CN118268897A - Machining reference self-locating method for additive repairing bearing bush - Google Patents

Abstract

A processing reference self-locating method for an additive repairing bearing bush comprises the following steps: s1, acquiring point cloud data of the surface of a bearing bush; s2, preprocessing the point cloud data; s3, aligning the bearing bush surface with the point cloud model through point cloud registration; s4, after the curved surface is rebuilt, obtaining characteristic points of the surface of the bearing bush; s5, accurately extracting characteristic points based on the point cloud data, and further calculating the geometric center of the bearing bush; s6, determining a machining axis of the bearing bush through a curve fitting algorithm; s7, acquiring bearing bush positioning data in real time to determine machining reference and axis parameters, automatically adjusting the position and parameters of machining equipment through compensation logic and a three-dimensional reconstruction algorithm to enable the machining equipment to be matched with the actual shape and position of the bearing bush, enabling the bearing bush I to be placed on a machining platform, enabling a computer to automatically identify and determine parameters such as the machining reference and the axis, and enabling the position of the machining equipment to be matched with the bearing bush through the compensation method in real time, so that machining efficiency and machining precision are improved, and the bearing bush positioning device has good practical value and economic value.

Description

Machining reference self-locating method for additive repairing bearing bush

Technical Field

The invention relates to the technical field of mechanical intelligent machining, in particular to a machining reference self-locating method for an additive repairing bearing bush.

Background

In the machining process, determination of a machining reference is critical to machining accuracy and machining quality. The traditional processing reference determination method needs to rely on manual adjustment and measurement, and has the problems of low processing efficiency, low precision and the like. Therefore, a machining reference self-locating method based on bearing bush machining is urgently needed to improve machining efficiency and machining precision.

The bearing bush is an important part in mechanical equipment, and the bearing bush can be directly put into use after material reduction processing is carried out after material addition. The existing material reduction processing mainly relies on manual tile scraping, so that the requirements on the skills of constructors are high, the processing precision is poor, and popularization is not facilitated.

Disclosure of Invention

The invention aims to solve the technical problems that: the machining reference self-locating method for the material-increasing and repairing bearing bush can improve the machining precision of the material-decreasing of the bearing bush and solve the technical problem in the existing machining process of the material-decreasing of the bearing bush.

The technical scheme adopted by the invention is as follows: a processing reference self-locating method for an additive repairing bearing bush comprises the following steps:

S1, a machine tool performs point cloud scanning on a bearing bush clamped on a machine tool workbench through vision measuring equipment to acquire point cloud data on the surface of the bearing bush;

s2, preprocessing point cloud data, including filtering denoising, outlier rejection and coordinate conversion operation;

S3, aligning the bearing bush surface with the point cloud model through point cloud registration;

s4, after the curved surface is rebuilt, obtaining characteristic points of the surface of the bearing bush, wherein the characteristic points are determined through curvature, normal vector and curvature change indexes; wherein points with significant geometric features are selected as feature points;

S5, accurately extracting characteristic points based on the point cloud data, and further calculating the geometric center of the bearing bush to determine a machining reference;

s6, determining a machining axis of the bearing bush through a curve fitting algorithm;

S7, acquiring bearing bush positioning data in real time to determine machining reference and axis parameters, and automatically adjusting the position and parameters of machining equipment through compensation logic and a three-dimensional reconstruction algorithm to enable the machining equipment to be matched with the actual shape and position of the bearing bush.

The invention has the following beneficial effects:

1. According to the invention, an online measurement technology is utilized to generate the three-dimensional model of the bearing bush after material addition, the machining allowance is reasonably distributed through comparison with a theoretical model, and automatic programming is carried out, so that the automatic material reduction machining of the bearing bush is realized.

2. The machining reference self-locating method can realize automatic locating of the machining reference, so that the first bearing bush is clamped on the machining platform, the position of machining equipment can be adjusted to be matched with the bearing bush in real time through the compensation method, the system can automatically identify and determine parameters such as the machining reference, an axis and the like, and the machining efficiency and the machining precision are improved.

Drawings

The invention is further described below with reference to the drawings and examples.

Fig. 1 is a schematic diagram of feature point extraction according to the present invention.

