A control system for multi-axis machine tool processing based on CNC

Abstract

The present invention relates to the technical field of multi-axis machine tools, and in particular to a control system for multi-axis machine tool processing based on CNC, comprising a control terminal, a construction layer, an identification layer and a judgment layer; the control terminal is the main control end of the system, and is used to issue control commands; the specification parameters of manufactured parts are uploaded through the construction layer, and the construction layer synchronously constructs a three-dimensional model of the part based on the uploaded part specification parameters, and obtains model images of the three-dimensional model of the part from various perspectives based on the three-dimensional model of the part. The present invention compares the collected part image with the model image to make a qualified judgment on the part, and when the judgment result is unqualified, it can also use the unqualified judgment source part image as data support to reversely sniff the unqualified position on the part, thereby providing data reference for system end users, and obtaining the source workstation of the unqualified position of the part on the multi-axis machine tool, thereby bringing error and fault prediction effects to the multi-axis machine tool.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-axis machine tools, and in particular to a CNC-based multi-axis machine tool processing control system. Background Art

[0002] A multi-axis machine tool refers to a machine tool that has at least 1-2 rotating coordinate axes in addition to the three moving coordinate axes of x, y, and z, that is, a 4-5-axis CNC machine tool;

[0003] Multi-spindle machine tools are widely used in the machinery industry for drilling and tapping porous parts. These include automotive and motorcycle porous parts such as engine cases, aluminum casting housings, brake drums, brake discs, steering gears, wheel hubs, differential housings, axle heads, half shafts, axles, pumps, valves, hydraulic components, and solar components.

[0004] However, when current multi-axis machine tools process workpieces, slight looseness or positional deviation may occur in the components of each station on the machine tool, which will cause errors in the workpieces produced by the multi-axis machine tool. Although these errors will not temporarily cause the multi-axis machine tool to produce unqualified workpieces, the accumulation of errors will eventually lead to excessive errors in the workpieces produced by the multi-axis machine tool, resulting in unqualified workpieces.

[0005] In the existing technology, the focus is often on detecting existing faults in multi-axis machine tools, but it is unable to capture workstations on multi-axis machine tools that are prone to failure. As a result, during the process of producing workpieces on multi-axis machine tools, the quality control process of producing workpieces consumes a relatively large proportion of cost expenditure. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides a CNC-based multi-axis machine tool control system, which solves the technical problems raised in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] A CNC-based multi-axis machine tool processing control system includes a control terminal, a construction layer, an identification layer, and a determination layer;

[0009] The control terminal is the main control terminal of the system and is used to issue control commands;

[0010] The specification parameters of the manufactured parts are uploaded through the construction layer. The construction layer synchronously constructs a three-dimensional model of the part based on the uploaded specification parameters of the part, and obtains model images of the three-dimensional model of the part at each perspective based on the three-dimensional model of the part. The obtained images of the three-dimensional model of the part at each perspective form a comparison data group. The recognition layer synchronously selects a model image from the comparison data group, and based on the selected model image source perspective, performs image acquisition on the part output by the multi-axis machine tool. Based on the model image source perspective, a similarity comparison is performed on the model image and the acquired part image. The judgment layer further receives the similarity comparison result, determines whether the part is qualified, and when the result of the judgment is that the part is unqualified, sniffs the processing station with errors on the multi-axis machine tool based on the part image and the model image;

[0011] The recognition layer includes a selection module, an identification module, an acquisition module, and a comparison module. The selection module is used to obtain the control data group obtained in the construction layer and select a model image from the control data group. The identification module is used to receive the model image selected by the selection module and identify the source perspective of each model image. The acquisition module is used to receive the source perspective of each model image identified in the identification module and perform image acquisition of the parts output from the multi-axis machine tool based on the source perspective of the model image. The comparison module is used to compare the similarity of the corresponding model image and the acquired part image under the same source perspective, and feed back the similarity comparison result to the decision layer.

