Positioning method for machining with a numerical control machine tool

Abstract

The application relates to the technical field of point cloud registration, in particular to a positioning method for numerical control machine tool machining, which comprises the following steps: taking the curvatures of standard data points in a standard point cloud as machining complexity; obtaining a real-time point cloud of a raw material to be machined; in one iteration of an ICP algorithm, a transformation matrix of the real-time point cloud is updated with the minimum matching error as the target until the change of the minimum matching error between two adjacent iterations is less than a preset value or the iteration number reaches a maximum number, and the transformation matrix is taken as a target matrix; and the raw material to be machined is adjusted based on the target matrix to complete positioning. According to the technical scheme, the influence of the machining complexity of each position point on the machining allowance of a structural member can be considered, and accurate positioning of the raw material to be machined is realized.

Description

Technical Field

[0001] This application relates to the field of point cloud registration technology, and in particular to a positioning method for CNC machine tool processing. Background Technology

[0002] Before using CNC machining to produce automotive structural parts, because automotive structural parts usually have complex curved surfaces, in order to ensure that each part of the raw material to be processed has sufficient machining allowance to produce a complete structural part that meets the design requirements, it is necessary to accurately position the raw material to be processed on a CNC machine tool. Therefore, how to achieve the positioning of the raw material to be processed on a CNC machine tool is an urgent problem to be solved.

[0003] Currently, patent application CN120147385A discloses a method and system for optimizing the machining allowance of a blank based on 3D scanning measurement. The method includes: collecting blank measurement point cloud data and design model point cloud data as raw point cloud data; simplifying the raw point cloud data to obtain point cloud normal vectors and redirecting them; aligning the blank measurement point cloud data and design model point cloud after point cloud simplification; performing initial segmentation on the point cloud feature surfaces, followed by further segmentation to establish a correspondence between the measurement point cloud and the design model point cloud feature surfaces; and establishing a mathematical model for machining allowance based on the correspondence between the measurement point cloud and the design model point cloud feature surfaces, constraining variance minimization to match allowance optimization. The method used to align the blank measurement point cloud data and the design model point cloud is the ICP algorithm.

[0004] The above method optimizes the rotation matrix and translation vector in the ICP algorithm by establishing a mathematical model of machining allowance and minimizing the constraint variance to match the allowance. However, the above method ignores the influence of machining complexity at each position point on the structural component on the machining allowance, and cannot guarantee that each position point on the structural component can obtain the required machining allowance, resulting in inaccurate positioning of the raw material to be processed. Summary of the Invention

[0005] To address the technical problem of inaccurate positioning of raw materials to be processed, this application provides a positioning method for CNC machine tool processing that can consider the influence of the machining complexity of each position point on the structural component on the machining allowance, thereby achieving accurate positioning of the raw materials to be processed.

[0006] In a first aspect, this application provides a positioning method for CNC machine tool machining. The positioning method includes: using the curvature of each standard data point in a standard point cloud as the machining complexity; acquiring the real-time point cloud of the raw material to be processed; updating the transformation matrix of the real-time point cloud in one iteration of the ICP algorithm with the goal of minimizing the matching error, until the change in the minimum matching error between two adjacent iterations is less than a preset value or the number of iterations reaches the maximum number, and then using the transformation matrix as the target matrix; adjusting the raw material to be processed based on the target matrix to complete the positioning; the matching error in one iteration is: acquiring the nearest standard data point in the standard point cloud for any real-time data point in the real-time point cloud to obtain multiple data point pairs; calculating the machining allowance for each real-time data point; comparing the machining allowance and the safety allowance to calculate the allowance penalty for each real-time data point in this iteration; using the product of the machining complexity and the allowance penalty in the data point pair as the attention of the data point pair in this iteration; transforming the real-time data points according to the transformation matrix, and then weighting and summing the Euclidean distances of each data point pair according to the normalized attention to obtain the matching error.

[0007] By introducing two key factors, processing complexity and processing allowance, into the ICP algorithm, the attention of different real-time data points can be dynamically adjusted in each iteration based on processing complexity and allowance penalty. This can take into account both structural complexity and actual processing requirements, effectively improve registration accuracy, and thus achieve high-precision positioning of raw materials on CNC machine tools, ensuring the processing quality of subsequent processing stages.

