Intelligent grinding wheel laser finishing system and method

Abstract

An intelligent grinding wheel laser finishing system belongs to the field of grinding wheel finishing and comprises an input end, a machining system and an output end. The input end inputs grinding wheel data into the machining system, and the machining system is composed of a database, a CAM system and a monitoring system. The output end obtains products generated by the machining system. The method for trimming the grinding wheel through the laser comprises the steps that 1, grinding wheel data are input into the system; 2, a monitoring system automatically generates tool setting point and allowance information; 3, optimizing machining process parameters; 4, calculating a laser beam attitude adjustment mode and a laser energy compensation mode, and generating a trajectory plan; 5, performing interpolation motion, and trimming the grinding wheel; 6, the machining process is monitored, and allowance changes are observed; 7, after the allowance is met, the monitoring system sends out shutdown detection information; according to the method, laser beam posture adjustment and energy compensation can be carried out according to the outline dimension of the grinding wheel blank in the drawing design and finishing process, and autonomous iteration of the whole process can be achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of grinding wheel dressing, and in particular relates to an intelligent system and method for grinding wheel laser dressing. Background Art

[0002] Currently, laser dressing of grinding wheels is highly dependent on manual labor, including laser parameter setting, grinding wheel material analysis, processing parameter determination, some dressing steps (such as tool setting), dressing automation and completion judgment, loading and unloading, etc.

[0003] The intelligent processing system focuses on manual substitution, including manual operation, process experience judgment, loading and unloading, etc. The difference between this and traditional automation is that automated manual substitution mainly replaces manual operation, such as automatic feed, automatic interpolation, etc. Intelligence is different. It uses data analysis from the database to form the selection of empirical parameters. It uses visual, sound, thermal, electromagnetic, force and other judgments to monitor the entire processing process. It uses the CAM core algorithm to plan and customize the processing grinding wheel. Therefore, intelligence is not a simple upgrade in automation, but a comprehensive replacement of manual processing, including automation.

[0004] The laser dressing system for formed grinding wheels is currently in the process trial stage and has not formed an integrated, automated and intelligent processing system. Summary of the invention

[0005] The purpose of the present invention is to propose an intelligent system and method for laser dressing of grinding wheels. The intelligent system can adjust the laser beam posture and energy according to the contour requirements of the drawing design and the real-time changing external dimensions of the grinding wheel blank during the dressing process. It can also realize autonomous iteration of the whole process and iteratively upgrade the entire processing system. The method realizes automatic generation of trajectory planning by determining dressing parameters, compensating for defocus and avoiding laser interference.

[0006] The technical solution adopted by the present invention is:

[0007] An intelligent system for laser dressing of grinding wheels, comprising an input end, a processing system and an output end;

[0008] The input end inputs the grinding wheel data into the processing system according to product requirements, and the grinding wheel data includes the contour requirement parameters and model designed on the drawing, as well as the material information and external dimensions of the grinding wheel blank; the drawing is in digital form, and the material information and external dimensions of the grinding wheel blank are obtained using various sensors, cameras and other hardware facilities on the machine tool.

[0009] The processing system is composed of a database, a CAM system and a monitoring system; the database can realize four functions: expert prediction, parameter optimization, digital twin and machine learning; the CAM system receives the processing parameters from the database, and performs two functions: laser beam posture adjustment and energy compensation based on the processing parameters; the functions of the monitoring system include judging the allowance of grinding wheel dressing, identifying tool points and judging abnormal situations. On the one hand, the monitoring system feeds back the real-time processing data to the database to help upgrade the database, and on the other hand, it interacts with the CAM to output the real-time processing position and allowance information required by the CAM system, and accepts the follow-up control of the CAM system; during the operation of the processing system, the autonomous iteration of the entire process is realized, and the entire processing system is iteratively upgraded;

[0010] The output end obtains the products produced by the processing system, including qualified grinding wheels and test reports, and outputs qualified grinding wheels (customers), processing test results (customers), and processing cost consumption (factory).

[0011] A method for intelligent laser dressing of a grinding wheel comprises the following steps:

[0012] Step 1: Input the grinding wheel data into the grinding wheel laser dressing intelligent system through the input end;

[0013] Step 2: The monitoring system compares and analyzes the outer dimensions of the grinding wheel blank and the contour requirement parameters of the drawing, and then automatically generates tool setting points and allowance information;

[0014] The monitoring system mainly analyzes the actual dimensions of the blank, and then compares the two, which is equivalent to blindly looking for differences.

[0015] Step 3: Compare the material information and dimensions of the grinding wheel blank with the dressing cases stored in the database to find similar or consistent dressing cases, and then optimize the processing parameters through expert prediction and parameter optimization functions; if there are no similar cases, perform machine learning through known cases to generate a set of processing parameters; the generated processing parameters are prepared for digital twins;

[0016] Step 4: Send the processing parameters generated by the database to the CAM system. The CAM system calculates the laser beam posture adjustment method and laser energy compensation method according to the grinding wheel profile requirements and processing parameters, and generates the trajectory planning of the laser processing process;

[0017] Step 5: The machine tool performs interpolation motion according to the trajectory planning generated by the CAM system and starts dressing the grinding wheel;

[0018] Step 6: After the interpolation motion starts, the monitoring system monitors the machining process and observes the change of the blank's allowance in real time. Then, through template matching and target recognition, the real-time allowance and grinding wheel blank position information are sent to the CAM system and database;

[0019] The CAM system uses the position information of the allowance and blank to compensate for abnormal processing points, and the database uses the real-time position and allowance to perform digital twin representation;

[0020] Step 7: After the margin meets the processing requirements, the monitoring system sends a message, stops the machine, and performs inspection;

[0021] Step 8: Output the test results. If the product is qualified, the qualified product and test report are output; if it is unqualified, re-analyze the grinding wheel information and plan the next round of dressing;

[0022] Step 9: If the special grinding wheel is difficult to pass after repeated dressing, use the system iteration function to optimize the system database and find more optimized parameters.

