CN-121977649-A - Online grinding wheel surface detection equipment and method based on laser multi-mode collaborative detection

CN121977649ACN 121977649 ACN121977649 ACN 121977649ACN-121977649-A

Abstract

The invention provides on-line grinding wheel surface detection equipment and method based on laser multi-mode collaborative detection, comprising a magnetic suction bench assembly, a modularized detection unit, a multi-degree-of-freedom motion platform and a master controller, wherein the modularized detection unit is arranged on the multi-degree-of-freedom motion platform, the multi-degree-of-freedom motion platform is arranged on the magnetic suction bench assembly, the modularized detection unit comprises a pneumatic dust removal module and a detection module, the detection module comprises a laser displacement sensor, an optical imaging assembly and a synchronous trigger controller, and the synchronous trigger controller is configured to send synchronous trigger signals to the laser displacement sensor and the optical imaging assembly so as to realize the strict synchronous execution of laser sampling and optical image acquisition actions of the laser displacement sensor and the optical imaging assembly on the grinding wheel surface.

Inventors

WANG SHUAI
ZHANG JIAHUA
WANG NINGCHANG
Chi Binzhou
HUANG YONGQI
Shao Xiaoshuang

Assignees

郑州磨料磨具磨削研究所有限公司

Dates

Publication Date: 20260505
Application Date: 20260211

Claims (10)