[0012] The similarity between the model image and the collected part image in the comparison module is calculated using the following formula:

[0013]

[0014] Where: sim(P,Q) is the similarity between the model image P and the part image Q; w is the set of feature points in the image; is the weighted shape factor of the εth feature point in the model image P; is the weighted shape factor mean of all feature points in the model image P; is the weighted shape factor of the εth feature point in the part image Q; is the weighted shape factor mean of all feature points in the part image Q;

[0015] Among them, the larger the value of sim(P,Q), the more similar the model image is to the part image; conversely, the smaller the value, the less similar the model image is to the part image.

[0016] Furthermore, the construction layer includes an upload module, a construction module, and a generation module. The upload module is used to upload the specification parameters of the part. The construction module is used to obtain the specification parameters of the part uploaded in the upload module and build a three-dimensional model of the part based on the specification parameters. The generation module is used to set the model image capture logic and capture the model image on the three-dimensional model of the part based on the model image capture logic.

[0017] Among them, the part specification parameters uploaded in the upload module include: the length of each edge on the part, and the inclination angle and distance between each edge. The three-dimensional part model constructed based on the part specification parameters in the construction module is composed of several groups of points and lines. The combination of model images captured based on the model image capture logic in the generation module is the reference data group.

[0018] Furthermore, the model image capture logic set in the generation module is set based on the complexity of the part's three-dimensional model. The model image capture logic is expressed as:

[0019]

[0020] Where: M is the number of model image captures; η is the total number of faces of the part 3D model; f is the total number of corner points on the part 3D model; n is the set of corner points on the part 3D model; α i is the relative position coordinate of the i-th corner point in the model construction space on the three-dimensional model of the part; β near is the relative position coordinate of the corner point closest to the i-th corner point on the three-dimensional model of the part in the model construction space; τ is a constant;

[0021] Among them, the number of model image captures M is further normalized to integers, y(x) represents the distance calculation function, then y(α i -β near ) represents α i and β near The straight-line distance between them is constant 2 ≥ τ ≥ 1, and obeys The larger the value of , the larger the value of constant τ.

[0022] Furthermore, It is used to represent the complexity of the three-dimensional model of the part. After the model image capture number M is obtained, the output part revolves around the multi-axis machine tool based on the specified viewing angle. The model image capturing operation is performed once, and all captured model image combinations are recorded as the control data group;

[0023] The designated viewing angle used when capturing the model image is a set of viewing angles around the part's three-dimensional model that can capture the largest number of parts on the three-dimensional model.

[0024] Furthermore, when selecting a model image from the reference data group, the selection module calculates the information entropy of each model image, sorts the model images in the reference data group based on the calculated information entropy of the model image, and further selects the model image from the sorted reference data group;

[0025] When selecting model images, the more complex the part's 3D model is, the more model images you will select.

[0026] When selecting model images, based on the model image sorting results, select model images at the first and last positions of the sorted model images. The higher the complexity of the part 3D model, the higher the complexity of the model image. The larger the value of , the smaller it is;

[0027] Among them, u is the number of model images selected in the front position in the sorted model images; v is the number of model images selected in the back position in the sorted model images; the larger the information entropy value of the model image, the higher the sorting position in the reference data group.

[0028] Furthermore, the information entropy of the model image is obtained by the following formula:

[0029]

[0030] Where: H is the information entropy of the model image; f(i, j) is the number of joint features (i, j) in the model image; N is the scale of the model image;

[0031] Among them, the larger the information entropy H of the model image is, the more information the model image contains, and the greater the impact on the model qualification probability is. Conversely, the smaller the information entropy H of the model image is, the smaller the impact on the model qualification probability is.

[0032] Furthermore, the placement posture of the part three-dimensional model constructed in the construction layer in the model construction space is the same as the placement posture of the part output on the multi-axis machine tool;

[0033] Among them, the source perspectives of the model image identified in the recognition module include: a specified perspective and an orbital angle around the output part of the multi-axis machine tool.

