration set comprises a control frequency parameter set, a structural response mapping table and a key period regulation scheme.
  9. 9. The AI-vision-based linear motion module precise control method of claim 8, wherein the specific step of S5 is: S501, calling the control pulse synchronization time table, extracting a pulse time sequence configuration interval marked as a key control period of a high-speed sliding table, sequencing and calibrating control pulse numbers and time points in the pulse time sequence configuration interval, extracting load peak value change trend data in a corresponding time period from an operation record of a sliding table actuator, carrying out normalization processing on a time axis, and then carrying out interval correspondence to generate a key control period load trend alignment sequence; S502, based on the key control period load trend alignment sequence, extracting a load peak value time period in a corresponding control period, comparing and judging the load peak value time period with a deformation range limit threshold of a high-speed slide main body material, screening continuous time periods of which the load peak value exceeds the deformation range limit threshold, and aggregating the continuous time periods into a structure response active interval according to a time sequence to obtain a structure response interval range set; S503, calling the structure response interval range set, mapping the response interval to a corresponding time interval on the controller pulse output frequency curve, and carrying out statistics and interval marking on the pulse frequency change gradient in the mapping section to generate a slipway control output configuration set.
  10. 10. The AI vision-based linear motion module accurate control system, characterized in that the system is used for realizing the AI vision-based linear motion module accurate control method according to any one of claims 1 to 9, and the system comprises: The image acquisition module irradiates an interference fringe image on the surface of the high-speed slide table actuator, detects the variation amplitude of a gray level distribution area in an image frame, identifies a fringe central axis formed by the continuity of gray level differences in the image, positions gray level peak point coordinates on the same interference fringe in the continuous frame, and generates an interference peak gray level path set; The pixel track module extracts the transverse and longitudinal pixel positions of the interference center point in the image coordinate system according to the interference peak gray level path set, acquires the pixel offset value and the frame sampling time interval of the center point between the real-time frame and the previous frame in the horizontal and axial direction, classifies and calibrates the sliding direction, and generates a sliding table execution coordinate migration sequence; The coordinate migration module executes a coordinate migration sequence according to the sliding table, extracts a gray level distribution curve of an edge contour line of a high-speed sliding table structure in an image frame, acquires gray level gradient direction values of continuous pixels at edge positions, judges consistency of gray level gradient directions of boundary areas between frames, records an edge characteristic point coordinate interval influenced by illumination disturbance and structural deformation, and generates a characteristic position set after interference repair; The characteristic repairing module calls the characteristic position set after interference repairing, obtains a matching item of a position point sequence and a control pulse signal timestamp issued by the controller, counts time difference between effective time of instructions in three continuous frames and a position response frame in an image, identifies response hysteresis change information in a control period, constructs a corresponding relation between a hysteresis change coefficient and the control period, and generates a control pulse synchronization time table; And the time sequence configuration module extracts a control pulse signal sequence corresponding to a key control period based on the control pulse synchronization time table, calls a load peak change curve of the corresponding period in the operation record of the high-speed sliding table executor, compares the load peak change curve with a deformation threshold interval of the sliding table main body material, and acquires a sliding table control output configuration set.

Description

Linear motion module accurate control method and system based on AI vision Technical Field The invention relates to the technical field of image recognition control, in particular to an AI vision-based linear motion module accurate control method and system. Background The technical field of image recognition control relates to the technical field of image recognition control, which is characterized in that an image sensor is used for collecting image information of an external environment or a target object, an image recognition algorithm is used for analyzing and recognizing image content, and a recognition result is used as a control instruction basis so as to realize closed-loop control of a system or equipment. The core matters in the technical field comprise image acquisition and preprocessing, feature extraction and recognition, target positioning and tracking, visual feedback control and the like. The image recognition control is widely applied to scenes such as industrial automation, robot navigation, intelligent manufacturing, traffic monitoring and the like, has high-precision sensing and dynamic response capabilities, is a cross technology system integrating artificial intelligence, image processing and control engineering, and plays a key role in realizing high-precision control tasks based on visual feedback. The traditional accurate control method of the linear motion module is to acquire motion state information through detection devices such as an encoder and a limit switch, and to regulate the output of a driving motor by combining a preset control program, so as to control linear motion components such as a sliding table to move according to a specified track. In the process, position detection is mainly finished by a hardware sensor, and the control precision is limited by the resolution of the sensor and the precision of a mechanical structure. In the traditional method, the rotating speed or the stepping frequency of the motor is adjusted in real time according to the position signal returned by the sensor by using a PID (proportion integration differentiation) adjustment mode so as to realize target displacement, but the problems of lag control response, easiness in interference, lack of environment adaptability and the like exist. In the prior art, a hardware sensor such as an encoder and a limit switch is relied on to acquire a position state, the sensing precision is limited by the matching degree of the resolution of the sensor and a mechanical structure, so that the detection capability is insufficient in a high-frequency response or tiny displacement scene, the updating frequency of position information is limited, the high-precision control of a fast moving part is difficult to realize, the control mode depends on PID feedback adjustment, signal lag and error accumulation are easy to occur in dynamic response, the abrupt change of a target state cannot be timely dealt with, the adaptability to external disturbance is weak in a strong interference environment, a multidimensional image information utilization means in the moving process is lacked, the real-time accurate identification and adjustment of an executing part are difficult to realize, and the closed-loop control performance of a linear motion device is restricted. Disclosure of Invention In order to solve the technical problems in the prior art, the embodiment of the invention provides an AI vision-based linear motion module accurate control method, which comprises the following steps: S1, acquiring interference fringe images irradiated on the surface of an actuator of a high-speed sliding table, collecting continuous image frames in an image processing unit, extracting gray level change areas of interference fringes in the image frames, and identifying the central axial direction of the interference fringes in the image to obtain an interference peak gray level path set; S2, calling the interference peak gray level path set, extracting the horizontal and longitudinal pixel positions of the interference center point of the continuous frame under the image coordinate system, and judging the difference value according to the variation distance of the center point coordinate between the real-time frame and the previous frame in the horizontal direction and the sampling time interval to obtain a coordinate migration sequence executed by the sliding table; S3, executing a coordinate migration sequence according to the sliding table, identifying gray gradient direction values corresponding to the edges of the high-speed sliding table structure in the image frame and brightness distribution of the boundary area, and calculating pixel gray gradient direction consistency in continuous frames at the boundary to obtain a feature position set after interference repair; And S4, aligning the repaired position difference value with the control pulse signal time sequence through the interference repaired characteristic position set, a