A real-time evaluation and adjustment system for polishing quality of a reducer gear

Abstract

The application discloses a kind of real-time evaluation and adjustment system for grinding quality of speed reducer gear, and the application relates to the technical field of speed reducer gear grinding quality evaluation and adjustment, by industrial camera combining infrared light curtain surveying and mapping, first determine the geometric center of gear and split gear tooth image, and then analyze pixel deviation rate and angle deviation rate, generate grinding quality index;At the same time, the digital twin technology is used to construct the grinding machine model, and the optimal grinding parameter combination is identified by combining the merit index algorithm;The system adjusts other machine tools according to these parameters, ensures that the grinding quality of each machine tool is consistent and continuously optimized;The whole process realizes closed-loop control from image acquisition to quality evaluation, and then to parameter adjustment, effectively improves the grinding accuracy and efficiency of speed reducer gear.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of gear grinding quality evaluation and adjustment of speed reducer, and particularly relates to a grinding quality real-time evaluation and adjustment system for gears of speed reducer. BACKGROUND

[0002] With the transformation of global economy to high-end manufacturing, the demand for precision machining technology is increasing; as a core component in mechanical transmission system, the gear of speed reducer plays a vital role in key fields such as automobiles, aerospace, industrial automation, etc.

[0003] In the prior art, the evaluation and regulation of the grinding quality of the gear of speed reducer usually rely on discrete, post-event detection methods and isolated machine tool parameter management; first, in the image acquisition link, the traditional method often uses a fixed camera to shoot, lacks an active positioning and calibration mechanism, and cannot accurately determine the geometric center of the gear, resulting in image reference drift, which may introduce errors in subsequent analysis; at the same time, the two-dimensional image lacks the auxiliary mapping of three-dimensional profile information, and it is difficult to accurately segment the single gear region, especially for complex structure gears, the segmentation accuracy is limited; secondly, in the quality evaluation level, the existing technology relies on manual sampling inspection or simple image comparison, which is not only low in efficiency, but also subjective, and cannot real-time and quantitatively evaluate the key indicators such as tooth volume uniformity and graduation accuracy, the defect detection is lagging, and it is impossible to form a continuous quality index to support online decision-making; in addition, for the cooperative management of multiple grinding machines, the existing technology can only adjust parameters based on single machine historical data or experience, and cannot real-time correlate processing parameters, environmental parameters and final grinding quality across machines, resulting in isolated and lagging optimization and adjustment, and it is difficult to realize the quality consistency and efficiency optimization of the whole production line.

[0004] To solve the above problems, the application provides a grinding quality real-time evaluation and adjustment system for gears of speed reducer. SUMMARY

[0005] In view of the deficiencies of the prior art, the application provides a grinding quality real-time evaluation and adjustment system for gears of speed reducer, which solves the problems of quality evaluation lag and isolated equipment regulation in the grinding quality evaluation and adjustment of gears of speed reducer in the prior art.

[0006] The purpose of the application can be achieved by the following technical solutions:

[0007] A grinding quality real-time evaluation and adjustment system for gears of speed reducer, the system comprises:

[0008] The reducer gear image acquisition module acquires the reducer gear image through an industrial camera, and measures and maps the reducer gear by combining a pre-constructed infrared light curtain to determine the geometric center of the reducer gear and map the geometric center to the reducer gear image to obtain the reducer gear image.

[0009] The reducer gear polishing quality evaluation module segments the reducer gear image after mapping the geometric center, determines a plurality of tooth images associated with the reducer gear, and performs correlation analysis on all tooth images to evaluate the polishing quality of the corresponding reducer gear.

[0010] The polishing machine digital twin construction module acquires all polishing machines and extracts the machining parameter group and the environment parameter group associated with all polishing machines to construct the polishing machine digital twin model associated with all polishing machines.

[0011] The polishing quality of the reducer gear polished by any one polishing machine is bound to the polishing machine digital twin model associated with the corresponding polishing machine to construct the polishing processing feature.

[0012] The polishing machine machining parameter adjustment module performs correlation analysis based on the determined plurality of polishing processing features, evaluates the polishing machine digital twin model associated with the optimal polishing quality, and acquires the machining parameter group and the environment parameter group associated with the polishing machine digital twin model to combine as the optimal adjustment parameter set. The polishing machine digital twin model associated with the remaining polishing machines is adjusted, and the adjustment result is output by each polishing machine digital twin model. The remaining polishing machines are adjusted based on the adjustment result.

[0013] As a further scheme of the present application, in the reducer gear image acquisition module, the specific way to acquire the reducer gear image is:

[0014] Take any reducer gear A;

[0015] Place A on the industrial shooting table and use an industrial camera to take a picture to obtain the reducer gear image G_A of A;

[0016] The industrial shooting table includes a built-in three-axis MEMS gravity sensor for bearing the reducer gear, and the object carrier has a horizontal calculation function to determine the inclination of the object carrier in the horizontal plane and output the normal vector n of the object carrier;

[0017] When the object carrier detects A, a detection start signal is sent to the infrared light curtain;

[0018] The infrared light curtain is a four-beam infrared fan-shaped light curtain, which simultaneously scans the tooth profile outer edge of A to obtain four polar coordinate sets {P1, P2, P3, P4} of the addendum circle of A in the light curtain coordinate system.

[0019] determine the geometric center Q1 of the addendum circle based on {P1, P2, P3, P4}, and send an adjustment signal to the industrial camera;

[0020] The industrial camera adjusts the optical axis direction to coincide with the normal vector n, vertically ascends along the Z axis, and performs a shooting operation when the pixel distance between the projection point of the geometric center Q1 on the imaging target surface of the industrial camera and the physical center of the target surface is less than a preset pixel distance threshold.

[0021] As a further scheme of the application, in the reducer gear polishing quality evaluation module, the specific way of determining the several tooth images associated with the reducer gear is:

[0022] Project the three-dimensional point cloud data of the reducer gear A measured by the infrared light curtain onto the imaging plane of the industrial camera using a preset spatial mapping matrix to determine the reducer gear pixel image G_A` of A in G_A;

[0023] Determine the geometric center Q1 in G_A`, and expand the circle U outward with Q1 as the center and the radius increasing by one pixel point as the step until U reaches the maximum state in G_A`, wherein the pixel points in U all belong to G_A`;

[0024] Crop U from G_A` and remove it to obtain the remaining pixel region;

[0025] The remaining pixel region contains several closed regions, and the total number of closed regions is counted and denoted as j;

[0026] The j closed regions are the tooth images of the j teeth on A;

[0027] Denote the tooth image directly above the remaining pixel region as TG1, arrange the j tooth images in clockwise order to obtain the tooth image sequence TG1, TG2,..., TGj.