[0023] Furthermore, in step 2, the monitoring system automatically generates tool setting point and margin information according to the following method:

[0024] For the judgment of grinding wheel blanks: judge the size of the blank machining allowance by visual observation and checking the allowance;

[0025] For tool point detection: use the trial burning method, and then judge the position of the molten pool and the laser tool tip based on visual observation; when using the mechanical method, judge the tool depth and tool position through sound sensing and vision;

[0026] Monitoring of the machining process: determine whether the energy absorption of the machining points is uniform based on the energy of the laser optical spectrometer; compare the actual contour observed by visual observation with the contour required by the drawing, and determine whether the machining is completed based on the difference between the two; determine the circularity, runout, and contour accuracy of the finished product based on visual observation and edge detection synthesis algorithm; determine the operating status of the entire equipment based on sound and current.

[0027] The monitoring system is a full-process monitoring system. It needs to be assembled by yourself according to different methods such as vision, optics, and acoustics. Common methods include direct visual observation and backlighting. Optical methods such as spectrometer detection and sound vibration sensor detection for tool setting. It is used to judge the remaining amount of grinding wheel blanks, tool setting point detection, abnormal processing point judgment, shutdown judgment after processing, finished product detection, and full equipment monitoring.

[0028] Furthermore, in step 3, the particle swarm PSO algorithm is used to optimize the processing parameters, and the algorithm includes the following steps:

[0029] S301, using a random method to initialize the position and speed of the particle swarm; including the historical optimal position of a single particle and the historical optimal position of the swarm, the historical optimal fitness value of a single particle and the historical optimal fitness value of the swarm;

[0030] S302, update the position and velocity of a single particle using formulas (1) and (2):

[0031]

[0032] The distance and direction of each particle's next step in the PSO algorithm is a position vector, called the speed of the next iteration, expressed as:

[0033]

[0034] in, represents the position vector of particle i in the dth dimension at the k+1th iteration; represents the position vector of particle i in the dth dimension in the kth iteration; represents the velocity vector of the dth dimension of particle i in the k+1th iteration; represents the velocity vector of particle i in the dth dimension in the kth iteration; ω represents the inertia weight; c1 and c2 represent learning factors; r1 and r2 represent two random numbers, ranging from 0 to 1; represents the historical optimal position of particle i in the dth dimension in the kth iteration, that is, after the kth iteration, the optimal solution obtained by the i-th particle (individual) in the dth dimension search; It represents the historical optimal position of the group (globally) in the dth dimension in the kth iteration, that is, the optimal solution of the entire particle group in the dth dimension after the kth iteration;

[0035] S303, calculate the fitness value of a single particle, and the fitness value must meet the requirements of formula (3):

[0036]

[0037] Among them, η m represents the fitness value of each particle; m represents the particle number; the fitness value represents the optimization target, and if the fitness values ​​of each particle meet the tolerance requirements, it can be considered that the fitting result has a good detection credibility;

[0038] S304, updating the historical best fitting value of a single particle;

[0039] S305, updating the historical best fitting value and position of the group;

[0040] S306, update the inertia weight and the number of iterations:

[0041] Using the linearly changing inertia weight strategy, the inertia weight is updated as follows (4):

[0042]

[0043] Among them, ω max and ω min They represent the maximum inertia weight and the minimum inertia weight respectively; iter represents the current number of iterations; iter max Indicates the maximum number of iterations;

[0044] S307, when the terminal condition is met, output the final optimal solution, otherwise repeat steps S302 to S306.

[0045] Furthermore, in step 5, the process of generating trajectory planning by the CAM system includes the following steps:

[0046] The CAM trajectory planning system is used to plan and adjust the multi-pass optimization of the laser trajectory during dressing, real-time posture adjustment, energy compensation, abnormal processing point compensation and other issues. The system needs to be established by itself, because there is no CAM software, algorithm, or planning system for grinding wheel laser dressing on the market. These adjustments need to provide data on the grinding wheel material processing process based on the database, such as single-pass ablation speed, the relationship between the inclination angle and the forming angle, the corresponding relationship between the grinding wheel diameter-the original laser cutting depth-the laser beam divergence angle, the effect of the feed rate on the single-pass ablation rate, etc., and make trajectory planning based on this.

[0047] S501, trimming parameters are determined:

[0048] In tangential dressing, the laser spot is irradiated onto the grinding wheel surface in a pulsed manner, and the shape of the spot on the grinding wheel surface is approximately an ellipse; according to literature and industry consensus, laser dressing of grinding wheels uses tangential shaping and radial sharpening. For accuracy and forming, only tangential dressing is considered.

[0049] The spot overlap rate in the circumferential direction of the grinding wheel is defined as: the ratio of the spot overlap area to the spot area under two pulses, expressed as Oc, and Oc is described as:

[0050]

[0051] Among them, S overlap Indicates the area of ​​the overlapping part of the light spot; S ellipse Indicates the spot area; a e and b e Indicates the parameters of the ellipse; l c Indicates the offset of the center of the light spot in the circumferential direction;

[0052] The spot overlap rate must reach more than 85% to obtain a better grinding wheel surface quality; the grinding wheel speed when the spot overlap rate reaches 85% is given by formula (6):

[0053]

[0054] Where D is the diameter of the grinding wheel; T represents the pulse period of the laser, which is the inverse of the repetition frequency;

[0055] The spot overlap rate in the direction of the grinding wheel generatrix is ​​defined as the ratio of the spot overlap area to the spot area, expressed as A Indicates that O A Described as:

[0056]

[0057] Among them, l A is the offset of the center of the light spot on the busbar;

[0058] Then, at the rotation speed n, the grinding wheel feed rate F is calculated by formula (8):

[0059] F = nl A (8)

[0060] The trimming parameters can be determined by formula (5) to formula (8);

[0061] S502, defocus compensation:

[0062] The edge of the grinding wheel is taken as the origin of the coordinates, the generatrix direction of the grinding wheel cylinder is taken as the X direction, and the diameter direction is taken as the Y direction and the Z direction; before the grinding wheel is dressed, its surface cylindrical annulus is described as:

[0063]

[0064] The mathematical model of defocus compensation is constructed by describing the trajectory of the laser movement through formula (9). In the formula, x, y, and z are the three directions after the coordinate system is established, and the origin of the coordinate is at the edge of the grinding wheel; B is the thickness of the grinding wheel (equivalent to the generatrix length of the cylinder);

[0065] The spot is irradiated on the surface of the material, generally within the Rayleigh length range, and the laser beam can be considered cylindrical at this time. During the material removal process, the vertically irradiated laser beam needs to rotate in the XOZ plane with the laser beam focus O as the rotation center. The state of rotation angle α is described as:

[0066]

[0067] Where r represents the radius of the laser beam at a certain position; h represents the distance from the position to the laser focus; r0 represents the radius of the spot at the laser focus (h = 0 position). The laser beam adopts the near-field cylindrical model and r is always consistent with r0; L ray represents the Rayleigh length of the laser beam, and e0 is the initial defocusing amount of the laser beam in the vertical state.

[0068] For vertically irradiated laser, the focus is on the XOY plane. As the laser beam rotates, the intersection of the laser beam and the grinding wheel (i.e. the actual irradiation position) P T In order to maintain a stable defocus, compensation needs to be made in the X and Z directions. The compensation amount is:

[0069]

[0070] When defocus compensation is performed, it is assumed that the focus always remains on the XOY plane.

[0071] S503, Avoidance of laser interference:

[0072] When the point of action between the laser beam and the material exceeds the Rayleigh length, the laser beam is considered to be a far-field laser composed of a hyperbola;

[0073] In the far-field model, using the asymptotic model, the radius value r at a certain height h in the beam direction is b It can be roughly described as:

[0074]

[0075] in, is the divergence angle of the laser beam; h b is the beam length;

[0076] In the vertical direction, the beam length h b The following geometric relationship is satisfied between the height h:

[0077] h b =h / cosα h (13)

[0078] Among them, α h Indicates the angle of rotation of the A axis when the laser is half blocked;

[0079] The shape profile of the wafer chamfering grinding wheel is described by the following piecewise function (the contour shape of the wafer chamfering grinding wheel can be understood as a combination of straight lines and arcs):

[0080]

[0081] Among them, K srepresents the intercept of the hypotenuse, (x r ,y r ) represents the coordinates of the center of the arc, which changes with the arc number; θ represents the angle between the contour line and the horizontal direction; y represents the coordinate system position;

[0082] During the grinding wheel ablation process, the laser tip needs a certain cutting depth. Assuming that the cutting depth is fixed during processing, the actual position of the laser and the grinding wheel will be raised by the circular effect of the grinding wheel. Therefore, the contour of the actual effect of the laser spot should be obtained by deforming the forming contour (in the actual processing process, the laser spot is not in the absolute tangent direction, but has a certain cutting depth. In the previous method, it was believed that the laser interpolation trajectory is the drawing contour trajectory. However, the present invention found through theoretical calculation that the cutting depth caused the laser spot to be raised on the grinding wheel surface, causing the removal point trajectory to be deformed relative to the laser interpolation trajectory). The height Δh of the raised part is obtained by the following geometric relationship:

[0083]

[0084] in is the radius of the grinding wheel; a represents the cutting depth of the laser tool tip;

[0085] At a height of Δh, the shape profile y of the forming grinding wheel h Transformed into:

[0086]

[0087] When the laser beam deflection angle (the deflection used in this method, the laser beam deflection center, the A-axis rotation center and the laser beam focus are coincident, so the rotation angle is consistent. The laser head is on the A-axis, and the laser beam deflection angle is the angle of rotation of the A-axis) continues to increase, the laser beam will intersect with the outer circle of the grinding wheel, thereby causing the outer circle to block the laser light. At this time, the angle α h The calculation is done using the following formula:

[0088]

[0089] In the formula, α h It indicates the angle of rotation of the A axis when the laser is half blocked; because from experience, when the laser beam is half blocked, the energy is insufficient to continue processing, and it is considered that the shielding effect exists completely at this time. Therefore, αh can be considered as the position of the center line of the beam participating in the shielding effect. L t represents the width of the groove, D represents the diameter, x l Indicates the distance from a certain position to the center of the groove, mainly to facilitate mathematical calculations of the arc part. It can also be changed to x position; y is the y-axis position;

[0090] When the light blocking effect occurs, the corresponding critical angle α2 is described as:

[0091]

[0092] After the laser beam intersects the outer circle of the grinding wheel, the laser beam cannot penetrate the grinding wheel material and continue to irradiate the contour surface of the grinding wheel. Therefore, part of the laser energy is dispersed to the outer circle surface that should not be processed, and the forming contour cannot be copied.