Detailed Description

A processing reference self-locating method for an additive repairing bearing bush comprises the following steps:

S1, the machine tool scans point cloud of a bearing bush clamped on a workbench of the machine tool through vision measuring equipment, and point cloud data of the surface of the bearing bush are obtained.

And (3) installing visual measurement equipment on the machine tool, and performing point cloud scanning on the bearing bush by using visual measurement to acquire the point cloud data on the surface of the bearing bush. The acquisition of target three-dimensional point cloud data is to reconstruct dense point cloud by utilizing multiple groups of images, and the method is a method for reconstructing a 3D scene from multiple 2D images. The basic principle is that images of the same object or scene are captured through multiple viewing angles, and then the images are processed through computer vision and geometric algorithms to generate a dense 3D point cloud. When the data acquisition is carried out on the laser scanning equipment during the measurement of the position coordinates of the axle bush, a three-dimensional local coordinate system taking the central point of the signal transmitter as the origin of coordinates O is automatically constructed. In this coordinate system, the axis is defined by the axis of rotation of the instrument, with the X axis in the horizontal plane and the Z axis in the vertical plane, and the Y axis is defined according to the right hand rule. The three-dimensional coordinates of the object are obtained through a certain mathematical calculation by measuring elements collected by a scanner. These measurement elements mainly include the linear distance r from the center of the signal emitter to the object surface, the horizontal included angle phi and the vertical included angle ϴ,

；

And calculating through the formula to obtain the coordinates of the three-dimensional points on the surface of the object.

S2, preprocessing the point cloud data, including filtering denoising, outlier rejection and coordinate conversion operation, so as to improve accuracy and efficiency of subsequent processing.

Firstly, counting distribution intervals, namely three lengths, in different directions by acquiring a maximum value coordinate and a minimum value coordinate of bearing bush point cloud data on a coordinate axis; then, dividing the whole point cloud data into voxel grids with equal size and resolution ratioAnd calculating a point cloud density within each voxel grid; then, the point cloud with the density higher than the set threshold value in each voxel grid is downsampled, and only one coordinate point is reserved to represent the central point of the point cloud in the voxel grid. According to the requirement of simplified sampling, two sampling point selection methods exist in the voxels: one is a gravity center selection method, which calculates the gravity center point coordinates of points falling into a voxel grid, wherein the coordinates are the coordinates of voxel downsampling; the other is to find out the nearest point from the center coordinates according to the center coordinates of the voxels, and the point is the voxel downsampling point. Both methods can realize the simplified processing of the point cloud data, but the sampling effect and the processing speed of the point cloud data are different. The gravity center grid downsampling process is as follows:

a. And searching a coordinate extremum of the point cloud in a coordinate space. Obtaining 。

B. calculating the size of grid space, and the size of voxels isVoxel(s)The size of (a) affects the degree of compaction of the point cloud,The setting is too small, the point cloud data volume is larger, the quality simplifying effect is poor,If the setting is too large, the building point cloud characteristics are lost.

；

；

；

In the method, in the process of the invention,Representing the size of the voxels in the X-axis direction; Representing the size of the voxels in the Y-axis direction; the voxel size in the Z-axis direction is indicated, Is an upward rounding operation.

C. the point cloud is mapped to the voxel grid.

；

；

；

；

In the method, in the process of the invention,Representing the number of points in the voxels in the X-axis direction; Representing the number of points in the voxels in the Y-axis direction; the number of points in the voxels in the Z-axis direction is indicated.

Grid toThe arrangement is carried out in sequence and,For the purpose of the grid index,Is an upward rounding operation. And calculating the position of each point falling on the grid, and taking the center point of the grid as the sampled point.

D. Ordering according to the index in c, and sampling according to a gravity center method is expressed as:

；

Wherein, For the number of points in a grid voxel,Is the coordinates sampled by the gravity center method,To map to points in the grid.