[0008] Preferably, using the curvature of each standard data point in the standard point cloud as the processing complexity includes: calculating the angle between the normal vectors of any standard data point and each adjacent data point, and using the average value of the angles as the curvature value of the standard data point, and using the normalized curvature value of each standard data point as the processing complexity.

[0009] Preferably, calculating the processing margin for each real-time data point includes: obtaining the normal vector of the standard data point and the margin vector pointing from the standard data point to the real-time data point in the data point pair to which the real-time data point belongs; the processing margin of the real-time data point is the product of the cosine of the angle between the margin vector and the normal vector and the magnitude of the margin vector.

[0010] Machining allowance can characterize the amount of cutting required to process the raw material into the target structural part. The cosine of the angle between the allowance vector and the normal vector can characterize whether the corresponding real-time data point is in a material shortage state. The magnitude of the allowance vector can characterize the deviation distance between the real-time data point and the standard point cloud, thus realizing the accurate quantification of machining allowance.

[0011] Preferably, adjusting the raw material to be processed based on the target matrix includes: the target matrix includes a target rotation matrix and a target translation matrix; the raw material to be processed is rotated according to the target rotation matrix; the raw material to be processed is translated according to the target translation matrix to complete the positioning.

[0012] Preferably, this iteration Real-time data points Residual penalty for:

[0013] ;

[0014] in, and The maximum and minimum values ​​of the margin penalty satisfy the following conditions: ; For real-time data points The machining allowance. This is for a safety margin.

[0015] If the machining allowance of a real-time data point is negative, it will lead to a defect at that data point, and the allowance penalty for that data point should be set to the maximum value. If the machining allowance of a real-time data point is positive and less than the safety allowance, a larger allowance penalty should be assigned to that data point to increase its impact on the matching error and achieve the desired result in this iteration. Precise quantification of residual penalties for each real-time data point.

[0016] Preferably, the residual penalty of any real-time data point in this iteration is also positively correlated with the upward trend of the residual penalty of the real-time data points in the previous iteration; the residual penalty of the real-time data points in the historical iterations is linearly fitted, and the slope of the linear fit and the maximum value of 0 are taken as the upward trend.

[0017] By linearly fitting the historical margin penalty value of each real-time data point, we can analyze its changing trend and determine whether the real-time data point is in a risky state where the margin cannot be met for a long time. This trend value is then used in the calculation of margin penalty in this iteration to give more attention to real-time data points with processing defect risks and improve the overall positioning accuracy.

[0018] Preferably, this iteration Real-time data points Residual penalty for:

[0019] ;

[0020] in, and The maximum and minimum values ​​of the margin penalty satisfy the following conditions: ; For real-time data points The machining allowance. For safety margin, For this iteration Previous real-time data points The upward trend of margin penalty.

[0021] Preferably, the first Each data point pair includes real-time data points. and standard data points This iteration The Middle The attention given to each data point pair in this iteration Real-time data points Residual penalty and standard data points The product of processing complexity.

[0022] Standard data points The processing complexity is a fixed static feature, while the real-time data points in each iteration... The residual penalty is constantly changing, so the residual penalty is a dynamic feature in the iteration process. The attention of each data point pair in one iteration is adaptively determined by combining static and dynamic features.

[0023] Preferably, the matching error is also related to the transformation increment of the transformation matrix in the current iteration, wherein the transformation increment is the Euclidean distance between the transformation matrices in the current iteration and the previous iteration.

[0024] Preferably, this iteration Matching error for:

[0025] ; and These are the rotation and translation matrices in the transformation matrix, respectively. The set consisting of all data point pairs. and The first Real-time data points and standard data points in a data point pair; For this iteration The Middle The level of attention given to each data point pair For this iteration The sum of attention to all data point pairs in the data; and These are the transformation increments after standardization in the current iteration and the previous iteration, respectively. This is the regularization coefficient.

[0026] This is used to measure the difference between the transformation increment of the current iteration and the transformation increment of the previous iteration, i.e., the degree of "oscillation" in the transformation trajectory; As part of the matching error, it can help the ICP algorithm escape local optima and improve the convergence speed of the ICP algorithm.