[0093] The critical angle α2 of the grinding wheel blocking light is a series of critical values ​​that follow the change of x. For a certain point on the grinding wheel forming profile (except the outer cylindrical surface), there is always a laser blocking light critical angle α2, so that when the laser beam deflection angle is greater than the critical angle α2, the laser is blocked by the outer cylindrical surface of the forming grinding wheel, and the dressing margin in the forming profile cannot be effectively removed; so the condition for laser dressing of the forming grinding wheel without blocking light is: the laser deflection angle, at a certain point on the profile, is less than the critical angle of the light blocking at that point, that is:

[0094] α(x)<α2(x) (19)

[0095] In the α2 range, the laser power density drops sharply because the laser is shielded, and the material cannot be removed. Therefore, when profiling the grinding wheel, different deflection angles α need to be used at different contours.

[0096] It can be seen that different laser deflection angles can process different positions on the arc of the grinding wheel. In the actual processing technology, the deflection resolution M required for processing the arc is calculated using the following formula:

[0097]

[0098] Among them, K CW is a parameter, which takes an integer; L g Indicates the length of the secant line of the arc contour, L α is the arc ablation length corresponding to the critical angle of bevel ablation.

[0099] Beneficial effects of the present invention:

[0100] The intelligent system for laser dressing of grinding wheels of the present invention can adjust the attitude of the laser beam and perform energy compensation according to the contour requirements of the design drawings and the external dimensions of the grinding wheel blank that change in real time during the dressing process. It can also realize autonomous iteration of the entire process and iteratively upgrade the entire processing system. The method realizes automatic generation of trajectory planning by determining dressing parameters, compensating for defocus and avoiding laser interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 It is the overall technical scheme diagram of the intelligent system for laser dressing of grinding wheels of the present invention;

[0102] Figure 2 It is a schematic diagram of the monitoring system monitoring the trimming process;

[0103] Figure 3 This is a schematic diagram of interference under the laser far-field model. DETAILED DESCRIPTION

[0104] like Figure 1 As shown, an intelligent system for laser dressing of grinding wheels includes an input end, a processing system and an output end;

[0105] The input end inputs the grinding wheel data into the processing system according to product requirements, and the grinding wheel data includes the contour requirement parameters and model designed on the drawing, as well as the material information and external dimensions of the grinding wheel blank; the drawing is in digital form, and the material information and external dimensions of the grinding wheel blank are obtained using various sensors, cameras and other hardware facilities on the machine tool.

[0106] The processing system is composed of a database, a CAM system and a monitoring system; the database can realize four functions: expert prediction, parameter optimization, digital twin and machine learning; the CAM system receives the processing parameters from the database, and performs two functions: laser beam posture adjustment and energy compensation based on the processing parameters; the functions of the monitoring system include judging the allowance of grinding wheel dressing, identifying tool points and judging abnormal situations. On the one hand, the monitoring system feeds back the real-time processing data to the database to help upgrade the database, and on the other hand, it interacts with the CAM to output the real-time processing position and allowance information required by the CAM system, and accepts the follow-up control of the CAM system; during the operation of the processing system, the autonomous iteration of the entire process is realized, and the entire processing system is iteratively upgraded;

[0107] The output end obtains the products produced by the processing system, including qualified grinding wheels and test reports, and outputs qualified grinding wheels (customers), processing test results (customers), and processing cost consumption (factory).

[0108] A method for intelligent laser dressing of a grinding wheel comprises the following steps:

[0109] Step 1: Input the grinding wheel data into the grinding wheel laser dressing system through the input end;

[0110] Step 2: The monitoring system compares and analyzes the outer dimensions of the grinding wheel blank and the contour requirement parameters of the drawing, and then automatically generates tool setting points and allowance information;

[0111] The monitoring system mainly analyzes the actual dimensions of the blank, and after analysis, compares the two. It is equivalent to blindly looking for differences. Figure 2 The monitoring system automatically generates tool setting points and margin information according to the following methods:

[0112] For the judgment of grinding wheel blanks: judge the size of the blank machining allowance by visual observation and checking the allowance;

[0113] For tool point detection: use the trial burning method, and then judge the position of the molten pool and the laser tool tip based on visual observation; when using the mechanical method, judge the tool depth and tool position through sound sensing and vision;

[0114] Monitoring of the machining process: determine whether the energy absorption of the machining points is uniform based on the energy of the laser optical spectrometer; compare the actual contour observed by visual observation with the contour required by the drawing, and determine whether the machining is completed based on the difference between the two; determine the circularity, runout, and contour accuracy of the finished product based on visual observation and edge detection synthesis algorithm; determine the operating status of the entire equipment based on sound and current.

[0115] The monitoring system is a full-process monitoring system. It needs to be assembled by yourself according to different methods such as vision, optics, and acoustics. Common methods include direct visual observation and backlighting. Optical methods such as spectrometer detection and sound vibration sensor detection for tool setting. It is used to judge the remaining amount of grinding wheel blanks, tool setting point detection, abnormal processing point judgment, shutdown judgment after processing, finished product detection, and full equipment monitoring.

[0116] Step 3: Compare the material information and dimensions of the grinding wheel blank with the dressing cases stored in the database to find similar or consistent dressing cases, and then optimize the processing parameters through expert prediction and parameter optimization functions; if there are no similar cases, perform machine learning through known cases to generate a set of processing parameters; the generated processing parameters are prepared for digital twins;

[0117] The particle swarm PSO algorithm is used to optimize the machining process parameters. The algorithm includes the following steps:

[0118] The database requires a large amount of grinding wheel dressing data to make judgments on grinding wheel dressing process parameters. The judgment methods use prediction systems, parameter classification, parameter optimization, machine learning, etc. to complete the three tasks of laser parameter setting, grinding wheel material analysis, and processing parameter determination during the processing. The laser parameter setting includes laser power (10-200W), repetition frequency (50-200kHz), pulse width (100fs-200ns), wavelength (100-1064nm), etc. The grinding wheel material includes metal, resin, ceramic, etc. and their composite binders, and non-homogeneous materials of diamond, CBN and other abrasives. The processing technology includes pure laser methods such as galvanometer scanning, interpolation dressing, deflected laser, Deep-Cutting, pure EDM methods such as EDM, mixed powder EDM, dry EDM, as well as mechanical methods such as mechanical profiling and mechanical forming, and also includes the composite methods of these processing technologies. The parameters used in all these processing technologies are processing technology parameters.