Due to the noise problem which is inevitably present in the acquisition of the three-dimensional point cloud data. These noise points increase uncertainty of the point cloud data, affect the geometric structure and surface characteristics of the point cloud data, and thus reduce the information amount of the point cloud data, so that optimization processing of the point cloud quality is required. The algorithm principle is as follows:

a. At the original point Where n is the total number of building point clouds,For any point, searching for the current point using the kd-TreeK adjacent points of (x, y, z) are calculated by calculating the pointsDistance to k adjacent pointsFurther, the average value of the points is obtained;

；

；

B. the Gaussian distribution of the point density in the point set P is determined by the mean value (mu) and the standard deviation (sigma), and the points in P are calculatedΜ represents spatial distribution information of points in the original data, σ represents approximation of the mean and the overall mean;

；

；

c. by setting a proper distance threshold t, judging whether the point is an outlier according to the following formula:

；

wherein L is the maximum distance threshold, when the average distance Points below L remain and point programs above L will set the point as a noise point and remove the point.

And S3, aligning the point clouds of different parts by using ICP (Iterative Closest Point) algorithm, and aligning the bearing bush surface with the point cloud model. The method comprises the following specific steps:

step1: and constructing a matching point pair, searching the nearest points of all points in the source point cloud on the target point cloud by using a KD-Tree method, and constructing the matching point pair.

Step2: noise values are removed and outliers in the set of constructed matched point pairs are removed using a random sample consensus algorithm (random sample consensus, RANSAC).

Step3: and calculating an optimal rotation matrix R and a translation vector t by using SVD according to the constructed point pair set.

Step4: and performing rotation translation calculation on the source point cloud by using the rotation translation transformation matrix obtained in the last step to obtain a new point set after rigid transformation of the source point cloud.

Step5: and repeatedly calculating step 1-4 until the rotation translation result solved among 2 iterations converges or the maximum iteration number is reached.

S4, referring to FIG. 1, after the curved surface is rebuilt, utilizing a characteristic point extraction algorithm to obtain characteristic points of the surface of the bearing bush, and obtaining the characteristic points of the surface of the bearing bush, wherein the characteristic points are determined by indexes such as curvature, normal vector, curvature change and the like; wherein points with significant geometric features are selected as feature points.

S5, based on the point cloud data, feature points are accurately extracted, and then the geometric center of the bearing bush is calculated to determine a machining reference.

The characteristic point extraction algorithm is to extract a group of characteristic points from the three-dimensional bearing bush data. Coordinate information of the feature points, namely the positions of the feature points in the three-dimensional space, is obtained. The geometric centers of the feature points are calculated by a mathematical algorithm using the coordinate information of the points. The geometric center may be calculated by taking an average of the feature point coordinates. I.e. the X, Y, Z coordinates of the feature points are averaged respectively to obtain the coordinates of the geometric center.

Record a given source point cloudUsually, the point cloud is firstly extracted with a few key feature points, and the feature point extraction process is as follows:

；

Wherein the method comprises the steps of The representative feature point extractor is provided with a feature point extraction unit,For the extracted feature point set, there are。

The feature point extraction uses a method based on normal vector angles. The normal vector for each point in the point cloud model is first calculated. For normal vector calculation, firstly constructing KD-Tree for an original point cloud, then selecting each point in the point cloud, searching for adjacent points, fitting the adjacent points containing the point into a curved surface, and then calculating a covariance matrix of a field set to obtain a characteristic value and a corresponding characteristic vector of the matrix. The principal component analysis method shows that the feature vector corresponding to the minimum feature value is the normal vector of the fitting curved surface, specifically: assume sampling pointsIs set as (1)First, the center of gravity of the domain set is calculated according to the following formula:

；

Then, calculating covariance matrix of the domain set according to the following method ：

；

Determining the eigenvalue of the matrixFeature vectors corresponding to the feature vectorsThe feature vector corresponding to the minimum feature valueIs the sampling pointIs defined in the specification. And then calculating the included angle between the normal vector of each point and the normal vector of the points in the surrounding field, and extracting the integral feature points according to a certain threshold value.

Surface curvature is also an important geometric feature describing the degree of curvature of a curved surface. And obtaining the characteristic points on the curved surface by using the characteristic point extraction algorithm, and calculating the curvature information of the characteristic points by using a mathematical algorithm. And calculating the principal curvature and the average curvature of each characteristic point, thereby obtaining the local curvature distribution condition of the curved surface.