[0027] The technical solution of this application has the following beneficial technical effects:

[0028] The ICP algorithm is improved by introducing processing complexity and margin penalties. The processing complexity is calculated for each point in the standard point cloud based on its curvature. This complexity is then combined with the processing margin of each point in the real-time point cloud to dynamically adjust the attention weights during point cloud registration, thus giving higher matching priority to key processing areas. Simultaneously, the smoothness of the transformation trajectory is used as a regularization term in the calculation of matching error, effectively suppressing oscillations during the iteration process, prompting the ICP algorithm to escape local optima, and improving convergence speed and stability. The final target transformation matrix achieves precise positioning of the raw material while ensuring sufficient margin at each processing position, thereby significantly improving processing quality. Attached Figure Description

[0029] Figure 1 This is a flowchart of a positioning method for CNC machine tool machining according to an embodiment of this application.

[0030] Figure 2 This is a schematic diagram of the alignment effect after one iteration of the ICP algorithm according to an embodiment of this application.

[0031] Figure 3 This is a schematic diagram of the final point cloud alignment effect of the ICP algorithm according to an embodiment of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0033] According to a first aspect of this application, this application provides a positioning method for CNC machine tool machining. Figure 1 This is a flowchart illustrating a positioning method for CNC machine tool machining according to an embodiment of this application. Figure 1 As shown, the positioning method for CNC machine tool machining includes steps S101 to S103, which are described in detail below.

[0034] S101 uses the curvature of each standard data point in the standard point cloud as the processing complexity.

[0035] In one embodiment, the standard point cloud is the point cloud data of the target structural component, and the standard point cloud includes multiple standard data points. The target structural component can be a workpiece requiring processing, such as an automotive structural component or an automotive mold.

[0036] Specifically, using the curvature of each standard data point in the standard point cloud as the processing complexity includes: calculating the angle between the normal vectors of any standard data point and each adjacent data point, and using the average value of the angles as the curvature value of the standard data point, and using the normalized curvature value of each standard data point as the processing complexity.

[0037] In this embodiment, the neighboring data points of a standard data point are the K nearest standard data points in the standard point cloud. In this embodiment, K is set to 5. The curvature values ​​of each standard data point can be normalized using the Sigmoid function, making the processing complexity range from [value missing]. .

[0038] In this way, the processing complexity of each standard data point in the standard point cloud can be accurately quantified. The greater the processing complexity, the more complex the surface features at the corresponding standard data point, and the greater the processing difficulty. In the subsequent CNC machining process, sufficient processing allowance should be allocated to the standard data points with greater processing complexity to ensure the machining accuracy of the structural parts.

[0039] S102: Obtain the real-time point cloud of the raw material to be processed. In one iteration of the ICP algorithm, update the transformation matrix of the real-time point cloud with the goal of minimizing the matching error. When the change in the minimum matching error between two adjacent iterations is less than the preset value or the number of iterations reaches the maximum number, the transformation matrix is ​​used as the target matrix.

[0040] In one embodiment, a real-time point cloud of the raw material to be processed is obtained. The raw material to be processed is a blank, and it needs to be CNC machined to process it into a target structural component.

[0041] To facilitate understanding, a complete iteration process in the traditional ICP algorithm is described here: First, obtain the nearest standard data point in the standard point cloud for any real-time data point, resulting in multiple data point pairs; then, calculate the matching error, which reflects the sum of the Euclidean distances between each data point pair after transforming the real-time data points in the real-time point cloud (i.e., the deviation between the real-time point cloud and the standard point cloud). The smaller the deviation, the greater the overlap between the real-time point cloud and the standard point cloud, and the better the registration effect; therefore, with the goal of minimizing the matching error, solve for the transformation matrix corresponding to the minimum matching error, and the transformation matrix of the real-time point cloud in this iteration can be obtained. Thus, a complete iteration process ends.

[0042] Each real-time data point will find a corresponding standard data point. For real-time data points in the real-time point cloud... Its corresponding standard data points Satisfying the relation:

[0043] ;in, For standard point clouds, For any standard data point in the standard point cloud, For real-time data points, For real-time data points and standard data points The Euclidean distance between them. Thus, real-time data points and standard data points Formed the first Data point pairs.

[0044] The transformation matrix includes a rotation matrix and a translation matrix. In the ICP algorithm, the transformation matrix can be used to transform each real-time data point in the real-time point cloud. For example, real-time data points... The transformed coordinates are: ,in, For real-time data points coordinates and These are the rotation matrix and the translation matrix, respectively.