[0034] Furthermore, the determination layer includes a determination module, a segmentation module and a sniffing module. The determination module is used to receive the similarity between the model image and the part image obtained by the comparison module in the recognition layer, set a similarity determination threshold, apply the similarity comparison result and the similarity determination threshold to compare, and determine whether the part from the part image is qualified. The segmentation module is used to monitor the determination result in the determination module. When the determination result is no, the model image and the part image are segmented and fed back to the recognition layer. The sniffing module is used to receive the segmented model image and part image fed back to the recognition layer by the segmentation module, and the recognition layer outputs data, and sniffs the error position on the part based on the output data of the recognition layer.

[0035] Among them, if the similarity comparison result is within the similarity judgment threshold, the part corresponding to the part image from which the similarity comparison result comes is judged to be qualified, otherwise, it is unqualified. When the segmentation module performs segmentation processing on the model image and the part image, the number of sub-images obtained by segmenting the model image and the part image is the same, and the size of each sub-image is equal. The proportion of the image occupied by the part in the model image and the part image is the same. When the segmentation module performs segmentation on the model image and the part image, the number of segmentations is expressed as x×x, where x is an integer greater than 1. The higher the complexity of the three-dimensional model of the part, the larger the value of x, and vice versa.

[0036] Furthermore, after the segmentation module feeds back the segmented model image and part image to the recognition layer, the recognition layer synchronously applies the comparison module to compare the similarity between the sub-images of each source model image and the sub-images of the source part image, and further sends the similarity comparison results to the judgment module in the judgment layer. The judgment module is applied to determine whether the part structure corresponding to the sub-image of each source part image is qualified, and then the sub-image of the corresponding source part image with an unqualified judgment result is obtained through the sniffing module. The system end user confirms the area on the part corresponding to the sub-image in the sub-image source part image based on the sub-image obtained by the sniffing module, and identifies the processing station on the multi-axis machine tool that processes the area according to the area on the part. The identified processing station is the processing station on the multi-axis machine tool with an error.

[0037] Furthermore, the selection module is electrically connected to the identification module, the acquisition module and the comparison module through a medium, the selection module is interactively connected to the generation module through a local area network, the generation module is electrically connected to the upload module and the construction module through a medium, the comparison module is interactively connected to the judgment module through a local area network, the judgment module is electrically connected to the segmentation model and the sniffing module through a medium, the construction layer, the identification layer and the judgment layer are interactively connected to the control terminal through the local area network, and the control terminal transmits control commands based on the local area network to feed back to the construction layer, the identification layer or the judgment layer.

[0038] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0039] The present invention provides a control system for multi-axis machine tool processing based on CNC. During operation, the system constructs a three-dimensional model of a part by uploading part specification parameters, and uses specified model image acquisition logic to acquire a large number of model images and aggregate them into a comparison data set for use as a data reference. When the multi-axis machine tool outputs the manufactured part, the system further acquires the part image, and finally compares the acquired part image with the model image to make a qualified judgment on the part. When the judgment result is unqualified, the system can also use the unqualified judgment source part image as data support to reversely sniff the unqualified position on the part, thereby providing data reference for system end users, acquiring the source workstation of the unqualified position on the part on the multi-axis machine tool, thereby bringing error and fault prediction effects to the multi-axis machine tool, ensuring that the qualified rate of parts output by the multi-axis machine tool is stably guaranteed, thereby saving the cost of the multi-axis machine tool production workpiece quality control process to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0041] Figure 1 This is a schematic diagram of the structure of a control system for multi-axis machine tool processing based on CNC;

[0042] Figure 2 This is a schematic diagram of the operating principle of the recognition module in the recognition layer of the present invention;

[0043] The numbers in the figure represent: 1. CNC multi-axis machine tool; 2. Conveyor belt; 3. Sample parts; 4. Annular electric slide; 5. Part image acquisition equipment. DETAILED DESCRIPTION

[0044] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] The present invention will be further described below with reference to the embodiments.