[0119] The database-based expert system is an independent library built by itself. The expert system's data induction and mining needs to be based on the particle swarm optimization algorithm, because the fast convergence of the particle swarm optimization algorithm can meet the real-time and low-latency requirements of the trimming system.

[0120] S301, using a random method to initialize the position and speed of the particle swarm; including the historical optimal position of a single particle and the historical optimal position of the swarm, the historical optimal fitness value of a single particle and the historical optimal fitness value of the swarm;

[0121] S302, update the position and velocity of a single particle using formulas (1) and (2):

[0122]

[0123] The distance and direction of each particle's next step in the PSO algorithm is a position vector, called the speed of the next iteration, expressed as:

[0124]

[0125] in, represents the position vector of particle i in the dth dimension at the k+1th iteration; represents the position vector of particle i in the dth dimension in the kth iteration; represents the velocity vector of the dth dimension of particle i in the k+1th iteration; represents the velocity vector of particle i in the dth dimension in the kth iteration; ω represents the inertia weight; c1 and c2 represent learning factors; r1 and r2 represent two random numbers, ranging from 0 to 1; represents the historical optimal position of particle i in the dth dimension in the kth iteration, that is, after the kth iteration, the optimal solution obtained by the i-th particle (individual) in the dth dimension search; It represents the historical optimal position of the group (globally) in the dth dimension in the kth iteration, that is, the optimal solution of the entire particle group in the dth dimension after the kth iteration;

[0126] S303, calculate the fitness value of a single particle, and the fitness value must meet the requirements of formula (3):

[0127]

[0128] Among them, η m represents the fitness value of each particle; m represents the particle number; the fitness value represents the optimization target, and if the fitness values ​​of each particle meet the tolerance requirements, it can be considered that the fitting result has a good detection credibility;

[0129] S304, updating the historical best fitting value of a single particle;

[0130] S305, updating the historical best fitting value and position of the group;

[0131] S306, update the inertia weight and the number of iterations:

[0132] Using the linearly changing inertia weight strategy, the inertia weight is updated as follows (4):

[0133]

[0134] Among them, ω max and ω min They represent the maximum inertia weight and the minimum inertia weight respectively; iter represents the current number of iterations; iter max Indicates the maximum number of iterations;

[0135] S307, when the terminal condition is met, output the final optimal solution, otherwise repeat steps S302 to S306.

[0136] Step 4: Send the processing parameters generated by the database to the CAM system. The CAM system calculates the laser beam posture adjustment method and laser energy compensation method according to the grinding wheel profile requirements and processing parameters, and generates the trajectory planning of the laser processing process;

[0137] Step 5: The machine tool performs interpolation motion according to the trajectory planning generated by the CAM system and starts to dress the grinding wheel. The process of the CAM system generating the trajectory planning includes the following steps:

[0138] The CAM trajectory planning system is used to plan and adjust the multi-pass optimization of the laser trajectory during dressing, real-time posture adjustment, energy compensation, abnormal processing point compensation and other issues. The system needs to be established by itself, because there is no CAM software, algorithm, or planning system for grinding wheel laser dressing on the market. These adjustments need to provide data on the grinding wheel material processing process based on the database, such as single-pass ablation speed, the relationship between the inclination angle and the forming angle, the corresponding relationship between the grinding wheel diameter-the original laser cutting depth-the laser beam divergence angle, the effect of the feed rate on the single-pass ablation rate, etc., and make trajectory planning based on this.

[0139] S501, trimming parameters are determined:

[0140] In tangential dressing, the laser spot is irradiated onto the grinding wheel surface in a pulsed manner, and the shape of the spot on the grinding wheel surface is approximately an ellipse; according to literature and industry consensus, laser dressing of grinding wheels uses tangential shaping and radial sharpening. For accuracy and forming, only tangential dressing is considered.

[0141] The spot overlap rate in the circumferential direction of the grinding wheel is defined as: the ratio of the spot overlap area to the spot area under two pulses, expressed as Oc, and Oc is described as:

[0142]

[0143] Among them, S overlap Indicates the area of ​​the overlapping part of the light spot; S ellipse Indicates the spot area; a e and b e Indicates the parameters of the ellipse; l c Indicates the offset of the center of the light spot in the circumferential direction;

[0144] The spot overlap rate must reach more than 85% to obtain a better grinding wheel surface quality; the grinding wheel speed when the spot overlap rate reaches 85% is given by formula (6):

[0145]

[0146] Where D is the diameter of the grinding wheel; T represents the pulse period of the laser, which is the inverse of the repetition frequency;

[0147] The spot overlap rate in the direction of the grinding wheel generatrix is ​​defined as the ratio of the spot overlap area to the spot area, expressed as A Indicates that O A Described as:

[0148]

[0149] Among them, l A is the offset of the center of the light spot on the busbar;

[0150] Then, at the rotation speed n, the grinding wheel feed rate F is calculated by formula (8):

[0151] F = nl A (8)

[0152] The trimming parameters can be determined by formula (5) to formula (8);

[0153] S502, defocus compensation:

[0154] The edge of the grinding wheel is taken as the origin of the coordinates, the generatrix direction of the grinding wheel cylinder is taken as the X direction, and the diameter direction is taken as the Y direction and the Z direction; before the grinding wheel is dressed, its surface cylindrical annulus is described as:

[0155]