[0045] In the traditional ICP algorithm, the matching error in each iteration is the sum of the Euclidean distances of each data point pair, treating each data point pair equally. However, in the scenario of CNC machine tool machining, the machining complexity of each position in the target structural part is different, and after transforming each real-time data point in the real-time point cloud using the transformation matrix, the machining allowance of each real-time data point relative to the standard point cloud will also be different. Therefore, in each iteration, different attention should be assigned to each data point according to the machining complexity and the machining allowance of each real-time data point relative to the standard point cloud to ensure that each position in the target structural part can achieve the machining allowance that meets the safety margin, thereby ensuring machining accuracy.

[0046] Specifically, the matching error in one iteration is as follows: obtain the nearest standard data point in the standard point cloud for any real-time data point in the real-time point cloud, and obtain multiple data point pairs; calculate the processing margin for each real-time data point; compare the processing margin and the safety margin to calculate the margin penalty for each real-time data point in this iteration; take the product of the processing complexity and the margin penalty in the data point pair as the attention of the data point pair in this iteration; after transforming the real-time data points according to the transformation matrix, calculate the weighted sum of the Euclidean distance of each data point pair according to the normalized attention to obtain the matching error.

[0047] The machining allowance represents the amount of cutting required to process the raw material into the target structural part. When the machining allowance of a real-time data point is 0, it means that the real-time data point falls exactly on the surface of the target structural part, and no further finishing is required to meet the shape requirements at the corresponding position of the real-time data point. When the machining allowance of a real-time data point is greater than 0, it means that the real-time data point falls on the outside of the target structural part, and the raw material can be finished to remove the excess part so that the removed real-time data point falls on the surface of the target structural part. In this case, the machining allowance should be greater than the safety allowance to ensure the machining accuracy of the subsequent finishing. When the machining allowance of a real-time data point is less than 0, it means that the real-time data point falls on the inside of the target structural part, and the real-time data point is in a short-material state, which will lead to the target structural part being unqualified. Specifically, calculating the processing margin for each real-time data point includes: obtaining the normal vector of the standard data point and the margin vector pointing from the standard data point to the real-time data point in the data point pair to which the real-time data point belongs; the processing margin of the real-time data point is the product of the cosine of the angle between the margin vector and the normal vector and the magnitude of the margin vector.

[0048] Among them, real-time data points machining allowance for:

[0049] ;in, Standard data points The normal vector, To start from standard data points Pointing to real-time data points The remaining vector, Residual vector The length of the module.

[0050] In one embodiment, after obtaining the processing allowance for each real-time data point, the processing allowance and the safety allowance are compared to calculate the allowance penalty for each real-time data point. If the processing allowance of a real-time data point is negative, it will cause a defect at the real-time data point, and the allowance penalty for that real-time data point should be set to the maximum value. If the processing allowance of a real-time data point is positive and the processing allowance is less than the safety allowance, a larger allowance penalty should be assigned to that real-time data point to increase the degree of influence of that real-time data point on the matching error.

[0051] Specifically, this iteration Real-time data points Residual penalty for:

[0052] ;

[0053] in, and The maximum and minimum values ​​of the margin penalty satisfy the following conditions: ; For real-time data points The machining allowance. For safety margin. In the embodiments of this application, The value is 10. The value is set to 1 to ensure that the margin penalty for each real-time data point is greater than 0, thereby ensuring that the matching error can be constrained to all real-time data points in the real-time point cloud. The safety margin can be determined based on the tool diameter. According to engineering experience, the safety margin is 0.03 times the tool diameter, and the implementer can adjust it according to the requirements of machining accuracy.

[0054] In another optional embodiment, the margin penalty of a real-time data point can reflect whether the processing margin of the real-time data point meets the requirements. The smaller the margin penalty, the better the processing margin of the real-time data point. Ideally, the margin penalty of a real-time data point should gradually decrease until it stabilizes during multiple iterations. If the margin penalty of a real-time data point shows an upward trend, it means that the processing margin that meets the safety margin cannot be obtained at the real-time data point. At this time, more attention should be allocated to the real-time data point. Therefore, the attention of any data point pair is also related to the upward trend of the margin penalty of the real-time data point in the data point pair.

[0055] Specifically, the residual penalty for any real-time data point in this iteration is positively correlated with the upward trend of the residual penalty for the real-time data points mentioned in the previous iteration; the residual penalty for the real-time data points mentioned in the historical iterations is linearly fitted, and the slope of the linear fit and the maximum value of 0 are taken as the upward trend.