[0028] As a further scheme of the application, the specific way of evaluating the polishing quality of the corresponding reducer gear in the reducer gear polishing quality evaluation module includes:

[0029] Take the tooth image sequence TG1, TG2,..., TGj;

[0030] Count the total number of pixel points in each tooth image in TG1, TG2,..., TGj, and denote the total number of pixel points in the order of TG1, TG2,..., TGj as the total number of pixel points sequence SUM1, SUM2,..., SUMj;

[0031] Take the average of the j total numbers of pixel points in SUM1, SUM2,..., SUMj to obtain the average total number of pixel points SUM_avg;

[0032] Take the total number of pixels SUMi of any tooth image TGi, calculate the absolute value of the difference between SUMi and SUM_avg, divide the absolute value of the difference by SUM_avg, and multiply by 100% to obtain the pixel deviation rate δ of SUMi, where i is the counting index, 1≤i≤j;

[0033] Compare δ with the preset pixel deviation rate threshold δ_yu;

[0034] If δ≥δ_yu, it is determined that there is a defect in the inter-tooth volume difference of gear A of the reducer. The first type of defect flag F1 is incremented by one, where the initial value of F1 is 0.

[0035] Conversely, no action is taken;

[0036] Similarly, the j gear tooth images in TG1, TG2, ..., TGj are processed to update the first type of defect label F1.

[0037] As a further aspect of the present invention, the specific method for evaluating the grinding quality of the corresponding reducer gear in the reducer gear grinding quality evaluation module further includes:

[0038] Take the gear tooth image sequence TG1,TG2,...,TGj;

[0039] Determine the geometric center of each of the j gear tooth images to obtain the geometry of the j gear teeth;

[0040] Map the geometric centers of j gear teeth to the pixel image G_A` of the reducer gear and label them;

[0041] Obtain and label the geometric center Q1 of G_A`;

[0042] Connect j teeth with their geometric centers as the starting point and Q1 as the ending point to obtain j lines;

[0043] j lines form j angles with the geometric center Q1;

[0044] Divide the full angle by j to obtain the reference angle value;

[0045] Obtain the angle value of any one of the j angles, calculate the absolute value of the difference between this angle value and the reference angle value, divide it by the reference angle value, and multiply it by 100% to obtain the angle value difference rate between this angle value and the reference angle value;

[0046] If the angle difference rate is greater than or equal to the preset angle difference rate threshold, it is determined that gear A of the reducer has an uneven indexing defect, and the second type of defect flag F2 is incremented by one, where the initial value of F2 is 0;

[0047] Conversely, no action is taken;

[0048] Similarly, the angle values of the j angles are processed, and F2 is updated;

[0049] Taking F1 and F2, the polishing quality index E of A is calculated by E = w1*F1 + w2*F2, wherein w1 and w2 are preset calculation weights, w1 and w2 are greater than 0, and w1 + w2 = 1;

[0050] If E < the preset polishing quality index threshold E_yu, the gear A of the speed reducer is qualified, otherwise, the gear A is unqualified.

[0051] As a further scheme of the present application, in the polishing machine tool digital twin construction module, the specific way of constructing the polishing machine tool digital twin model associated with all polishing machine tools is:

[0052] All polishing machine tools are obtained, and the total number is recorded as u;

[0053] The u polishing machine tools are recorded in the order of acquisition as a polishing machine tool sequence MT1, MT2,..., MTu;

[0054] Obtain the environmental parameters of each polishing machine tool processing the gear of the speed reducer, and combine them into an environmental parameter group;

[0055] Obtain the processing parameters predefined by the operator for each polishing machine tool, and combine them into a processing parameter group;

[0056] Based on the digital twin technology, construct the polishing machine tool digital twin model associated with each of the u polishing machine tools, and perform parameter injection on the polishing machine tool digital twin model associated with each polishing machine tool based on the processing parameters and environmental parameters of each polishing machine tool.

[0057] Get the polishing machine tool digital twin model sequence PMT1, PMT2,..., PMTu.

[0058] As a further scheme of the present application, in the polishing machine tool digital twin construction module, the specific way of constructing the polishing processing feature is:

[0059] Obtain any one of the polishing machine tool MTo in the polishing machine tool sequence MT1, MT2,..., MTu and the corresponding polishing machine tool digital twin model PMTo, wherein o is a count index, 1 ≤ o ≤ u;

[0060] Take the total number of polishing H preset by the operator;

[0061] Determine the qualified total number Cout1, the unqualified total number Cout2 of the H gears of the speed reducer polished by the polishing machine tool MTo closest to the current time, and the average of the polishing quality indexes E associated with the H gears of the speed reducer, recorded as the average polishing quality index E_avg;

[0062] binding E_avg, Cout1, Cout2 and PMTo as polishing processing features PPFo;

[0063] Similarly, the synchronous processing is performed on each polishing machine in the polishing machine sequence MT1, MT2,..., MTu, to determine u polishing processing features, and the polishing processing features are recorded in sequence as the polishing processing feature sequence PPF1, PPF2,..., PPFu.

[0064] As a further scheme of the present application, in the polishing machine processing parameter adjustment module, the specific way of evaluating the polishing machine digital twin model associated with the optimal polishing quality is:

[0065] Taking the polishing processing feature sequence PPF1, PPF2,..., PPFu;

[0066] Extracting the average polishing quality index E_avg, the total number of qualified products Cout1, and the total number of unqualified products Cout2 in any one polishing processing feature PPFo;

[0067] Calculating the polishing machine merit index YL of PPFo using YL=(Cout1–Cout2) / (Cout1+Cout2)-λ*E_avg, wherein λ is a dimensionless scaling constant;

[0068] Similarly, the polishing machine merit index of each polishing processing feature in PPF1, PPF2,..., PPFu is determined;

[0069] Arranging PPF1, PPF2,..., PPFu in descending order according to the polishing machine merit index, to obtain the polishing processing feature sequence PPF1`, PPF2`,..., PPFu`;

[0070] Taking the first polishing processing feature PPF1` and the polishing machine digital twin model PMT1` associated therewith in PPF1`, PPF2`,..., PPFu`;

[0071] The polishing machine digital twin model PMT1` is the polishing machine digital twin model associated with the optimal polishing quality.

[0072] As a further scheme of the present application, in the polishing machine processing parameter adjustment module, the specific way of determining the dimensionless scaling constant λ is:

[0073] The operator pre-acquires k full-qualified zero-defect reduction machine gears and k full-unqualified high-defect reduction machine gears as limit samples, wherein k is a preset value;

[0074] The full-qualified zero-defect corresponds to Cout2=0 and E_avg=0, and λ is set to 1.