[0156] The mathematical model of defocus compensation is constructed by describing the trajectory of the laser movement through formula (9). In the formula, x, y, and z are the three directions after the coordinate system is established, and the origin of the coordinate is at the edge of the grinding wheel; B is the thickness of the grinding wheel (equivalent to the generatrix length of the cylinder);

[0157] The spot is irradiated on the surface of the material, generally within the Rayleigh length range, and the laser beam can be considered cylindrical at this time. During the material removal process, the vertically irradiated laser beam needs to rotate in the XOZ plane with the laser beam focus O as the rotation center. The state of rotation angle α is described as:

[0158]

[0159] Where r represents the radius of the laser beam at a certain position; h represents the distance from the position to the laser focus; r0 represents the radius of the spot at the laser focus (h = 0 position). The laser beam adopts the near-field cylindrical model and r is always consistent with r0; L ray represents the Rayleigh length of the laser beam, and e0 is the initial defocusing amount of the laser beam in the vertical state.

[0160] For vertically irradiated laser, the focus is on the XOY plane. As the laser beam rotates, the intersection of the laser beam and the grinding wheel (i.e. the actual irradiation position) P T In order to maintain a stable defocus, compensation needs to be made in the X and Z directions. The compensation amount is:

[0161]

[0162] When defocus compensation is performed, it is assumed that the focus always remains on the XOY plane.

[0163] S503, Avoidance of laser interference:

[0164] like Figure 3 As shown in Figure 1, when the point of action between the laser beam and the material exceeds the Rayleigh length, the laser beam is considered to be a far-field laser composed of a hyperbola.

[0165] In the far-field model, using the asymptotic model, the radius value r at a certain height h in the beam direction is b It can be roughly described as:

[0166]

[0167] in, is the divergence angle of the laser beam; h b is the beam length;

[0168] In the vertical direction, the beam length h b The following geometric relationship is satisfied between the height h:

[0169] h b =h / cosα h (13)

[0170] Among them, α h Indicates the angle of rotation of the A axis when the laser is half blocked;

[0171] The shape profile of the wafer chamfering grinding wheel is described by the following piecewise function (the contour shape of the wafer chamfering grinding wheel can be understood as a combination of straight lines and arcs):

[0172]

[0173] Among them, K s represents the intercept of the hypotenuse, (x r ,y r ) represents the coordinates of the center of the arc, which changes with the arc number; θ represents the angle between the contour line and the horizontal direction; y represents the coordinate system position;

[0174] During the grinding wheel ablation process, the laser tip needs a certain cutting depth. Assuming that the cutting depth is fixed during processing, the actual position of the laser and the grinding wheel will be raised by the circular effect of the grinding wheel. Therefore, the contour of the actual effect of the laser spot should be obtained by deforming the forming contour (in the actual processing process, the laser spot is not in the absolute tangent direction, but has a certain cutting depth. In the previous method, it was believed that the laser interpolation trajectory is the drawing contour trajectory. However, the present invention found through theoretical calculation that the cutting depth caused the laser spot to be raised on the grinding wheel surface, causing the removal point trajectory to be deformed relative to the laser interpolation trajectory). The height Δh of the raised part is obtained by the following geometric relationship:

[0175]

[0176] in is the radius of the grinding wheel; a represents the cutting depth of the laser tool tip;

[0177] At a height of Δh, the shape profile y of the forming grinding wheel h Transformed into:

[0178]

[0179] When the laser beam deflection angle (the deflection used in this method, the laser beam deflection center, the A-axis rotation center and the laser beam focus are coincident, so the rotation angle is consistent. The laser head is on the A-axis, and the laser beam deflection angle is the angle of rotation of the A-axis) continues to increase, the laser beam will intersect with the outer circle of the grinding wheel, thereby causing the outer circle to block the laser light. At this time, the angle α h The calculation is done using the following formula:

[0180]

[0181] In the formula, α h It indicates the angle of rotation of the A axis when the laser is half blocked; because from experience, when the laser beam is half blocked, the energy is insufficient to continue processing, and it is considered that the shielding effect exists completely at this time. Therefore, αh can be considered as the position of the center line of the beam participating in the shielding effect. L t represents the width of the groove, D represents the diameter, x l Indicates the distance from a certain position to the center of the groove, mainly to facilitate mathematical calculations of the arc part. It can also be changed to x position; y is the y-axis position;

[0182] When the light blocking effect occurs, the corresponding critical angle α2 is described as:

[0183]

[0184] After the laser beam intersects the outer circle of the grinding wheel, the laser beam cannot penetrate the grinding wheel material and continue to irradiate the contour surface of the grinding wheel. Therefore, part of the laser energy is dispersed to the outer circle surface that should not be processed, and the forming contour cannot be copied.

[0185] The critical angle α2 of the grinding wheel blocking light is a series of critical values ​​that follow the change of x. For a certain point on the grinding wheel forming profile (except the outer cylindrical surface), there is always a laser blocking light critical angle α2, so that when the laser beam deflection angle is greater than the critical angle α2, the laser is blocked by the outer cylindrical surface of the forming grinding wheel, and the dressing margin in the forming profile cannot be effectively removed; so the condition for laser dressing of the forming grinding wheel without blocking light is: the laser deflection angle, at a certain point on the profile, is less than the critical angle of the light blocking at that point, that is:

[0186] α(x)<α2(x) (19)

[0187] In the α2 range, the laser power density drops sharply because the laser is shielded, and the material cannot be removed. Therefore, when profiling the grinding wheel, different deflection angles α need to be used at different contours.

[0188] It can be seen that different laser deflection angles can process different positions on the arc of the grinding wheel. In the actual processing technology, the deflection resolution M required for processing the arc is calculated using the following formula:

[0189]

[0190] Among them, K CW is a parameter, which takes an integer; L g Indicates the length of the secant line of the arc contour, L α is the arc ablation length corresponding to the critical angle of bevel ablation.