S6, determining a machining axis of the bearing bush through a curve fitting algorithm.

And acquiring the main axis direction of the bearing bush by using a curve fitting algorithm. And fitting the characteristic points to obtain the main axis direction vector of the bearing bush. The fitting method is least square method and principal component analysis.

The curve fitting operation can be performed by two-dimensional bearing bush data to acquire the main axis direction of the bearing bush. And acquiring two-dimensional data points of the inner surface and the outer surface of the bearing bush by measuring the actual models of the inner surface and the outer surface of the bearing bush, and fitting the discrete data points into a curve by adopting a least square method by combining a curve fitting algorithm, so as to obtain the main axis direction of the bearing bush. And extracting geometric features of the bearing bush by using a mathematical algorithm based on the curved surface model of curved surface reconstruction, characteristic points extracted by the characteristic points and a curve fitted curve. This includes calculation of geometric center, determination of principal axis direction, calculation of surface curvature:

；

a. The sum of the distances from the point to the curve, i.e. the sum of the squares of the deviations ：

；

B. to obtain compliance with the conditionValue, right side of the opposite equationPartial derivative is calculated:

；

；

；

c. The left equation is simplified:

；

；

；

d. Matrixing the equation:

；

and solving the coefficient A and simultaneously solving a two-dimensional fitting curve. The curve is the fitted bearing bush axis, and the parameters of the curve can provide data support for the subsequent algorithm.

S7, acquiring bearing bush positioning data in real time to determine machining reference and axis parameters, and automatically adjusting the position and parameters of machining equipment through compensation logic and a three-dimensional reconstruction algorithm to enable the machining equipment to be matched with the actual shape and position of the bearing bush.

Once the machining reference and axis parameters of the bearing shell are determined, the control system can automatically adjust the position and parameters of the machining equipment to match the actual shape and position of the bearing shell. And (3) carrying out three-dimensional reconstruction on the bearing bush machining profile by using a compensation method, so that the system automatically adjusts machining parameters according to the bearing bush surface characteristics and positions. The main implementation process comprises the following two parts of contents: firstly, reconstructing a machining profile based on NURBS reconstruction technology, and further calculating phase difference; and then, modifying the coordinates of the tool position points in the tool path file according to the difference value to complete compensation.

The NURBS curve is a vector parameter defined by nodesIs a sequence of (2)Determined byAnd a second order polynomial. The equation is defined as:

；

Wherein, In order to control the vertexes, the convex hull polygons formed by connecting the points in sequence are control polygons,Is a k-degree B-spline basis function.

Given a givenA control vertex, wherein the control vertex presents array distribution, and hasLine sumColumns, denoted: the control points distributed by the arrays form a control grid. Definition of the definition The direction of each point isThe direction, the number of curves in this direction is k. Definition of the definitionThe direction of each point isDirection, the curve number of the direction is. And gives two node vectorsThus defining a sheetThe tensor product of the times is non-uniform and rational B-spline surface. The equation can be expressed as:

；

after the node vector is determined, the control vertex is known again And the number of times k, the definition of a NURBS curve is completed. For any one parameter value in NURBS curve definition domainIts corresponding point on NURBS curveCan be obtained by calculation of the following formula.

Interpolation for curve inversionData pointsThe k-th NURBS curve equation for (c) can be written as:

；

Defining a curve into a domain The values of the nodes in the interpolation are substituted into the equation, and the interpolation condition should be satisfied, namely

；

The position and parameters of the processing equipment are automatically adjusted, so that the actual shape and position of the bearing bush are matched, and the specific implementation process is as follows:

① Based on the phase difference calculation of the reconstructed profile, the coordinates and the tool point measured in the actual measurement process are not completely coincident, so that the machined profile needs to be reconstructed in the phase difference value solving process. Firstly, carrying out inverse calculation on the NURBS curve surface according to the actual measurement points to obtain the definition of the curve surface. And then carrying out forward solving on the NURBS curve surface, parameterizing the coordinates of the cutter position points, setting the head and tail of the cutter position points to coincide with the head and tail of the real measurement points, and calculating the actual coordinate values of the cutter position points according to a curve equation and the parameterized cutter position points so as to obtain the difference value of each cutter position point.