[0075] Full disqualification - high defect corresponds to Cout1=0 and E_avg=E_max, let lambda be -1;

[0076] In summary, solve lambda, lambda belongs to [-1,1];

[0077] Lambda is a fixed value and no longer changes.

[0078] As a further scheme of the application, in the polishing machine tool processing parameter adjustment module, the specific way of adjusting the remaining polishing machine tool based on the adjustment result is:

[0079] Obtain the machining parameter group and the environment parameter group of the polishing machine tool digital twin PMT1` with the optimal polishing quality, and combine them as the optimal adjustment parameter set;

[0080] Take u-1 polishing machine tool digital twins other than PMT1` among the u polishing machine tool digital twins;

[0081] Adjust the machining parameter group and the environment parameter group of the u-1 polishing machine tool digital twins to the machining parameter group and the environment parameter group in the optimal adjustment parameter set;

[0082] Continuously monitor the adjustment results output by the u-1 polishing machine tool digital twins after polishing H reducer gears, and the adjustment result is the average polishing quality index;

[0083] If the average polishing quality index output by any polishing machine tool digital twin is better than the current average polishing quality index of the corresponding polishing machine tool, the corresponding polishing machine tool is controlled according to the optimal adjustment parameter set;

[0084] On the contrary, do not process.

[0085] The beneficial effects of the application are:

[0086] (1) The application realizes high-precision positioning and self-adaptive shooting control of the reducer gear by integrating the object carrier disc with the built-in three-axis MEMS gravity sensor and the four-beam infrared fan-shaped light curtain; the inclination of the object carrier disc is monitored in real time, and the optical axis direction of the industrial camera is automatically calibrated to ensure that image acquisition is always perpendicular to the gear plane, effectively avoiding image distortion and measurement errors caused by tilting placement; at the same time, the infrared light curtain non-contact scans the tooth profile and quickly calculates the geometric center of the addendum circle to guide the camera to accurately center, thereby improving image consistency and detection reliability, and providing a high-quality image basis with complete structure and standard position for subsequent gear quality analysis;

[0087] (2) The present application accurately locates the independent image area of each tooth by using the tooth profile data obtained by the infrared light curtain, and effectively identifies the tooth shape volume difference defects caused by uneven polishing by calculating the total pixel deviation rate of the teeth; at the same time, by analyzing the uniformity of the included angle between the geometric center of each tooth and the overall center of the gear, the division accuracy defects can be reliably detected; this double-index fusion evaluation strategy (tooth shape consistency + division uniformity) significantly improves the comprehensiveness and accuracy of quality judgment, avoids human subjective errors, and is based on image and geometric data automatic processing, improves the detection efficiency and objectivity;

[0088] (3) The present application realizes the accurate digital mapping of the production process by constructing a digital twin model linked in real time with the physical polishing machine tool and injecting processing and environmental parameters; then, by collecting and analyzing the quality data of multiple gears polished by the machine tool in the near future, the polishing processing features reflecting the comprehensive processing state are automatically generated; this feature binding method based on historical performance data can objectively and quantitatively evaluate the stability and processing capacity of each machine tool;

[0089] (4) The present application automatically identifies the polishing machine tool with the best machining quality in the current production system and its digital twin model by defining a machine tool performance automatic sorting mechanism based on the polishing machine tool merit index; innovatively uses the comprehensive calculation of the pass rate and the average quality index to objectively evaluate the machine tool merit and demerit, and pre-calibrates the key parameter λ based on the limit sample to ensure the rationality and stability of the evaluation algorithm; then, by synchronizing the processing and environmental parameters of the optimal machine tool to the digital twin models of other machine tools and tracking and dynamically adjusting the effect, the cluster optimization and self-iteration of production parameters are realized, thereby improving the polishing quality consistency, equipment comprehensive efficiency and intelligent management level of the whole production line. BRIEF DESCRIPTION OF DRAWINGS

[0090] The present application will be further described below in conjunction with the drawings.

[0091] Figure 1 is a structural schematic diagram of the system described in the present application;

[0092] Figure 2 is a flowchart of the method described in embodiment 2 of the present application;

[0093] Figure 3 is a flowchart of the method described in embodiment 3 of the present application;

[0094] Figure 4 is a flowchart of the method described in embodiment 4 of the present application. DETAILED DESCRIPTION

[0095] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0096] Embodiment 1

[0097] A polishing quality real-time evaluation and adjustment system for a reducer gear, as shown in the accompanying drawings, the system comprises the following: Figure 1

[0098] The system is mainly implemented by combining four modules, i.e., a reducer gear image acquisition module, a reducer gear polishing quality evaluation module, a polishing machine digital twin construction module, and a polishing machine processing parameter adjustment module.

[0099] The reducer gear image acquisition module shoots the reducer gear after polishing operation in real time through an industrial camera, combines a pre-constructed infrared light curtain to measure and map the reducer gear, determines the geometric center of the reducer gear, maps the geometric center to the reducer gear image, and acquires the reducer gear image.

[0100] Specifically, first, any reducer gear after polishing operation is acquired, denoted as A. It needs to be noted that the reducer gear needs to be screened for faults before polishing operation, so as to ensure that only non-fault reducer gears are subjected to polishing operation and subsequent polishing quality real-time evaluation and adjustment steps. If any reducer gear has a fault, it is directly discarded without entering the polishing operation and subsequent process.

[0101] Next, the reducer gear A is placed on a pre-constructed industrial shooting table, and the industrial camera is used to shoot the reducer gear A, so that the reducer gear image associated with the reducer gear A, denoted as G_A, can be obtained.

[0102] The industrial shooting table mentioned above is composed of a built-in three-axis MEMS gravity sensor object carrier, an infrared light curtain, and an industrial camera.

[0103] The built-in three-axis MEMS gravity sensor object carrier can also determine whether the reducer gear is placed on the object carrier, and the object carrier also has a horizontal calculation function, which can output the inclination of the object carrier itself in the horizontal plane in real time, and determine the normal vector n associated with the current position of the object carrier.

[0104] If the object carrier detects that the reducer gear A is placed therein, a detection start signal is generated and transmitted to the infrared light curtain.​

[0105] The infrared light curtain is four infrared fan-shaped light curtains, and after the infrared light curtain receives the detection start signal, the four infrared fan-shaped light curtains will be controlled to simultaneously scan the tooth profile outer edge of the gear A of the speed reducer;

[0106] Based on the scanning results of the four infrared fan-shaped light curtains, four polar coordinate sets of the addendum circle of the gear A of the speed reducer in the light curtain coordinate system (the light curtain coordinate system formed after the four infrared fan-shaped light curtains are scanned) can be obtained, which are sequentially recorded as {P1, P2, P3, P4}.