[0191] Step 6: After the interpolation motion starts, the monitoring system monitors the machining process and observes the change of the blank's allowance in real time. Then, through template matching and target recognition, the real-time allowance and grinding wheel blank position information are sent to the CAM system and database;

[0192] The CAM system uses the position information of the allowance and blank to compensate for abnormal processing points, and the database uses the real-time position and allowance to perform digital twin representation;

[0193] Step 7: After the margin meets the processing requirements, the monitoring system sends a message, stops the machine, and performs inspection;

[0194] Step 8: Output the test results. If the product is qualified, the qualified product and test report are output; if it is unqualified, re-analyze the grinding wheel information and plan the next round of dressing;

[0195] Step 9: If the special grinding wheel is difficult to pass after repeated dressing, use the system iteration function to optimize the system database and find more optimized parameters.

Claims

1. An intelligent system for laser dressing of grinding wheels, characterized in that: It includes input end, processing system and output end; The input end inputs the grinding wheel data into the processing system according to the product requirements, wherein the grinding wheel data includes the contour requirement parameters and model designed on the drawing, and the material information and the outer dimensions of the grinding wheel blank; The processing system is composed of a database, a CAM system and a monitoring system; the database can realize four functions: expert prediction, parameter optimization, digital twin and machine learning; the CAM system receives the processing parameters from the database, and performs two functions: laser beam posture adjustment and energy compensation based on the processing parameters; the functions of the monitoring system include judging the allowance of grinding wheel dressing, identifying tool points and judging abnormal situations. On the one hand, the monitoring system feeds back the real-time processing data to the database to help upgrade the database, and on the other hand, it interacts with the CAM to output the real-time processing position and allowance information required by the CAM system, and accepts the follow-up control of the CAM system; during the operation of the processing system, the autonomous iteration of the entire process is realized, and the entire processing system is iteratively upgraded; The output end obtains the products produced by the processing system, including qualified grinding wheels and test reports.

2. A method for intelligent laser dressing of grinding wheels, characterized in that: The following steps are involved: Step 1: Input the grinding wheel data into the grinding wheel laser dressing intelligent system through the input end; Step 2: The monitoring system compares and analyzes the outer dimensions of the grinding wheel blank and the contour requirement parameters of the drawing, and then automatically generates tool setting points and allowance information; Step 3: Compare the material information and dimensions of the grinding wheel blank with the dressing cases stored in the database to find similar or consistent dressing cases, and then optimize the processing parameters through expert prediction and parameter optimization functions; if there are no similar cases, perform machine learning through known cases to generate a set of processing parameters; the generated processing parameters are prepared for digital twins; Step 4: Send the processing parameters generated by the database to the CAM system. The CAM system calculates the laser beam posture adjustment method and laser energy compensation method according to the grinding wheel profile requirements and processing parameters, and generates the trajectory planning of the laser processing process; Step 5: The machine tool performs interpolation motion according to the trajectory planning generated by the CAM system and starts dressing the grinding wheel; Step 6: After the interpolation motion starts, the monitoring system monitors the machining process and observes the change of the blank's allowance in real time. Then, through template matching and target recognition, the real-time allowance and grinding wheel blank position information are sent to the CAM system and database; Step 7: After the margin meets the processing requirements, the monitoring system sends a message, stops the machine, and performs inspection; Step 8: Output the test results. If the product is qualified, the qualified product and test report are output; if it is unqualified, re-analyze the grinding wheel information and plan the next round of dressing; Step 9: If the special grinding wheel is difficult to pass after repeated dressing, use the system iteration function to optimize the system database and find more optimized parameters.

3. A method for intelligent laser dressing of grinding wheels as claimed in claim 2, characterized in that: In step 2, the monitoring system automatically generates tool setting points and allowance information according to the following method: For the judgment of grinding wheel blanks: judge the size of the blank machining allowance by visual observation and checking the allowance; For tool point detection: use the trial burning method, and then judge the position of the molten pool and the laser tool tip based on visual observation; when using the mechanical method, judge the tool depth and tool position through sound sensing and vision; Monitoring of the machining process: determine whether the energy absorption of the machining points is uniform based on the energy of the laser optical spectrometer; compare the actual contour observed by visual observation with the contour required by the drawing, and determine whether the machining is completed based on the difference between the two; determine the circularity, runout, and contour accuracy of the finished product based on visual observation and edge detection synthesis algorithm; determine the operating status of the entire equipment based on sound and current.

4. The method for intelligent laser dressing of a grinding wheel as claimed in claim 2, characterized in that: In step 3, the particle swarm PSO algorithm is used to optimize the machining process parameters. The algorithm includes the following steps: S301, using a random method to initialize the position and speed of the particle swarm: including the historical optimal position of a single particle and the historical optimal position of the swarm, the historical optimal fitness value of a single particle and the historical optimal fitness value of the swarm; S302, update the position and velocity of a single particle using formulas (1) and (2): The distance and direction of each particle's next step in the PSO algorithm is a position vector, called the speed of the next iteration, expressed as: in, represents the position vector of particle i in the dth dimension at the k+1th iteration; represents the position vector of particle i in the dth dimension in the kth iteration; represents the velocity vector of the dth dimension of particle i in the k+1th iteration; represents the velocity vector of particle i in the dth dimension in the kth iteration; ω represents the inertia weight; c1 and c2 represent learning factors; r1 and r2 represent two random numbers, ranging from 0 to 1; represents the historical optimal position of particle i in the d-th dimension in the k-th iteration, that is, after the k-th iteration, the optimal solution obtained by the i-th particle in the d-th dimension; It represents the historical optimal position of the swarm in the dth dimension at the kth iteration, that is, the optimal solution of the entire particle swarm in the dth dimension after the kth iteration; S303, calculate the fitness value of a single particle, and the fitness value must meet the requirements of formula (3): Among them, η m represents the fitness value of each particle; m represents the particle number; S304, updating the historical best fitting value of a single particle; S305, updating the historical best fitting value and position of the group; S306, update the inertia weight and the number of iterations: Using the linearly changing inertia weight strategy, the inertia weight is updated as follows (4): Among them, ω max and ω min They represent the maximum inertia weight and the minimum inertia weight respectively; iter represents the current number of iterations; iter max Indicates the maximum number of iterations; S307, when the terminal condition is met, output the final optimal solution, otherwise repeat steps S302 to S306.