NURBS surface forward: when any set of parameter values in the definition domain of a given to-be-solved surfaceTo request the corresponding point on the NURBS curved surfaceThe method can be carried out as follows.

Step1, calculating an intermediate polygon. By means ofParameter value pair edgeDirection of parametersControlling the polygon to execute a Deboolean recursive algorithm to obtainThe points being intermediate vertices, the intermediate vertices forming a path alongIntermediate polygons of parameter directions.

Step2, calculating points on the curved surface. By means ofThe parameter value executes the Debz recursive algorithm on the intermediate polygon, and the obtained point is the parameter on the NURBS curved surfaceCorresponding point。

NURBS surface inversion: according to the method for determining the node vector in the inverse solving process of the NURBS curve, the node vectors in two parameter directions can be determined by using a given curved surface type value pointAnd. After the node vector is determined, the calculation of the NURBS curved control vertex can be performed as follows:

step1, calculating an intermediate control vertex. Can be obtained along the u parameter direction of the curved surface Data points of groups each containingData points. For this purposeThe group data points are calculated according to the curve inverse method respectively, and can be obtainedGroup-intermediate control vertices, each group containingAnd intermediate control vertices.

Step2, calculating a control grid. Then calculating the intermediate control vertex obtained by step1 along the v parameter direction of the curved surface according to a curve inversion method to obtainGroup control vertices, each group containingAnd control vertices. Thus, the product can be obtainedAnd a control polygon mesh formed by the control vertexes is the control vertex mesh of the NURBS curved surface.

② And (3) based on the compensation of the tool path file, modifying coordinates of the tool positions in the tool path file according to the calculated difference value of each tool position, generating a tool path file with compensation, outputting the tool path file to CAD software for post-processing, and finally generating NC codes with compensation for compensation processing.

By combining the steps, the processing reference self-locating method can realize automatic locating of the processing reference, and improves the processing efficiency and the processing precision. The machining state of the bearing bush is monitored in real time, and the machining datum point is adjusted according to the monitoring result, so that automatic locating of the machining datum point is realized, and the machining efficiency and the machining precision are improved.

The present invention is not limited to the above embodiments, and various changes or modifications are included in the scope of the present invention.

By implementing the technical scheme of the invention, the automatic locating of the processing reference can be realized, the processing efficiency and the processing precision are improved, and the method has good practical value and economic value.

Claims (13)

1. The machining reference self-locating method for the material-adding repairing bearing bush is characterized by comprising the following steps of:

S1, a machine tool performs point cloud scanning on a bearing bush clamped on a machine tool workbench through vision measuring equipment to acquire point cloud data on the surface of the bearing bush;

s2, preprocessing point cloud data, including filtering denoising, outlier rejection and coordinate conversion operation;

s3, aligning the surface of the bearing bush with the point cloud model, and carrying out point cloud registration;

s4, after the curved surface is rebuilt, obtaining characteristic points of the surface of the bearing bush, wherein the characteristic points are determined through curvature, normal vector and curvature change indexes; wherein points with significant geometric features are selected as feature points;

S5, accurately extracting characteristic points based on the point cloud data, and further calculating the geometric center of the bearing bush to determine a machining reference;

s6, determining a machining axis of the bearing bush through a curve fitting algorithm;

S7, acquiring bearing bush positioning data in real time to determine machining reference and axis parameters, and automatically adjusting the position and parameters of machining equipment through compensation logic and a three-dimensional reconstruction algorithm to enable the machining equipment to be matched with the actual shape and position of the bearing bush.

2. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S1, carrying out point cloud scanning on a bearing bush clamped on a machine tool workbench through vision measurement equipment arranged on the machine tool, measuring the position coordinates of the bearing bush, automatically constructing a three-dimensional local coordinate system taking the center point of the vision measurement equipment as a coordinate origin O by the system, wherein in the coordinate system, an X axis is positioned on a horizontal plane, a Z axis is positioned on a vertical plane, the Y axis is determined according to a right-hand rule, the three-dimensional coordinates of the bearing bush are obtained through measurement elements acquired by the vision measurement equipment, the measurement elements comprise a straight line distance r from the center of the vision measurement equipment to the surface of an object, a horizontal included angle phi and a vertical included angle ϴ, calculating through the following formula to obtain the coordinates of the three-dimensional point on the surface of the object,

。

3. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S2, the step of preprocessing the point cloud data is as follows,

S21, counting distribution intervals in different directions, namely three lengths, by acquiring a maximum value coordinate and a minimum value coordinate of the bearing bush point cloud data on a coordinate axis;

s22, dividing the whole point cloud data into voxel grids with equal size, wherein the resolution is that And calculating a point cloud density within each voxel grid;

S23, downsampling the point cloud with the density higher than a set threshold value in each voxel grid, and only reserving one coordinate point to represent the center point of the point cloud in the voxel grid.

4. A method of locating a machining datum of an additive repair bearing shell according to claim 3, wherein: in S23, the step of downsampling the point cloud within the voxel grid is as follows,

S231, searching a coordinate extremum of the point cloud in a coordinate space to obtain；

S232, calculating the size of the grid space, wherein the size of the voxels isVoxel(s)The size of (a) affects the degree of compaction of the point cloud,The setting is too small, the point cloud data volume is larger, the quality simplifying effect is poor,If the setting is too large, the building point cloud characteristics are lost;

；

；

；

In the method, in the process of the invention, Representing the size of the voxels in the X-axis direction; Representing the size of the voxels in the Y-axis direction; the voxel size in the Z-axis direction is indicated, Is an upward rounding operation;

s233, mapping the point cloud and the voxel grid,

；

；

；

；

In the method, in the process of the invention,Representing the number of points in the voxels in the X-axis direction; Representing the number of points in the voxels in the Y-axis direction; Representing the number of points in the voxels in the Z-axis direction;

Grid to The arrangement is carried out in sequence and,For the grid index, calculating the position of each point falling on the grid, taking the center point of the grid as the sampled point,

S234, sorting according to the index in S233, and sampling according to the barycenter method is expressed as:

；

Wherein, For the number of points in a grid voxel,Is the coordinates sampled by the gravity center method,To map to points in the grid.

5. The method for machining reference self-locating of an additive repairing bearing bush according to claim 4, wherein the method comprises the following steps: the optimization processing steps of the point cloud quality are as follows:

a. At the original point Where n is the total number of building point clouds,For any point, searching for the current point using the kd-TreeK adjacent points of (x, y, z) are calculated by calculating the pointsDistance to k adjacent pointsFurther, the average value of the points is obtained;

；

；

B. the Gaussian distribution of the point density in the point set P is determined by the mean value mu and the standard deviation sigma, and the points in P are calculatedΜ represents spatial distribution information of points in the original data, σ represents approximation of the mean and the overall mean;

；

；

c. By setting a proper distance threshold t, judging whether the point is an outlier according to the following formula:

；

wherein L is the maximum distance threshold, when the average distance Points smaller than L remain, points larger than L set the point as a noise point, and the point is removed.

6. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S3, the step of aligning the bearing shell surface with the point cloud model is as follows,

S31, constructing matching point pairs; searching the nearest points of all points in the source point cloud on the target point cloud by using a KD-Tree method, and constructing matching point pairs;

s32, removing noise values; removing abnormal values in the constructed matching point pair set by using a random sampling consistency algorithm;

s33, calculating an optimal rotation matrix R and a translation vector t by using SVD according to the constructed point pair set;

S34, performing rotation translation calculation on the source point cloud by using the rotation translation transformation matrix obtained in the previous step to obtain a new point set after rigid transformation of the source point cloud;

And S35, repeatedly calculating S31-S34 until the rotation translation result solved among 2 iterations converges or the maximum iteration number is reached.

7. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S5, a characteristic point extraction algorithm is to extract a group of characteristic points from the three-dimensional bearing bush data, wherein the coordinate information of the characteristic points is the position of the characteristic points in a three-dimensional space; and calculating the geometric center by utilizing the coordinate information of the feature points and by using the average value of the feature point coordinates, namely respectively averaging X, Y, Z coordinates of the feature points to obtain the coordinates of the geometric center.

8. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S6, performing curve fitting operation on two-dimensional bearing bush data to obtain the main axis direction of the bearing bush, measuring actual models of the inner and outer surfaces of the bearing bush to obtain two-dimensional data points of the inner and outer surfaces of the bearing bush, and fitting discrete data points into a curve by adopting a least square method in combination with a curve fitting algorithm to obtain the main axis direction of the bearing bush, and extracting geometric features of the bearing bush based on a curved surface model reconstructed by the curved surface, feature points extracted by the feature points and the curve fitted curve; the step of extracting the geometrical features of the bearing shell is as follows,

The initial curve expression for curve fitting is:

；

a. The sum of the distances from the point to the curve, i.e. the sum of the squares of the deviations ：

；

B. to obtain compliance with the conditionValue, right side of the opposite equationPartial derivative is calculated:

；

；

9.；

c. Simplifying the left equation:

；

；

10.；

d. Matrixing the equation:

；

And solving the coefficient A, and solving a two-dimensional fitting curve, wherein the curve is the axis of the fitted bearing bush, and the parameters can provide data support for the subsequent algorithm.

11. The method for machining reference self-locating of an additive repairing bearing bush according to claim 1, wherein the method comprises the following steps: in S7, a compensation method is used for carrying out three-dimensional reconstruction on the bearing bush machining molded surface, so that a system automatically adjusts machining parameters according to bearing bush surface characteristics and positions, and the implementation process comprises the following two parts: firstly, reconstructing a machining profile based on NURBS reconstruction technology, and further calculating phase difference; and then, modifying the coordinates of the tool position points in the tool path file according to the difference value to complete compensation.

12. The method for machining reference self-locating of an additive repairing bearing bush according to claim 9, wherein the method comprises the following steps: in S7, the position and parameters of the processing equipment are automatically adjusted through the compensation logic and the three-dimensional reconstruction algorithm, so that the actual shape and position of the processing equipment and the bearing bush are matched, and the implementation process is as follows:

Firstly, carrying out inverse calculation on a NURBS curve curved surface according to a real measurement point to obtain a definition of the curved surface, then carrying out positive calculation on the NURBS curve curved surface, parameterizing the coordinates of tool positions, setting the head and tail of the tool positions to coincide with the head and tail of the real measurement point, calculating the actual coordinate values of the tool positions according to a curve equation and the parameterized tool positions, and further obtaining the difference value of each tool position; and (3) based on the compensation of the tool path file, modifying coordinates of the tool positions in the tool path file according to the calculated difference value of each tool position, generating a tool path file with compensation, outputting the tool path file to CAD software for post-processing, and finally generating NC codes with compensation for compensation processing.

13. Wherein, NURBS curved surface is just asking: when any set of parameter values in the definition domain of a given to-be-solved surfaceTo request the corresponding point on the NURBS curved surfaceThe method is carried out according to the following steps,

Step1, calculating a middle polygon; by means ofParameter value pair edgeDirection of parametersControlling the polygon to execute a Deboolean recursive algorithm to obtainThe points being intermediate vertices, the intermediate vertices forming a path alongMiddle polygon of parameter direction;

step2, calculating points on the curved surface, using The parameter value executes the Debz recursive algorithm on the intermediate polygon, and the obtained point is the parameter on the NURBS curved surfaceCorresponding point；

NURBS surface inversion: according to the method for determining the node vector in the inverse solving process of the NURBS curve, the node vectors in two parameter directions can be determined by using a given curved surface type value pointAnd; After the node vector is determined, the calculation of the NURBS curved control vertex can be performed as follows:

Step1, calculating an intermediate control vertex; can be obtained along the u parameter direction of the curved surface Data points of groups each containingData points for thisThe group data points are calculated according to the curve inverse method respectively, and can be obtainedGroup-intermediate control vertices, each group containingA plurality of intermediate control vertices;

step2, calculating a control grid; then calculating the intermediate control vertex obtained by step1 along the v parameter direction of the curved surface according to a curve inversion method to obtain Group control vertices, each group containingControl vertices such that a control vertex is obtainedAnd a control polygon mesh formed by the control vertexes is the control vertex mesh of the NURBS curved surface.