[0107] Based on the four polar coordinate sets {P1, P2, P3, P4} determined, a closed circle is fitted, the center of the closed circle is determined based on the fitted closed circle, and the center is taken as the geometric center of the addendum circle associated with the gear A of the speed reducer, and is recorded as Q1.

[0108] At this point, the infrared light curtain determines the geometric center Q1 of the gear A of the speed reducer, generates an adjustment signal from the infrared light curtain, and transmits it to the industrial camera;

[0109] After the industrial camera receives the adjustment signal, the industrial camera will first be adjusted to coincide with the normal vector n determined by the above-mentioned turntable, and the imaging direction of the industrial camera will be adjusted to coincide with the normal vector n determined by the above-mentioned turntable, to ensure that the turntable (i.e. the gear of the speed reducer) is shot directly.

[0110] Then, after the imaging direction of the industrial camera is adjusted to coincide with the normal vector n determined by the above-mentioned turntable, the industrial camera is then operated vertically along the Z-axis until the pixel distance between the projection point of the geometric center Q1 on the imaging target surface of the industrial camera and the physical center of the target surface is less than the pixel distance threshold preset by the operator (this step is a focusing process for the industrial camera to achieve clear shooting), and then the shooting operation is performed to obtain the gear image of the speed reducer.

[0111] The gear grinding quality evaluation module segments the gear image after mapping the geometric center, determines a plurality of tooth images associated with the gear of the speed reducer, and performs correlation analysis on all tooth images to evaluate the grinding quality of the corresponding gear of the speed reducer.

[0112] Specifically, this module is a complete step from "one whole image" to "full-tooth defect quantification" of the gear image of the speed reducer, and the purpose of segmenting the gear image of the speed reducer to obtain all tooth images of the gear of the speed reducer is to analyze each tooth on the gear of the speed reducer. Based on the analysis results of the teeth on the gear of the speed reducer, the grinding quality of the gear of the speed reducer is comprehensively evaluated.

[0113] The polishing machine digital twin construction module acquires all polishing machines and extracts all polishing machine associated machining parameter groups and environment parameter groups to construct all polishing machine associated polishing machine digital twin models.

[0114] The polishing quality of any one polishing machine polished gear is bound to the polishing machine digital twin model associated with the corresponding polishing machine to construct the polishing processing feature.

[0115] Specifically, the method of constructing all polishing machine associated polishing machine digital twin models by extracting machining parameter groups and environment parameter groups from all polishing machines can avoid the trial and error cost brought by polishing machine parameter adjustment.

[0116] It should be noted that the machining parameter group includes machine tool motion parameters such as grinding wheel spindle speed and workpiece spindle speed, and grinding process parameters such as single grinding depth and grinding method.

[0117] The environment parameter group includes temperature, humidity, air cleanliness, and wind speed.

[0118] Next, the polishing quality of any one polishing machine polished gear is bound to the polishing machine digital twin model associated with the polishing machine to construct the polishing processing feature. The polishing machine digital twin model is bound to the actual polishing quality using the solidified polishing processing feature. This operation actually also binds the machining parameter group and the environment parameter group of the physical polishing machine to the polishing quality.

[0119] The polishing machine processing parameter adjustment module first performs correlation analysis according to the above determined polishing processing features, evaluates the optimal polishing quality and the polishing machine digital twin model associated with the optimal polishing quality through correlation analysis.

[0120] Next, the machining parameter group and the environment parameter group associated with the polishing machine digital twin model associated with the optimal polishing quality are obtained.

[0121] The obtained machining parameter group and environment parameter group are combined, and the combined result is recorded as the optimal adjustment parameter set. The optimal adjustment parameter set is used to adjust the polishing machine digital twin models associated with the remaining polishing machines (i.e., trial and error using the polishing machine digital twin model). Each polishing machine digital twin model outputs an adjustment result, and the remaining polishing machines (the machining parameter group and the environment parameter group of the actual polishing machine) are adjusted based on the adjustment result.

[0122] The embodiment binds the "quality-working condition" of a single gear to the virtual model of the machine tool through industrial vision, infrared mapping and digital twin cooperation, forms a reusable polishing processing feature, and then uses the feature to quickly test and correct in the twin body, reversely outputs the optimal parameter set, calibrates the remaining machine tools in batches, realizes "one-time yield replication", and finally stabilizes the polishing quality of the reducer gear in the optimal interval, and reduces the scrap rate and parameter adjustment cost.

[0123] Embodiment 2

[0124] The embodiment further discloses a method for evaluating the polishing quality of a reducer gear based on the embodiment 1, as shown in the formula (1), and specifically comprises the following steps. Figure 2

[0125] Based on the content described in embodiment 1, it can be known that the infrared curtain performs a mapping operation on the reducer gear A, so that the three-dimensional point cloud data associated with the reducer gear A can be obtained. The determined three-dimensional point cloud data is projected to the imaging plane of the industrial camera (that is, the reducer gear image G_A shot by the industrial camera) through the space mapping matrix (actually, it is a transformation matrix of three-dimensional and two-dimensional, which converts the 3D point in the world coordinate system into the 2D pixel coordinate on the imaging plane of the camera, which is a prior art and will not be described in detail in this scheme) preset by the operator, and the two-dimensional pixel region in the reducer gear image G_A is further determined, and the reducer gear pixel image is obtained, which is denoted as G_A`. The purpose of this step is to separate the part of the reducer gear A pixel from the reducer gear image G_A and remove the background pixels.

[0126] Then, the geometric center Q1 in the reducer gear pixel image G_A` is determined, that is, the geometric center Q1 in the reducer gear image G_A. The determined geometric center Q1 is taken as the center of the outward expanding circle U.

[0127] It should be noted that the radius of the outward expanding circle U increases by one pixel point as a step, and all the pixel points covered by the outward expanding circle U are checked in real time whether they are contained in the reducer gear pixel image G_A`. When the radius is increased by one pixel, the outward expanding circle U touches the pixels outside the G_A` region. At this time, the outward expanding circle U is the approximate maximum inscribed circle of the reducer gear A. At this time, the outward expanding circle U maximally covers the central body part (spokes / hub) of the gear, and does not contain any gear tooth.

[0128] Then, the outward expanding circle U is cropped from the reducer gear pixel image G_A` and removed, and only the remaining pixel region of the gear tooth is retained.

[0129] ​The pixel area of each tooth is a closed area, and the total number of all closed areas in the remaining pixel area is recorded as j, which is also the total number of all teeth on the gear A of the speed reducer. Each closed area is a tooth image, so finally j tooth images can be obtained.