[0056] This iteration Real-time data points Residual penalty for:

[0057] ;

[0058] in, and The maximum and minimum values ​​of the margin penalty satisfy the following conditions: ; For real-time data points The machining allowance. For safety margin, For this iteration Previous real-time data points The upward trend of margin penalty.

[0059] In one embodiment, the first Each data point pair includes real-time data points. and standard data points This iteration The Middle Attention to each data point pair for: ; For this iteration Real-time data points Residual penalty Standard data points The processing complexity is reduced. During the iteration of the ICP algorithm, greater attention is given to data points with higher processing complexity and larger margin penalties. This allows the ICP algorithm to prioritize reducing the deviation of standard data points with higher processing complexity, thereby avoiding subsequent processing of real-time data points with higher processing complexity and reducing processing difficulty.

[0060] Understandably, standard data points The processing complexity is fixed and can be considered a static feature. However, in each iteration, the transformation matrix changes continuously, leading to variations in the real-time data points. The processing allowance changes continuously, that is, the real-time data points in each iteration The residual penalty is constantly changing. It can be regarded as a dynamic feature in the iteration process. The attention of each data point pair in an iteration is adaptively determined by combining static and dynamic features.

[0061] In one embodiment, after determining the attention level of each data point pair in this iteration, this iteration... Matching error for: ; and These are the rotation and translation matrices in the transformation matrix, respectively. The set consisting of all data point pairs. and The first Real-time data points and standard data points in a data point pair; For this iteration The Middle The level of attention given to each data point pair For this iteration The sum of attention to all data point pairs in the data.

[0062] In another embodiment, each iteration of the ICP algorithm produces a transformation increment of a transformation matrix. These transformation increments constitute a transformation trajectory in a six-dimensional pose space (where the rotation matrix in the transformation matrix contains 3 parameters and the translation matrix in the transformation matrix contains 3 parameters). The smoothness of the transformation trajectory reflects the stability of the convergence process. If the transformation trajectory exhibits severe oscillations, it indicates that the ICP algorithm is wandering between multiple local optima, which will cause the ICP algorithm to fail to converge quickly. Therefore, to ensure that the ICP algorithm can converge quickly, the smoothness of the transformation trajectory is used as a regularization term in the calculation of the matching error.

[0063] Specifically, the matching error is also related to the transformation increment of the transformation matrix in the current iteration, where the transformation increment is the Euclidean distance between the transformation matrices in the current iteration and the previous iteration; in the current iteration Matching error for:

[0064] ; and These are the rotation and translation matrices in the transformation matrix, respectively. The set consisting of all data point pairs. and The first Real-time data points and standard data points in a data point pair; For this iteration The Middle The level of attention given to each data point pair For this iteration The sum of attention to all data point pairs in the data; As a normalization factor, used for... Perform normalization; and These are the transformation increments after standardization in the current iteration and the previous iteration, respectively. This is the regularization coefficient.

[0065] Wherein, regularization coefficient The value is 0.5; normalization factor The value is .

[0066] Understandably, Used to measure the difference between the transformation increment of the current iteration and the transformation increment of the previous iteration, i.e. the degree of "oscillation" of the transformation trajectory; The larger the value of , the greater the fluctuation of the ICP algorithm between local optima, and the more unstable the matching process. In this case, As part of the matching error, it can help the ICP algorithm escape local optima and improve the convergence speed of the ICP algorithm.

[0067] It should be noted that the change increment Includes rotation and translation matrices. To avoid inconsistencies in the dimensions of the parameters in the rotation and translation matrices, [the following is omitted]. The impact in calculation Previously, the transformation increment for each iteration was standardized to eliminate the dimensions of the parameters in the rotation and translation matrices, so that... It is a dimensionless numerical value.

[0068] Thus, the transformation matrix directly affects the magnitude of the matching error. With the goal of minimizing the matching error, the transformation matrix corresponding to the minimum matching error is obtained and used as the transformation matrix for this iteration. This updates the real-time point cloud transformation matrix, completing the current iteration. After transforming the real-time point cloud based on the transformation matrix of this iteration, the process moves to the next iteration, where multiple data point pairs are reacquired. Please refer to [link to previous section]. Figure 2 This is a schematic diagram of the alignment effect after one iteration of the ICP algorithm according to an embodiment of this application.