[0046] Example 1:

[0047] This embodiment is a CNC-based multi-axis machine tool processing control system, such as Figure 1 As shown, it includes a control terminal, a construction layer, an identification layer and a decision layer;

[0048] The control terminal is the main control terminal of the system and is used to issue control commands;

[0049] The specification parameters of the manufactured parts are uploaded through the construction layer. The construction layer synchronously constructs a three-dimensional model of the part based on the uploaded specification parameters of the part, and obtains model images of the three-dimensional model of the part at each perspective based on the three-dimensional model of the part. The obtained images of the three-dimensional model of the part at each perspective form a comparison data group. The recognition layer synchronously selects a model image from the comparison data group, and based on the selected model image source perspective, performs image acquisition on the part output by the multi-axis machine tool. Based on the model image source perspective, a similarity comparison is performed on the model image and the acquired part image. The judgment layer further receives the similarity comparison result, determines whether the part is qualified, and when the result of the judgment is that the part is unqualified, sniffs the processing station with errors on the multi-axis machine tool based on the part image and the model image;

[0050] The construction layer includes an upload module, a construction module, and a generation module. The upload module is used to upload part specifications and parameters. The construction module is used to obtain the part specifications and parameters uploaded in the upload module and build a 3D model of the part based on the part specifications and parameters. The generation module is used to set the model image capture logic and capture the model image on the part 3D model based on the model image capture logic.

[0051] The part specifications uploaded in the upload module include the length of each edge on the part, and the relative inclination angle and distance of each edge. The 3D model of the part constructed based on the part specifications in the construction module is composed of several groups of points and lines. The combination of model images captured based on the model image capture logic in the generation module is the reference data group.

[0052] The model image capture logic set in the generation module is set based on the complexity of the part's 3D model. The model image capture logic is expressed as:

[0053]

[0054] Where: M is the number of model image captures; η is the total number of faces of the part 3D model; f is the total number of corner points on the part 3D model; n is the set of corner points on the part 3D model; α i is the relative position coordinate of the i-th corner point in the model construction space on the three-dimensional model of the part; β near is the relative position coordinate of the corner point closest to the i-th corner point on the three-dimensional model of the part in the model construction space; τ is a constant;

[0055] Among them, the number of model image captures M is further normalized to integers, y(x) represents the distance calculation function, then y(α i -β near ) represents α i and β near The straight-line distance between them is constant 2 ≥ τ ≥ 1, and obeys The larger the value of , the larger the value of constant τ is.

[0056] The recognition layer includes a selection module, an identification module, an acquisition module and a comparison module. The selection module is used to obtain the control data group obtained in the construction layer and select a model image from the control data group. The identification module is used to receive the model image selected by the selection module and identify the source perspective of each model image. The acquisition module is used to receive the source perspective of each model image identified in the identification module and perform image acquisition of the parts output from the multi-axis machine tool based on the source perspective of the model image. The comparison module is used to compare the similarity of the corresponding model image and the acquired part image under the same source perspective, and feed back the similarity comparison result to the decision layer.

[0057] The similarity between the model image and the collected part image in the comparison module is calculated using the following formula:

[0058]

[0059] Where: sim(P,Q) is the similarity between the model image P and the part image Q; w is the set of feature points in the image; is the weighted shape factor of the εth feature point in the model image P; is the weighted shape factor mean of all feature points in the model image P; is the weighted shape factor of the εth feature point in the part image Q; is the weighted shape factor mean of all feature points in the part image Q;

[0060] The larger the value of sim(P,Q), the more similar the model image is to the part image; conversely, the smaller the value, the less similar the model image is to the part image.

[0061] The determination layer includes a determination module, a segmentation module and a sniffing module. The determination module is used to receive the similarity between the model image and the part image obtained by the comparison module in the recognition layer, set a similarity determination threshold, apply the similarity comparison result to the similarity determination threshold, and determine whether the part from which the part image comes is qualified. The segmentation module is used to monitor the determination result in the determination module. When the determination result is no, the model image and the part image are segmented and fed back to the recognition layer. The sniffing module is used to receive the segmented model image and part image fed back to the recognition layer by the segmentation module, and the recognition layer outputs data, and sniffs the error position on the part based on the output data of the recognition layer.