[0130] Then, the first tooth image directly above the remaining pixel area is recorded as TG1, and the subsequent tooth images are recorded and sorted in clockwise order to obtain a tooth image sequence, which is represented as TG1, TG2,..., TGj.

[0131] Next, the total number of pixels in each tooth image in the tooth image sequence TG1, TG2,..., TGj is calculated, so that the total number of j pixels associated with the j tooth images can be obtained. The total number of j pixels is recorded as a pixel total sequence in the order of the tooth image sequence TG1, TG2,..., TGj, and is represented as SUM1, SUM2,..., SUMj.

[0132] Next, the j pixel totals are extracted, and the average is calculated, and the obtained average is recorded as the pixel total average, which is represented as SUM_avg.

[0133] Next, any tooth image TGi is determined from the tooth image sequence TG1, TG2,..., TGj for example processing, and the remaining tooth images are processed in the same way as the tooth image TGi, where i is a count index, and the value range is 1 to j;

[0134] The pixel deviation rate δ between the pixel total SUMi and the pixel total average SUM_avg is calculated by using |SUMi-SUM_avg| / SUM_avg*100%=δ;

[0135] Then, the calculated pixel deviation rate δ is compared with the pixel deviation rate threshold δ_yu preset by the operator:

[0136] If there is a pixel deviation rate δ greater than or equal to the pixel deviation rate threshold δ_yu, it is determined that the gear A of the speed reducer has an inter-tooth volume difference defect, and the first defect flag F1 is incremented by one. The first defect refers to the inter-tooth volume difference defect, and the initial value of the first defect flag F1 is 0.

[0137] If there is a pixel deviation rate δ less than the pixel deviation rate threshold δ_yu, no processing is performed, and the other tooth images are traversed. All tooth images in the tooth image sequence TG1, TG2,..., TGj are processed in the same way, and the first defect flag F1 is updated after processing is completed.

[0138] Then, j gear tooth geometric centers in the gear tooth image sequence TG1, TG2,..., TGj are determined, and the j gear tooth geometric centers are recorded in the order of the gear tooth image sequence TG1, TG2,..., TGj as a gear tooth geometric center sequence, denoted as GC1, GC2,..., GCj.

[0139] Then, the j gear tooth geometric centers in the gear tooth geometric center sequence GC1, GC2,..., GCj are mapped to the reducer gear pixel image G_A` and labeled.

[0140] The geometric center Q1 of the reducer gear pixel image G_A` is determined and labeled.

[0141] At this time, j+1 labeled points, i.e., j gear tooth geometric centers and 1 geometric center Q1, can be obtained in the reducer gear pixel image G_A`.

[0142] The j gear tooth geometric centers in the gear tooth geometric center sequence GC1, GC2,..., GCj are connected as starting points and the geometric center Q1 is connected as an ending point in the direction from the starting point to the ending point, and finally j lines are obtained.

[0143] At this time, the j lines and the geometric center Q1 will form j angles.

[0144] The reference angle value is obtained by dividing the circumferential angle (360 degrees) by j (if each gear tooth is standard, then the angle values of the j angles obtained will be close to the reference angle value, otherwise it indicates that the gear tooth is not standard).

[0145] Then, the angle value of any one of the j angles is obtained, and the angle value difference rate calculation is performed between the angle value and the reference angle value, and the calculation steps are as follows:

[0146] First, the absolute value of the difference between the angle value and the reference angle value is calculated, and then the absolute value of the difference is divided by the reference angle value and multiplied by 100%, and finally the obtained value is the angle value difference rate between the angle value of the angle and the reference angle value.

[0147] The angle value difference rate threshold value set by the operator in combination with the actual situation is obtained, and the calculated angle value difference rate is compared with the angle value difference rate threshold value set by the operator.

[0148] If the angle value difference rate is greater than or equal to the angle value difference rate threshold value, it indicates that the reducer gear A has a division uneven defect, and the second type of defect flag F2 is incremented by one, wherein the second type of defect refers to the division uneven defect, and the initial value of the second type of defect flag F2 is 0.

[0149] If the angle value difference rate is less than the angle value difference rate threshold, no processing is performed, and the angle value difference rates of other angles are calculated until the angle values of all angles are processed, and the second type of defect flag F2 is updated.

[0150] At this point, the first type of defect flag F1 and the second type of defect flag F2 (it should be noted that the first type of defect flag F1 and the second type of defect flag F2 have no dimension concept, and are only numerical values) are obtained.

[0151] The grinding quality index E associated with the reducer gear A is calculated using E = w1*F1 + w2*F2, wherein the calculation weights w1 and w2 are obtained by being preset by an operator, and the calculation weights w1 and w2 satisfy w1, w2 > 0 and w1 + w2 = 1.

[0152] Next, the grinding quality index E is determined. If the grinding quality index E is less than the grinding quality index threshold E_yu preset by the operator, the reducer gear A is determined to be qualified (the smaller the grinding quality index E, the smaller the defect, and vice versa).

[0153] If the grinding quality index E is greater than or equal to the grinding quality index threshold E_yu preset by the operator, the reducer gear A is determined to be unqualified.

[0154] It should be noted that although the embodiment is to determine the qualification of the reducer gear by counting defects, it is clear from Embodiment 1 that the qualified reducer gear with obvious faults will be invalidated, so this embodiment mainly processes the reducer gear without obvious faults, or the reducer gear that cannot be identified by artificial means. Therefore, there is no case where a reducer gear is invalidated due to a significant fault in this step.

[0155] In this embodiment, three-dimensional point cloud data of the reducer gear is collected by an infrared curtain and an industrial camera, and is projected onto a two-dimensional image plane to separate the gear pixel area to remove background interference. Based on the gear pixel image, the number of teeth is accurately identified and counted by geometric center determination, outer expansion circle operation and tooth pixel area clipping, and the total number of pixel points of each tooth and the deviation rate from the average value are calculated to determine the inter-tooth volume difference defect. In addition, the angle between the tooth geometric center and the gear geometric center is analyzed to evaluate the division uniformity and determine the division unevenness defect. Finally, the grinding quality index is calculated using a weighted formula based on the two types of defect flags to automatically quantify the grinding quality of the reducer gear and quickly determine whether the gear is qualified.

[0156] Embodiment 3

[0157] The embodiment continues to disclose a method for building a digital twin model of a grinding machine and grinding processing features based on embodiment 2, as shown, which specifically includes the following: Figure 3

[0158] First, all grinding machines are obtained, and the total number is counted as u. It should be noted here that the grinding machines are grinding machines for processing the same reducer gear, and need to meet the same model, otherwise they are not comparable.