5. The method for intelligent laser dressing of grinding wheels as claimed in claim 2, characterized in that: In step 5, the CAM system generates a trajectory planning process, including the following steps: S501, trimming parameters are determined: During tangential dressing, the laser spot is irradiated onto the grinding wheel surface in a pulsed manner, and the shape of the spot on the grinding wheel surface is approximately an ellipse. The spot overlap rate in the circumferential direction of the grinding wheel is defined as: the ratio of the area of ​​the overlapping part of the spot to the spot area under two pulses, expressed as Oc, and Oc is described as: Among them, S overlap Indicates the area of ​​the overlapping part of the light spot; S ellipse Indicates the spot area; a e and b e Indicates the parameters of the ellipse; l c Indicates the offset of the center of the light spot in the circumferential direction; The grinding wheel speed when the spot overlap rate reaches 85% is given by formula (6): Where D is the diameter of the grinding wheel; T represents the pulse period of the laser, which is the inverse of the repetition frequency; The spot overlap rate in the direction of the grinding wheel generatrix is ​​defined as the ratio of the spot overlap area to the spot area, expressed as A Indicates that O A Described as: Among them, l A is the offset of the center of the light spot on the busbar; Then, at the rotation speed n, the grinding wheel feed rate F is calculated by formula (8): F=nl A (8) The trimming parameters can be determined by formula (5) to formula (8); S502, defocus compensation: The edge of the grinding wheel is taken as the origin of the coordinates, the generatrix direction of the grinding wheel cylinder is taken as the X direction, and the diameter direction is taken as the Y direction and the Z direction; before the grinding wheel is dressed, its surface cylindrical annulus is described as: In the formula, x, y, and z are the three directions after the coordinate system is established, and the origin of the coordinate is at the edge of the grinding wheel; B is the thickness of the grinding wheel; During the material removal process, the vertically irradiated laser beam needs to rotate in the XOZ plane with the laser beam focus O as the rotation center. The state of rotation angle α is described as: Where r represents the radius of the laser beam at a certain position; h represents the distance from the position to the laser focus; r0 represents the radius of the spot at the laser focus; L ray represents the Rayleigh length of the laser beam, e0 is the initial defocus of the laser beam in the vertical state; For vertically irradiated laser, the focus is on the XOY plane. As the laser beam rotates, the intersection point P of the laser beam and the grinding wheel T In order to maintain a stable defocus, compensation needs to be made in the X and Z directions. The compensation amount is: When defocus compensation is performed, it is assumed that the focus always remains on the XOY plane; S503, Avoidance of laser interference: When the point of action between the laser beam and the material exceeds the Rayleigh length, the laser beam is considered to be a far-field laser composed of a hyperbola. In the far-field model, the radius value r at a certain height h in the direction of the beam is calculated using the asymptotic model. b It can be roughly described as: in, is the divergence angle of the laser beam; h b is the beam length; In the vertical direction, the beam length h b The following geometric relationship is satisfied between the height h: h b =h / cosα h (13) Among them, α h Indicates the angle of rotation of the A axis when the laser is half blocked; The shape profile of the wafer chamfering wheel is described by the following piecewise function: Among them, K s represents the intercept of the hypotenuse, (x r ,y r ) represents the coordinates of the center of the arc, which changes with the arc number; θ represents the angle between the contour line and the horizontal direction; y represents the coordinate system position; During the grinding wheel ablation process, the laser tip needs a certain cutting depth. Assuming that the cutting depth is fixed during processing, the actual position of the laser and the grinding wheel will be raised by the circular effect of the grinding wheel. The height Δh of the raised part is obtained by the following geometric relationship: in is the radius of the grinding wheel; a represents the cutting depth of the laser tool tip; At a height of Δh, the shape profile y of the forming grinding wheel h Transformed into: When the laser beam deflection angle continues to increase, the laser beam will intersect with the outer circle of the grinding wheel, causing the outer circle to block the laser light. At this time, the angle α h The calculation is done using the following formula: In the formula, α h Indicates the angle of rotation of the A axis when the laser is half blocked; L t represents the width of the groove, D represents the diameter, x l Indicates the distance between a certain position and the center of the groove; When the light blocking effect occurs, the corresponding critical angle α2 is described as: For a certain point on the grinding wheel forming profile, there is always a laser light blocking critical angle α2, so that when the laser beam deflection angle is greater than the critical angle α2, the laser is blocked by the outer cylindrical surface of the forming grinding wheel, and the dressing margin in the forming profile cannot be effectively removed; so the non-light blocking condition for laser dressing of the forming grinding wheel is: the laser deflection angle is less than the light blocking critical angle of the point at a certain point on the profile, that is: α(x)<α2(x) (19) In the actual processing technology, the deflection resolution M required for processing the arc is calculated using the following formula: Among them, K CW is a parameter, which takes an integer; L g Indicates the length of the secant line of the arc contour, L α is the arc ablation length corresponding to the critical angle of bevel ablation.

Images