[0062] Among them, if the similarity comparison result is within the similarity determination threshold, then the part corresponding to the part image from which the similarity comparison result is derived is determined to be qualified, otherwise, it is determined to be unqualified. When the segmentation module performs segmentation processing on the model image and the part image, the number of sub-images obtained by segmenting the model image and the part image is the same, and the sizes of the sub-images are equal. The proportion of the image occupied by the part in the model image and the part image is the same. When the segmentation module performs segmentation on the model image and the part image, the number of segmentations is expressed as x×x, where x is an integer greater than 1. The higher the complexity of the part 3D model, the larger the value of x, and vice versa.

[0063] The selection module is electrically connected to the identification module, the acquisition module and the comparison module through a medium. The selection module is interactively connected to the generation module through a local area network. The generation module is electrically connected to the upload module and the construction module through a medium. The comparison module is interactively connected to the judgment module through the local area network. The judgment module is electrically connected to the segmentation model and the sniffing module through a medium. The construction layer, the identification layer and the judgment layer are interactively connected to the control terminal through the local area network. The control terminal transmits control commands based on the local area network to feed back to the construction layer, the identification layer or the judgment layer.

[0064] In this embodiment, the control terminal controls the operation of the construction layer, the upload module runs to upload the specification parameters of the parts, the construction module synchronously obtains the specification parameters of the parts uploaded in the upload module, and builds the three-dimensional model of the parts based on the specification parameters of the parts. The generation module post-operates to set the model image capture logic, and based on the model image capture logic, the model image is captured on the three-dimensional model of the part. The selection module synchronously runs to obtain the comparison data group obtained in the construction layer, selects the model image in the comparison data group, receives the model image selected by the selection module, and identifies the source perspective of each model image. The acquisition module then receives the source perspective of each model image identified in the recognition module, and performs image acquisition on the parts output on the multi-axis machine tool based on the source perspective of the model image. The comparison module runs to compare the similarity of the corresponding model image and the collected part image under the same source perspective, and feeds back the similarity comparison result to the judgment layer. Finally, the judgment module receives the similarity of the model image and the part image obtained by the comparison module in the recognition layer, sets the similarity judgment threshold, and applies the similarity comparison result to the similarity judgment threshold to determine whether the part from the part image is qualified. The segmentation module monitors the judgment result in the judgment module in real time. When the judgment result is no, the model image and the part image are segmented and fed back to the recognition layer. The sniffing module further receives the segmented model image and part image fed back to the recognition layer by the segmentation module, and the recognition layer outputs data. The error position on the part is sniffed based on the output data of the recognition layer.

[0065] The above settings enable the detection of qualified parts and the detection of errors and faulty workstations during the production process of multi-axis machine tools. This provides data support for system users, enables more predictable management and control of multi-axis machine tools, and guarantees the qualified rate of parts produced by multi-axis machine tools.

[0066] See also Figure 2 As shown in the figure, the operation logic of the acquisition module in the recognition layer to acquire part images is further demonstrated: the conveyor belt 2 is a group of workstations on the CNC multi-axis machine tool 1, and the sample part 3 produced by the multi-axis machine tool 1 is output by the conveyor belt 2. When the sample part 3 is transmitted on the conveyor belt 2, the annular electric slide rail 4 carries the part image acquisition device 5 to complete the continuous acquisition of the part image based on the specified viewing angle and the orbital angle of the part output by the multi-axis machine tool.

[0067] Example 2:

[0068] In terms of specific implementation, based on Example 1, this example refers to Figure 1 The following is a further detailed description of a CNC-based multi-axis machine tool control system in Example 1:

[0069] It is used to represent the complexity of the three-dimensional model of the part. After the model image capture number M is obtained, the output part revolves around the multi-axis machine tool based on the specified viewing angle. The model image capturing operation is performed once, and all captured model image combinations are recorded as the control data group;

[0070] The designated viewing angle used when capturing the model image is a set of viewing angles around the part's three-dimensional model that can capture the largest number of parts on the three-dimensional model.

[0071] Through the above settings, a limited logic is brought to the complexity of the three-dimensional model of the part, which provides a limit for the complexity of the three-dimensional model of the part used in the system operation and ensures the stable operation of the system.