[0159] The obtained u grinding machines are recorded in the order of acquisition as a grinding machine sequence, denoted as MT1, MT2,..., MTu.

[0160] According to the content described in embodiment 1, the environmental parameters associated with each of the u grinding machines in the grinding machine sequence MT1, MT2,..., MTu when processing the reducer gear are extracted in real time, and are combined as an environmental parameter group.

[0161] Similarly, the processing parameters associated with each of the u grinding machines in the grinding machine sequence MT1, MT2,..., MTu when processing the reducer gear are extracted in real time, and are combined as a processing parameter group.

[0162] Based on the digital twin technology, the grinding machine digital twin model associated with each of the u grinding machines is built (at this time, the grinding machine digital twin model built does not have the ability to simulate), and the processing parameter group and the environmental parameter group obtained in the above steps are extracted. The parameter group injection is performed on the grinding machine digital twin model associated with each grinding machine, that is, the grinding machine digital twin model is matched with the actual grinding machine.

[0163] Finally, the injected u grinding machine digital twin models are obtained, and the u grinding machine digital twin models are sorted in the order of the grinding machine sequence MT1, MT2,..., MTu. The sorted result is recorded as a grinding machine digital twin model sequence, denoted as PMT1, PMT2,..., PMTu.

[0164] Then, any grinding machine MTo is extracted from the grinding machine sequence MT1, MT2,..., MTu, and the grinding machine digital twin model PMTo associated with the grinding machine MTo is obtained, where o is a count index, and the value range is 1 to u.

[0165] The total number of grinding H preset by the operator is obtained;

[0166] ​Then, taking the current time as the end time, the grinding quality of the H ground reducer gears closest to the current time in the past is obtained from the grinding machine MTo, and the total number of qualified Cout1, the total number of unqualified Cout2 and the grinding quality index E associated with each of the H reducer gears are determined.

[0167] And calculate the average of the H grinding quality indexes E, denoted as the average grinding quality index E_avg.

[0168] Then, the determined average grinding quality index E_avg, the total number of qualified Cout1, the total number of unqualified Cout2 and the grinding machine digital twin model PMTo are combined and bound, and the result after combination and binding is taken as a grinding processing feature, denoted as PPFo.

[0169] In this way, the grinding processing features associated with each of the u grinding machines in the grinding machine sequence MT1, MT2,..., MTu are determined, and a total of u grinding processing features are obtained. The u grinding processing features are denoted as a grinding processing feature sequence according to the order of the grinding machine sequence MT1, MT2,..., MTu, denoted as: PPF1, PPF2,..., PPFu.

[0170] Embodiment 4

[0171] This embodiment continues to disclose a method for adjusting the grinding machine based on embodiment 3, as shown in Figure 4 Specifically, the method comprises the following steps:

[0172] Based on the content described in embodiment 3, the grinding processing feature sequence PPF1, PPF2,..., PPFu associated with the grinding machine sequence MT1, MT2,..., MTu and the grinding machine digital twin model sequence PMT1, PMT2,..., PMTu can be obtained;

[0173] From the grinding processing feature sequence PPF1, PPF2,..., PPFu, any grinding processing feature PPFo is extracted, and the average grinding quality index E_avg, the total number of qualified Cout1 and the total number of unqualified Cout2 in the grinding processing feature PPFo are further determined.

[0174] According to the formula:

[0175] YL=(Cout1–Cout2) / (Cout1+Cout2)-λ*E_avg

[0176] The grinding machine merit index YL associated with the grinding processing feature PPFo is calculated, wherein λ is a dimensionless scaling constant.

[0177] The determination method of the dimensionless scaling constant λ comprises the following steps:

[0178] Firstly, k full qualified-zero defect reducer gears and k full unqualified-high defect reducer gears need to be pre-acquired by the operator as limit samples, wherein the value of k is determined by the operator in combination with the actual situation;

[0179] Based on the determined limit samples, it is known that when full qualified-zero defect, Cout2=0 and E_avg=0, at this time, let λ be 1;

[0180] When full unqualified-high defect, Cout1=0 and E_avg=E_max, at this time, let λ be-1;

[0181] In summary, λ can be solved, and λ∈[-1,1], the dimensionless scaling constant λ is a fixed value and no longer changes, and λ*E_avg will eliminate the dimension of E_avg itself to obtain a numerical part;

[0182] Here, it needs to be explained that the purpose of determining the dimensionless scaling constant λ is to balance the qualified proportion and polishing quality index E_avg, which are two different dimensional data, eliminate the dimensional difference between them, ensure that the weights in the formula are reasonable, and accurately reflect the advantages and disadvantages of the machine tool. By limiting the dimensionless scaling constant λ in the interval [-1,1], it can adapt to the two extreme cases of full qualified and full unqualified, ensure that the formula can reasonably output under various quality scenarios, and avoid logical contradictions or misjudgments.

[0183] According to the method for determining the polishing machine tool advantage and disadvantage index YL associated with the polishing processing feature PPFo, the remaining u-1 polishing processing features in the polishing processing feature sequence PPF1, PPF2,..., PPFu are processed in the same way, and finally the polishing machine tool advantage and disadvantage indexes associated with the u polishing processing features are obtained;

[0184] Then, the polishing processing feature sequence PPF1, PPF2,..., PPFu is reordered according to the polishing machine tool advantage and disadvantage indexes corresponding to each polishing processing feature from large to small, and the reordered polishing processing feature sequence is represented as: PPF1`, PPF2`,..., PPFu`.

[0185] The first polishing processing feature PPF1` and its associated polishing machine tool digital twin model PMT1` and polishing machine tool MT1` in the polishing processing feature sequence PPF1`, PPF2`,..., PPFu` are extracted;

[0186] At this time, the extracted polishing machine tool digital twin model PMT1' is the polishing machine tool digital twin model associated with the optimal polishing quality, and the polishing machine tool MT1' is the polishing machine tool associated with the optimal polishing quality.

[0187] Then, the polishing machine tool digital twin model PMT1' associated with the optimal polishing quality and its associated processing parameter set and environmental parameter set are obtained, and the obtained processing parameter set and environmental parameter set are combined again, and the result of the combination is recorded as the optimal adjustment parameter set.

[0188] The u-1 polishing machine tool digital twin models other than the polishing machine tool digital twin model PMT1' among the u polishing machine tool digital twin models, i.e. the polishing machine tool digital twin models with the second best polishing quality, are obtained.

[0189] The processing parameter set and the environmental parameter set of the u-1 polishing machine tool digital twin models are adjusted to the processing parameter set and the environmental parameter set in the optimal adjustment parameter set, and the adjustment results (here, the adjustment results are the polishing quality indexes of the H reduction gear teeth) output by the u-1 polishing machine tool digital twin models after polishing H reduction gear teeth are continuously monitored.