1. On-line grinding wheel surface detection equipment based on laser multi-mode collaborative detection is characterized by comprising: the magnetic attraction bench assembly, the modularized detection unit, the multi-degree-of-freedom motion platform and the master controller; The modularized detection unit is arranged on the multi-degree-of-freedom motion platform which is arranged on the magnetic attraction rack assembly; The modularized detection unit comprises a pneumatic dust removal module and a detection module, the detection module comprises a laser displacement sensor, an optical imaging assembly and a synchronous trigger controller, and the synchronous trigger controller is configured to send synchronous trigger signals to the laser displacement sensor and the optical imaging assembly so as to realize the strict synchronous execution of laser sampling and optical image acquisition actions on the surface of a grinding wheel; The device comprises a laser displacement sensor, a pneumatic dust removal module, a laser displacement sensor, a fixed device, a blowing nozzle, an optical imaging assembly, a rigid bracket, a laser displacement sensor and a laser imaging assembly, wherein the pneumatic dust removal module and the laser displacement sensor are arranged on the multi-degree-of-freedom motion platform through the fixed device, and the distance between the blowing nozzle of the pneumatic dust removal module and the laser emission port of the laser displacement sensor meets a set distance so as to realize space cooperation of dust removal and detection; the master controller is in control connection with the multi-freedom-degree motion platform and is used for dynamically setting and executing a scanning strategy according to geometric parameters of the grinding wheel and controlling the multi-freedom-degree motion platform to move according to a set motion track; The master controller is in control connection with the laser displacement sensor, the pneumatic dust removal module and the optical imaging assembly, and is used for controlling the pneumatic dust removal module, the laser displacement sensor and the optical imaging assembly to cooperatively operate, realizing detection area pretreatment and multi-mode data synchronous acquisition, and carrying out quantitative evaluation and defect positioning on the surface appearance of the grinding wheel based on the multi-mode data.
2. The on-line grinding wheel surface detection device based on laser multi-mode collaborative detection according to claim 1, wherein the multi-degree-of-freedom motion platform comprises an X displacement adjustment assembly and a Z displacement adjustment assembly; The X-displacement adjusting assembly comprises an X-axis linear guide rail arranged on the magnetic attraction bench assembly and an X-axis driving assembly capable of horizontally reciprocating along the X-axis linear guide rail; The Z displacement adjusting assembly comprises a Z-axis linear guide rail arranged on the X-axis driving assembly and a Z-axis driving assembly capable of vertically and reciprocally moving along the Z-axis linear guide rail.
3. The online grinding wheel surface detection device based on laser multi-mode collaborative detection according to claim 1 or 2, wherein the distance between a purge air tap of the pneumatic dust removal module and a laser emission port of the laser displacement sensor is less than or equal to 50mm, and the angular interval between the optical imaging assembly and the laser emission direction of the laser displacement sensor is less than or equal to 5 degrees.
4. The online grinding wheel surface detection device based on laser multi-mode collaborative detection according to claim 3, wherein the pneumatic dust removal module is driven by stable air pressure above 0.2MPa and blows the surface of the grinding wheel by an air tap with an injection angle of 30 degrees plus or minus 2 degrees.
5. The detection method of the online grinding wheel surface detection equipment based on the laser multi-mode cooperative detection is characterized by comprising the following steps of: Controlling the multi-degree-of-freedom motion platform to move so as to adjust the position of the laser displacement sensor, so that the laser beam emitted by the laser displacement sensor is perpendicular to the end face of the grinding wheel to be detected; The pneumatic dust removal module is controlled to conduct directional blowing on the surface of the grinding wheel at a preset pressure and a preset angle, the laser displacement sensor is synchronously triggered to conduct real-time scanning on the surface of the grinding wheel to obtain laser point cloud data, and the optical imaging assembly is triggered to obtain RGB/near infrared optical images of the surface of the grinding wheel in real time; fusing the RGB/near infrared optical image and laser point cloud data through a space-time registration algorithm to generate three-dimensional point cloud model data with visual textures; and importing the three-dimensional point cloud model data into a visualization platform for display, and superposing visual features in the three-dimensional point cloud model to form a multi-dimensional detection result and outputting the multi-dimensional detection result.
6. The detection method of the online grinding wheel surface detection device based on the laser multi-mode collaborative detection according to claim 5, wherein before fusing RGB/near infrared optical images and laser point cloud data through a space-time registration algorithm to generate a three-dimensional point cloud model with visual textures, the following steps are executed: correcting polar coordinate data based on accumulated arc length calculation and spline interpolation by using an arc length resampling technology, and eliminating sampling angle errors caused by uneven rotation of a grinding wheel; Based on a preset laser displacement sensor and a grinding wheel coordinate system conversion model, converting corrected polar coordinate data into standard Cartesian three-dimensional coordinates in real time to generate an initial three-dimensional point cloud; Extracting a complete grinding wheel main body area according to target point groups selected by operator interaction or identified automatically, and leveling a reference surface of the selected area by adopting a least square method to eliminate installation inclination errors; The method comprises the steps of determining a dynamic height threshold value based on preset parameters, and preliminarily removing obvious outlier noise points, wherein the dynamic height threshold value is extracted from laser point cloud data based on the preset parameters, and the preset parameters comprise inner and outer diameters of a grinding wheel, rotating speed, scanning frequency, scanning radial speed, a displacement starting point and a displacement end point; setting an intelligent threshold value to filter abnormal reflection points, and removing residual noise by adopting a statistical outlier removing algorithm; and smoothing the three-dimensional morphology data through a spatial filtering algorithm to generate a three-dimensional geometric model of the grinding wheel.
7. The method for detecting the online grinding wheel surface detection equipment based on the laser multi-mode collaborative detection according to claim 6, wherein the step of fusing the RGB/near infrared optical image and the laser point cloud data through a space-time registration algorithm to generate a three-dimensional point cloud model with visual textures comprises the following steps: preprocessing an RGB/near infrared optical image, unifying a color space through graying conversion, and effectively inhibiting image noise by adopting median filtering; Carrying out space-time registration on the preprocessed image and the generated three-dimensional geometric model of the grinding wheel, precisely matching corresponding pixel information for each three-dimensional point, and endowing the three-dimensional point with color and texture attributes; And taking the fusion three-dimensional model obtained after space-time registration as a standardized multi-modal data set, and storing the standardized multi-modal data set in a standardized format.
8. The detection method of the on-line grinding wheel surface detection device based on the laser multi-mode collaborative detection according to claim 7, wherein the performing space-time registration of the preprocessed image and the generated three-dimensional geometric model of the grinding wheel, precisely matching corresponding pixel information for each three-dimensional point, and endowing the three-dimensional point with color and texture attributes comprises the following steps: calculating and extracting high-dimensional texture feature vectors which quantitatively represent surface roughness, directivity and uniformity from the preprocessed image based on a preset feature extraction method; establishing a spatial correspondence between an original optical image and a three-dimensional geometric model through a space-time registration algorithm; The color information of the original optical image is endowed to corresponding points of the three-dimensional geometric model; and synchronously mapping the data in the texture feature component diagram to corresponding points of the three-dimensional model as an additional attribute channel.
9. The detection method of the online grinding wheel surface detection device based on the laser multi-mode collaborative detection according to claim 8, wherein the method is characterized in that a three-dimensional point cloud model is imported into a visualization platform for display, visual features are superimposed in the three-dimensional point cloud model, and the method comprises the following steps: after the three-dimensional point cloud model is imported to a visualization platform, discrete three-dimensional point cloud data are converted into continuous surface cloud images based on voxel rendering and curved surface reconstruction technology, and color information and texture characteristic data of an optical image are synchronously mapped to the surface of the surface cloud images, so that a fused visualization model with complete color and material information is generated.
10. The detection method of the online grinding wheel surface detection device based on the laser multi-mode collaborative detection according to claim 9, wherein after the visual features are superimposed in the three-dimensional point cloud model, the following steps are further performed: Reconstructing three-dimensional point cloud data into a continuous curved surface model through a real-time triangle gridding and grid interpolation algorithm, performing pseudo-color mapping according to a height value or a local curvature calculation result, and synchronously generating a color height distribution cloud image and a curvature characteristic cloud image to form a visual reference capable of intuitively reflecting surface topology and geometric characteristics; Edge feature detection and curvature field quantification calculation are executed in parallel to locate contour step boundaries, groove side walls and fracture lines caused by abrasion and identify abnormal protrusions, microscopic pits and local plastic deformation areas; Establishing a feature fusion model of the edge intensity field and the curvature distribution field, carrying out region growth and connected domain analysis on the fusion feature map by adopting a self-adaptive double-threshold segmentation algorithm, and automatically classifying and extracting regions of interest conforming to different failure modes by setting geometric constraint conditions related to wear morphology; and marking the identified region of interest in the three-dimensional point cloud model in real time in a dynamic highlighting outline, a semitransparent mask or a three-dimensional highlighting rendering mode.