[0072] like Figure 1 As shown, when the selection module selects a model image from the reference data group, it calculates the information entropy of each model image, sorts the model images in the reference data group based on the result of the information entropy calculation of the model image, and further selects the model image from the sorted reference data group;

[0073] When selecting model images, the more complex the part's 3D model is, the more model images you will select.

[0074] When selecting model images, based on the model image sorting results, select model images at the first and last positions of the sorted model images. The higher the complexity of the part 3D model, the higher the complexity of the model image. The larger the value of , the smaller it is;

[0075] Wherein, u is the number of model images selected at the front position in the sorted model images; v is the number of model images selected at the back position in the sorted model images; the larger the information entropy value of the model image, the higher the sorting position in the reference data group;

[0076] The information entropy of the model image is obtained by the following formula:

[0077]

[0078] Where: H is the information entropy of the model image; f(i, j) is the number of joint features (i, j) in the model image; N is the scale of the model image;

[0079] Among them, the larger the information entropy H of the model image is, the more information the model image contains, and the greater the impact on the model qualification probability is. Conversely, the smaller the information entropy H of the model image is, the smaller the impact on the model qualification probability is.

[0080] Through the above settings and the calculation of the information entropy of the model image, selection logic support is provided for the selection module in the system to select the model image in the reference data group, ensuring that the system can use model images with different data volumes according to the different parts currently produced by the multi-axis machine tool, and provide an appropriate amount of reference data for the output parts of the multi-axis machine tool.

[0081] like Figure 1 As shown, the placement posture of the part 3D model constructed in the construction layer in the model construction space is the same as the placement posture of the part output on the multi-axis machine tool;

[0082] Among them, the source perspectives of the model image identified in the recognition module include: a specified perspective and an orbital angle around the output part of the multi-axis machine tool.

[0083] like Figure 1 As shown, after the segmentation module feeds back the segmented model image and part image to the recognition layer, the recognition layer synchronously applies the comparison module to compare the similarity between the sub-images of each source model image and the sub-images of the source part image, and further sends the similarity comparison results to the judgment module in the judgment layer. The judgment module is applied to determine whether the part structure corresponding to the sub-image of each source part image is qualified, and then the sub-image of the corresponding source part image with an unqualified judgment result is obtained through the sniffing module. The system end user confirms the area on the part corresponding to the sub-image in the sub-image source part image based on the sub-image obtained by the sniffing module, and identifies the processing station on the multi-axis machine tool that processes the area according to the area on the part. The identified processing station is the processing station on the multi-axis machine tool with an error.

[0084] Through the above settings, further operation data support is provided for the operation of the judgment layer in the system, so that the judgment layer can make effective qualified judgments on the output parts of the multi-axis machine tool, and have a predictive monitoring effect on the workstations with errors and faults on the multi-axis machine tool, providing data reference for system users, and further controlling the multi-axis machine tool to eliminate predicted errors or faults.