[0069] In one embodiment, after completing one iteration, the change in the minimum matching error between two adjacent iterations is calculated. If the change in the minimum matching error is less than a preset value, or the maximum number of iterations is reached, it indicates that the optimal transformation matrix has been obtained. In this case, the transformation matrix is ​​used as the target matrix. The preset value is 0.5, and the maximum number of iterations is 50. See also... Figure 3 This is a schematic diagram of the final point cloud alignment effect of the ICP algorithm according to an embodiment of this application.

[0070] In this way, the target matrix is ​​obtained. Based on the target matrix, the raw material to be processed is rotated and translated to adjust its position, thus achieving precise positioning of the raw material to be processed.

[0071] S103, adjust the raw material to be processed based on the target matrix to complete the positioning.

[0072] In one embodiment, the target matrix includes a target rotation matrix and a target translation matrix. The material to be processed is rotated according to the target rotation matrix and translated according to the target translation matrix to complete the machining positioning of the CNC machine tool, ensuring that sufficient machining allowance can be obtained at each position of the material to be processed, thereby improving the machining accuracy of the target structural parts.

[0073] It should be noted that any modifications and improvements made without departing from the concept of this application fall within the scope of protection of this application. Therefore, the scope of protection of this patent application shall be determined by the appended claims.

Claims

1. A positioning method for CNC machine tool machining, characterized in that, Location methods include: The curvature of each standard data point in the standard point cloud is used as the processing complexity, including: Calculate the angle between any standard data point and the normal vector between each adjacent data point, and use the average of the angles as the curvature value of the standard data point. Use the normalized curvature value of each standard data point as the processing complexity. The real-time point cloud of the raw material to be processed is obtained. In one iteration of the ICP algorithm, the transformation matrix of the real-time point cloud is updated with the goal of minimizing the matching error. The transformation matrix is ​​used as the target matrix when the change of the minimum matching error between two adjacent iterations is less than the preset value or the number of iterations reaches the maximum number. The raw material to be processed is adjusted based on the target matrix to complete the positioning; The matching error in one iteration is as follows: obtain the nearest standard data point in the standard point cloud for any real-time data point in the real-time point cloud, and obtain multiple data point pairs; calculate the processing margin for each real-time data point; compare the processing margin with the safety margin to calculate the margin penalty for each real-time data point in this iteration. The matching error is also related to the transformation increment of the transformation matrix in the current iteration. The transformation increment is the Euclidean distance between the transformation matrices in the current iteration and the previous iteration. Matching error for: ; and These are the rotation and translation matrices in the transformation matrix, respectively. The set consisting of all data point pairs. and The first Real-time data points and standard data points in a data point pair; For this iteration The Middle The level of attention given to each data point pair For this iteration The sum of attention to all data point pairs in the data; Normalization factor; and These are the transformation increments after standardization in the current iteration and the previous iteration, respectively. The regularization coefficient is used. The residual penalty of any real-time data point in this iteration is positively correlated with the upward trend of the residual penalty of real-time data points before this iteration; linear fitting is performed on the residual penalty of real-time data points in historical iterations, and the slope of the linear fitting and the maximum value of 0 are taken as the upward trend. This iteration Real-time data points Residual penalty for: ； in, and The maximum and minimum values ​​of the margin penalty satisfy the following conditions: ; For real-time data points The machining allowance. For safety margin, For this iteration Previous real-time data points The upward trend of margin penalty; The product of the processing complexity and the margin penalty in the data point pair is used as the attention of the data point pair in this iteration; after transforming the real-time data points according to the transformation matrix, the Euclidean distance of each data point pair is weighted and summed according to the normalized attention to obtain the matching error.

2. The positioning method for CNC machine tool machining according to claim 1, characterized in that, The calculation of the processing margin for each real-time data point includes: In the data point pair to which the real-time data point belongs, obtain the normal vector of the standard data point and the margin vector pointing from the standard data point to the real-time data point; the processing margin of the real-time data point is the product of the cosine of the angle between the margin vector and the normal vector and the magnitude of the margin vector.

3. The positioning method for CNC machine tool machining according to claim 1, characterized in that, Adjusting the raw materials to be processed based on the target matrix includes: The target matrix includes a target rotation matrix and a target translation matrix. The raw material to be processed is rotated according to the target rotation matrix and translated according to the target translation matrix to complete the positioning.

4. The positioning method for CNC machine tool machining according to claim 1, characterized in that, No. Each data point pair includes real-time data points. and standard data points This iteration The Middle The attention given to each data point pair in this iteration Real-time data points Residual penalty and standard data points The product of processing complexity.

Images