[0190] If the average polishing quality index (i.e. the adjustment result) output by any one of the polishing machine tool digital twin models is better than the current average polishing quality index of the corresponding polishing machine tool, the current processing parameter set and the environmental parameter set of the corresponding polishing machine tool are immediately adjusted to the processing parameter set and the environmental parameter set in the optimal adjustment parameter set.

[0191] If the average polishing quality index (i.e. the adjustment result) output by any one of the polishing machine tool digital twin models is not better than the current average polishing quality index of the corresponding polishing machine tool, the current processing parameter set and the environmental parameter set are kept unchanged.

[0192] In addition to the above steps, the operator can also select cascade adjustment, obtain the optimal adjustment parameter set associated with the polishing machine tool digital twin model associated with the second polishing processing feature PPF2' in the polishing processing feature sequence PPF1', PPF2',..., PPFu', and adjust the polishing machine tool digital twin model associated with the polishing machine tool associated with PPF1' and PPF2' other than PPF1' and PPF2', and repeat this step until the last polishing processing feature PPFu' associated with the polishing machine tool, so as to realize cascade adjustment.

[0193] The embodiment optimizes the processing parameters of the polishing machine tool to improve the polishing quality through digital modeling and quality evaluation. First, based on the merit index calculation method of the polishing machine tool, the number of qualified products and the quality index are combined to determine the merit ranking of each machine tool. The dimensionless scaling constant lambda is calibrated through the limit sample to ensure the scientificity of the merit index calculation. Then, the digital twin model is used to extract the machine tool and its parameter set corresponding to the optimal polishing quality, and the parameters of other machine tools are adjusted based on this. The polishing quality after adjustment is monitored, and the parameter setting is dynamically optimized. If the adjustment effect is better than the current quality, the machine tool parameters are updated; otherwise, the original parameters are kept. In addition, a cascade adjustment mechanism is introduced to gradually optimize the suboptimal machine tool, and finally the overall polishing quality is improved.

[0194] Some data in the formulas described above are dimensionless for numerical calculation, and the contents not described in detail in the specification all belong to the existing technology known to those skilled in the art.

[0195] The above is only an example and description of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the invention or exceed the scope defined by the present claims, and they should belong to the protection scope of the present application.

[0196] It needs to be declared that all user data collected in this application are collected with the consent and authorization of the user. The use of user data is legal and compliant, and the use and processing of user data comply with relevant laws, regulations and standards in the relevant region.

Claims

1. A real-time evaluation and adjustment system for grinding quality of a gear of a speed reducer, characterized in that, The system comprises: A reducer gear image acquisition module, which captures the reducer gear after polishing operation through an industrial camera in real time, measures and maps the reducer gear in combination with a pre-constructed infrared light curtain, determines the geometric center Q1 of the reducer gear, maps the geometric center Q1 to the reducer gear image, and acquires the reducer gear image G_A; The three-dimensional point cloud data of the reducer gear A measured and mapped by the infrared light curtain is projected to the imaging plane of the industrial camera by using a preset spatial mapping matrix, and the reducer gear pixel image G_A` of A in G_A is determined; A reducer gear polishing quality evaluation module, which segments the reducer gear image after mapping the geometric center, determines a plurality of tooth images associated with the reducer gear, obtains a tooth image sequence TG1, TG2,..., TGj, and performs correlation analysis on all the tooth images to evaluate the polishing quality of the corresponding reducer gear in the following specific manner: The total amount of pixel points in each tooth image in TG1, TG2,..., TGj is counted, and the total amount of pixel points is sequentially recorded as a total amount of pixel points sequence SUM1, SUM2,..., SUMj according to TG1, TG2,..., TGj; The average value of the j total amounts of pixel points in SUM1, SUM2,..., SUMj is obtained, and the average value of the total amount of pixel points SUM_avg is obtained; The total amount of pixel points SUMi of any tooth image TGi is taken, the absolute value of the difference between SUMi and SUM_avg is calculated, and the absolute value is divided by SUM_avg and multiplied by 100% to obtain the pixel deviation rate δ of SUMi, wherein i is a count index, and 1≤i≤j; The δ is compared with a preset pixel deviation rate threshold δ_yu; If δ≥δ_yu, it is determined that the reducer gear A has an inter-tooth volume difference defect, and the first type of defect flag F1 is incremented by one, wherein the initial value of F1 is 0; The geometric centers of the j tooth images are determined, and j tooth geometric centers are obtained; The j tooth geometric centers are mapped to the reducer gear pixel image G_A` and labeled; The geometric center Q1 of G_A` is obtained and labeled; J lines are obtained by connecting the j tooth geometric centers as starting points and the geometric center Q1 as an ending point, respectively; The j angles are formed by the j lines and the geometric center Q1; The reference angle value is obtained by dividing the central angle by j; The angle value difference rate between the angle value of any one of the j angles and the reference angle value is obtained by calculating the absolute value of the difference between the angle value and the reference angle value, dividing the absolute value by the reference angle value, and multiplying by 100%; If the angle value difference rate is greater than or equal to a preset angle value difference rate threshold, it is determined that the reducer gear A has a non-uniform division defect, and the second type of defect flag F2 is incremented by one, wherein the initial value of F2 is 0; The polishing quality index E of A is calculated by E=w1*F1+w2*F2, wherein w1 and w2 are preset calculation weights, w1 and w2 are both greater than 0, and w1+w2=1; If E is less than a preset polishing quality index threshold E_yu, the reducer gear A is qualified, otherwise, the reducer gear A is unqualified. The polishing machine digital twin construction module acquires all polishing machines and extracts all polishing machine associated machining parameter groups and environment parameter groups to construct all polishing machine associated polishing machine digital twin models; The polishing quality of any one polished reduction gear is bound to the polishing machine digital twin model associated with the corresponding polishing machine to construct a polishing processing feature; The polishing machine machining parameter adjustment module performs correlation analysis based on the determined polishing processing features, evaluates the polishing machine digital twin model associated with the optimal polishing quality, and acquires the machining parameter group and the environment parameter group associated with the polishing machine digital twin model as an optimal adjustment parameter set to adjust the polishing machine digital twin models associated with the remaining polishing machines. Each polishing machine digital twin model outputs an adjustment result, and the remaining polishing machines are adjusted based on the adjustment result.