[0085] In summary, in the above embodiment, during operation, the system constructs a three-dimensional model of the part by uploading part specification parameters, and uses the specified model image acquisition logic to collect a large number of model images and aggregate them into a comparison data set for use as a data reference. When the multi-axis machine tool outputs the manufactured part, the part image is further collected, and finally the collected part image is compared with the model image to make a qualified judgment on the part. When the judgment result is unqualified, the unqualified source part image can be used as data support to reversely sniff the unqualified position on the part, thereby providing data reference for the system end user, and obtaining the source workstation of the unqualified position on the part on the multi-axis machine tool, thereby bringing error and fault prediction effects to the multi-axis machine tool, ensuring that the qualified rate of parts produced by the multi-axis machine tool is stably guaranteed, thereby saving the cost of the multi-axis machine tool production workpiece quality control process to a certain extent.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A CNC-based multi-axis machine tool control system, characterized in that: Includes control terminal, construction layer, identification layer and decision layer; The control terminal is the main control terminal of the system and is used to issue control commands; The specification parameters of the manufactured parts are uploaded through the construction layer. The construction layer synchronously constructs a three-dimensional model of the part based on the uploaded specification parameters of the part, and obtains model images of the three-dimensional model of the part at each perspective based on the three-dimensional model of the part. The obtained images of the three-dimensional model of the part at each perspective form a comparison data group. The recognition layer synchronously selects a model image from the comparison data group, and based on the selected model image source perspective, performs image acquisition on the part output by the multi-axis machine tool. Based on the model image source perspective, a similarity comparison is performed on the model image and the acquired part image. The judgment layer further receives the similarity comparison result, determines whether the part is qualified, and when the result of the judgment is that the part is unqualified, sniffs the processing station with errors on the multi-axis machine tool based on the part image and the model image; The recognition layer includes a selection module, an identification module, an acquisition module, and a comparison module. The selection module is used to obtain the control data group obtained in the construction layer and select a model image from the control data group. The identification module is used to receive the model image selected by the selection module and identify the source perspective of each model image. The acquisition module is used to receive the source perspective of each model image identified in the identification module and perform image acquisition of the parts output from the multi-axis machine tool based on the source perspective of the model image. The comparison module is used to compare the similarity of the corresponding model image and the acquired part image under the same source perspective, and feed back the similarity comparison result to the decision layer. The similarity between the model image and the collected part image in the comparison module is calculated using the following formula: Where: sim(P,Q) is the similarity between the model image P and the part image Q; w is the set of feature points in the image; is the weighted shape factor of the εth feature point in the model image P; is the weighted shape factor mean of all feature points in the model image P; is the weighted shape factor of the εth feature point in the part image Q; is the weighted shape factor mean of all feature points in the part image Q; Among them, the larger the value of sim(P,Q), the more similar the model image is to the part image; conversely, the smaller the value, the less similar the model image is to the part image.

2. A CNC-based multi-axis machine tool control system according to claim 1, characterized in that: The construction layer includes an upload module, a construction module, and a generation module. The upload module is used to upload the specification parameters of the part. The construction module is used to obtain the specification parameters of the part uploaded in the upload module and build a three-dimensional model of the part based on the specification parameters. The generation module is used to set the model image capture logic and capture the model image on the three-dimensional model of the part based on the model image capture logic. Among them, the part specification parameters uploaded in the upload module include: the length of each edge on the part, and the inclination angle and distance between each edge. The three-dimensional part model constructed based on the part specification parameters in the construction module is composed of several groups of points and lines. The combination of model images captured based on the model image capture logic in the generation module is the reference data group.

3. A CNC-based multi-axis machine tool control system according to claim 2, characterized in that: The model image capture logic set in the generation module is set based on the complexity of the part's three-dimensional model. The model image capture logic is expressed as: Where: M is the number of model image captures; η is the total number of faces of the part's three-dimensional model; f is the total number of corner points on the three-dimensional model of the part; n is the set of corner points on the three-dimensional model of the part; α i is the relative position coordinate of the i-th corner point in the model construction space on the three-dimensional model of the part; β near is the relative position coordinate of the corner point closest to the i-th corner point on the three-dimensional model of the part in the model construction space; τ is a constant; Among them, the number of model image captures M is further normalized to integers, y(x) represents the distance calculation function, then y(α i -β near ) represents α i and β near The straight-line distance between them is constant 2 ≥ τ ≥ 1, and obeys The larger the value of , the larger the value of constant τ.

4. A CNC-based multi-axis machine tool control system according to claim 3, characterized in that: It is used to represent the complexity of the three-dimensional model of the part. After the model image capture number M is obtained, the output part revolves around the multi-axis machine tool based on the specified viewing angle. The model image capturing operation is performed once, and all captured model image combinations are recorded as the control data group; The designated viewing angle used when capturing the model image is a set of viewing angles around the part's three-dimensional model that can capture the largest number of parts on the three-dimensional model.