Description

Online grinding wheel surface detection equipment and method based on laser multi-mode collaborative detection Technical Field The invention relates to the technical field of ultra-precise machining, in particular to an on-line grinding wheel surface detection device and method based on laser multi-mode cooperative detection. Background In the field of high-end precision machining, the large-size grinding wheel is widely applied to key scenes such as ultra-precision machining of optical elements, high-precision grinding of ceramic parts, packaging of semiconductor chips and the like. As the length of the grinding wheel increases in use, the working surface thereof may undergo geometric changes due to wear, resulting in gradual deterioration of grinding performance and machining accuracy. If the worn grinding wheel is continuously used, the processing quality is directly affected, so that the periodic detection and the dressing of the grinding wheel are necessary links for guaranteeing the precision processing precision. The traditional flatness detection method relies on off-line operation, namely the grinding wheel needs to be disassembled from the main shaft of the machine tool and then moved to a special detection platform, and the single complete detection process takes more than several hours. The frequent disassembly not only seriously affects the production efficiency, but also causes accumulated damage to the precision of the spindle. The currently mainstream grinding wheel detection flow generally comprises the steps of grinding wheel disassembly, off-line geometric parameter and abrasion state detection, targeted dressing, reinstallation and dynamic balance adjustment. Although the detection technology is mature, the detection efficiency is low due to the inherent mode of off-line operation, and the production requirements of modern high-precision processing on real-time performance and continuity are difficult to meet, so that the detection technology becomes one of the key bottlenecks for restricting the development and efficiency improvement of the ultra-precision processing industry. At present, a sensor based on a laser triangulation principle is commonly adopted in the industry to detect the three-dimensional shape of the grinding wheel, and the method can efficiently acquire geometric profile data of the surface of the grinding wheel and accurately calculate key size parameters such as radial abrasion loss, profile error and the like. However, a single geometrical measurement has significant limitations in that, first, laser measurements cannot penetrate or effectively identify oil stains, coolant residues and fine dust adhering to the surface of the grinding wheel, which can be recorded as false heights, severely distorting the real wear data. This not only results in misalignment of the wear assessment, but may also lead to erroneous maintenance decisions. To make up for the lack of geometric information, machine vision techniques are introduced into the inspection system. Conventional visible light imaging can provide visual records of macroscopic texture, color and obvious defects on the surface of the grinding wheel, but the imaging quality is easily affected by field illumination conditions, oil mist interference and surface reflection, and has limited identification capability on pollutants or early microcracks similar to the color and reflection characteristics of a matrix. In recent years, near infrared imaging technology has begun to be explored for industrial detection. Since certain materials have different absorption and reflection properties in the near infrared band than visible light, the technique is capable of effectively penetrating portions of the oil mist, enhancing the contrast of certain residues or materials, thereby revealing surface features that are imperceptible under visible light, such as transparent cooling liquid films, microscopic thermal damage areas, and the like. However, existing detection schemes mostly perform the geometric measurement and visual imaging as separate or simple sequential steps, lacking depth information fusion. The data island mode causes two core problems that firstly, time sequence is asynchronous, the geometric data and the image data which are collected sequentially are mismatched in space due to tiny vibration or state change of a grinding wheel in the rotating process, and accurate correlation at a pixel level cannot be achieved, secondly, the analysis dimension is split, a system cannot automatically correlate and analyze a depth abnormality at a certain position with a visual texture feature at the position, so that whether the abnormality is material abrasion, pollutant coverage or physical collision is difficult to judge, and intelligent diagnosis capability of the system is limited. Therefore, an online detection method and system capable of synchronously acquiring and precisely fusing three-dimensional geometric information