2. The system of claim 1, wherein, In the reduction gear image acquisition module, the specific way to acquire the reduction gear image is: Take any reduction gear A; Place A on the industrial shooting table and use an industrial camera to take a picture to acquire the reduction gear image G_A of A; The industrial shooting table includes a built-in three-axis MEMS gravity sensor for bearing the reduction gear, and the object carrier has a horizontal calculation function to determine the inclination of the object carrier in the horizontal plane and output the normal vector n of the object carrier; When the object carrier detects A, it sends a detection start signal to the infrared light curtain; The infrared light curtain is a four-beam infrared fan-shaped light curtain that simultaneously scans the outer edge of the tooth profile of A to obtain four polar coordinate sets {P1, P2, P3, P4} of the addendum circle of A in the light curtain coordinate system; Determine the geometric center Q1 of the addendum circle based on {P1, P2, P3, P4} and send an adjustment signal to the industrial camera; Adjust the optical axis direction of the industrial camera to coincide with the normal vector n, and vertically raise and lower along the Z axis. When the pixel distance between the projection point of the geometric center Q1 on the imaging target surface of the industrial camera and the physical center of the target surface is less than a predetermined pixel distance threshold, perform a shooting operation.

3. The system of claim 1, wherein, In the reduction gear polishing quality evaluation module, the specific way to determine the several tooth images associated with the reduction gear is: Determine the geometric center Q1 in G_A`, and expand the circle U outward with Q1 as the center and a pixel point as the step length. Continue to increase the radius until U reaches the maximum state in G_A`, where all pixel points in U belong to G_A`; Crop U from G_A` and remove it to obtain the remaining pixel region; The remaining pixel region contains several closed regions, and the total number of closed regions is j; The j closed regions are the tooth images of the j teeth on A; Arrange the j tooth images in clockwise order to obtain the tooth image sequence TG1, TG2,..., TGj.

4. The system of claim 3, wherein, In the reduction gear polishing quality evaluation module, the specific way to evaluate the polishing quality of the corresponding reduction gear also includes: If δ < δ_yu, do not process. Similarly, j tooth images in TG1, TG2,..., TGj are processed, and the first type of defect flag F1 is updated.

5. The system of claim 4, wherein, In the reducer gear polishing quality evaluation module, the specific way of evaluating the polishing quality of the corresponding reducer gear further comprises: If the angle value difference rate is less than the preset angle value difference rate threshold, no processing is performed; Similarly, the angle values of the j angles are processed, and F2 is updated.

6. The system of claim 1, wherein, In the polishing machine digital twin construction module, the specific way of constructing the polishing machine digital twin model associated with all polishing machines is: All polishing machines are obtained, and the total number is recorded as u; The u polishing machines are recorded as a polishing machine sequence MT1, MT2,..., MTu in the order of acquisition; Obtain the environmental parameters of each polishing machine processing the reducer gear, and combine them as the environmental parameter group; Obtain the processing parameters predefined by the operator for each polishing machine, and combine them as the processing parameter group; Based on the digital twin technology, construct the polishing machine digital twin model associated with each of the u polishing machines, and perform parameter injection on the polishing machine digital twin model associated with each polishing machine based on the processing parameters and environmental parameters of each polishing machine. Get the polishing machine digital twin model sequence PMT1, PMT2,..., PMTu.

7. The system of claim 6, wherein, In the polishing machine digital twin construction module, the specific way of constructing the polishing processing feature is: Obtain any one of the polishing machines MTo in the polishing machine sequence MT1, MT2,..., MTu and its corresponding polishing machine digital twin model PMTo, where o is a count index, 1≤o≤u; Take the total number of polishing H preset by the operator; Determine the total number of qualified Cout1, the total number of unqualified Cout2, and the average polishing quality index E of the H reducer gears associated with the polishing machine MTo closest to the current time, and record it as the average polishing quality index E_avg; Bind E_avg, Cout1, and Cout2 with PMTo as the polishing processing feature PPFo. Similarly, each polishing machine in the polishing machine sequence MT1, MT2,..., MTu is processed synchronously, u polishing processing features are determined, and the polishing processing feature sequence PPF1, PPF2,..., PPFu is recorded in the order of MT1, MT2,..., MTu.

8. The system of claim 7, wherein, In the polishing machine processing parameter adjustment module, the specific way of evaluating the polishing machine digital twin model associated with the optimal polishing quality is: Take the polishing processing feature sequence PPF1, PPF2,..., PPFu; Extract the average polishing quality index E_avg, the total number of qualified Cout1, and the total number of unqualified Cout2 in any one polishing processing feature PPFo; Calculate the polishing machine merit index YL of PPFo using YL=(Cout1–Cout2) / (Cout1+Cout2)-λ*E_avg, where λ is a dimensionless scaling constant; Similarly, determine the polishing machine merit index of each polishing processing feature in PPF1, PPF2,..., PPFu; PPF1, PPF2, …, PPFu are arranged in descending order of polishing machine bed merit index, and a polishing treatment feature sequence PPF1`, PPF2`, …, PPFu` is obtained; PPF1` and the polishing machine digital twin model PMT1` associated with PPF1` are taken from the first polishing treatment feature PPF1` in PPF1`, PPF2`, …, PPFu` and the polishing machine digital twin model PMT1` associated with PPF1`; The polishing machine digital twin model PMT1` is the polishing machine digital twin model associated with the optimal polishing quality.

9. The system of claim 8, wherein, In the polishing machine processing parameter adjustment module, the specific way to determine the dimensionless scaling constant λ is: The operator pre-acquires k full qualified-zero defect reduction gears and k full unqualified-high defect reduction gears as limit samples, and k is a preset value; Full qualified-zero defect corresponds to Cout2=0 and E_avg=0, and λ is 1; Full unqualified-high defect corresponds to Cout1=0 and E_avg=E_max, and λ is -1; In summary, λ is solved, and λ∈[-1, 1]; λ is a fixed value and no longer changes.

10. The system of claim 8, wherein, In the polishing machine processing parameter adjustment module, the specific way to adjust the remaining polishing machines based on the adjustment result is: Obtain the processing parameter set and the environment parameter set of the polishing machine digital twin model PMT1` with the optimal polishing quality, and combine them as the optimal adjustment parameter set; Take u-1 polishing machine digital twin models from the u polishing machine digital twin models except PMT1`; Adjust the processing parameter set and the environment parameter set of the u-1 polishing machine digital twin models to the processing parameter set and the environment parameter set in the optimal adjustment parameter set; Continuously monitor the adjustment results output by the u-1 polishing machine digital twin models after polishing H reduction gears, and the adjustment result is the average polishing quality index; If the average polishing quality index output by any polishing machine digital twin model is better than the current average polishing quality index of the corresponding polishing machine, the corresponding polishing machine is adjusted according to the optimal adjustment parameter set; Otherwise, no processing is performed.

Images