5. The CNC-based multi-axis machine tool control system according to claim 1, characterized in that: When selecting a model image from the reference data group, the selection module calculates the information entropy of each model image, sorts the model images in the reference data group based on the calculated information entropy of the model image, and further selects the model image from the sorted reference data group; When selecting model images, the more complex the part's 3D model is, the more model images you will select. When selecting model images, based on the model image sorting results, select model images at the first and last positions of the sorted model images. The higher the complexity of the part 3D model, the higher the complexity of the model image. The larger the value of , the smaller it is; Among them, u is the number of model images selected in the front position in the sorted model images; v is the number of model images selected in the back position in the sorted model images; the larger the information entropy value of the model image, the higher the sorting position in the reference data group.

6. A CNC-based multi-axis machine tool control system according to claim 5, characterized in that: The information entropy of the model image is obtained by the following formula: Where: H is the information entropy of the model image; f(i, j) is the number of joint features (i, j) in the model image; N is the scale of the model image; Among them, the larger the information entropy H of the model image is, the more information the model image contains, and the greater the impact on the model qualification probability is. Conversely, the smaller the information entropy H of the model image is, the smaller the impact on the model qualification probability is.

7. The CNC-based multi-axis machine tool control system according to claim 1, characterized in that: The placement posture of the part three-dimensional model constructed in the construction layer in the model construction space is the same as the placement posture of the part output on the multi-axis machine tool; Among them, the source perspectives of the model image identified in the recognition module include: a specified perspective and an orbital angle around the output part of the multi-axis machine tool.

8. The CNC-based multi-axis machine tool control system according to claim 1, characterized in that: The determination layer includes a determination module, a segmentation module and a sniffing module. The determination module is used to receive the similarity between the model image and the part image obtained by the comparison module in the recognition layer, set a similarity determination threshold, apply the similarity comparison result to the similarity determination threshold, and determine whether the part from the part image is qualified. The segmentation module is used to monitor the determination result in the determination module. When the determination result is no, the model image and the part image are segmented and fed back to the recognition layer. The sniffing module is used to receive the segmented model image and part image fed back to the recognition layer by the segmentation module, and the recognition layer outputs data, and sniffs the error position on the part based on the output data of the recognition layer; Among them, if the similarity comparison result is within the similarity judgment threshold, the part corresponding to the part image from which the similarity comparison result comes is judged to be qualified, otherwise, it is unqualified. When the segmentation module performs segmentation processing on the model image and the part image, the number of sub-images obtained by segmenting the model image and the part image is the same, and the size of each sub-image is equal. The proportion of the image occupied by the part in the model image and the part image is the same. When the segmentation module performs segmentation on the model image and the part image, the number of segmentations is expressed as x×x, where x is an integer greater than 1. The higher the complexity of the three-dimensional model of the part, the larger the value of x, and vice versa.

9. The CNC-based multi-axis machine tool control system according to claim 8, characterized in that: After the segmentation module feeds back the segmented model image and part image to the recognition layer, the recognition layer synchronously applies the comparison module to compare the similarity between the sub-images of each source model image and the sub-images of the source part image, and further sends the similarity comparison results to the judgment module in the judgment layer. The judgment module is applied to determine whether the part structure corresponding to the sub-image of each source part image is qualified, and then the sub-image of the corresponding source part image with an unqualified judgment result is obtained through the sniffing module. The system end user confirms the area on the part corresponding to the sub-image in the sub-image source part image based on the sub-image obtained by the sniffing module, and identifies the processing station on the multi-axis machine tool that processes the area according to the area on the part. The identified processing station is the processing station on the multi-axis machine tool with an error.

10. The CNC-based multi-axis machine tool control system according to claim 1, characterized in that: The selection module is electrically connected to the identification module, the acquisition module and the comparison module through a medium, the selection module is interactively connected to the generation module through a local area network, the generation module is electrically connected to the upload module and the construction module through a medium, the comparison module is interactively connected to the determination module through the local area network, the determination module is electrically connected to the segmentation model and the sniffing module through a medium, the construction layer, the identification layer and the determination layer are interactively connected to the control terminal through the local area network, and the control terminal transmits control commands based on the local area network to feed back to the construction layer, the identification layer or the determination layer.

Images