Positioning method of fan main shaft forging based on machine vision

Abstract

The present invention relates to the field of spindle forging positioning, and more specifically, the present invention relates to a fan spindle forging positioning method based on machine vision, the method comprising: obtaining three views and depth data of the spindle forging, identifying the forging area using semantic segmentation technology, obtaining the extension area of ​​the spindle forging according to the front view from the initial moment to the final moment of the hydraulic hammer pressing down, calculating the degree of imbalance of the extension area after each hydraulic hammer pressing down, obtaining an imbalance degree sequence, performing second-order difference on the imbalance degree sequence, and judging whether the hydraulic hammer stops running, adjusting the hydraulic position of the hydraulic hammer; performing binary classification based on the adjustment matrix obtained by the depth value difference, calculating the classification coefficient, guiding the axial and vertical axial movement of the hydraulic hammer, so as to achieve precise positioning and optimize the forging process. The present invention can timely discover deviations in the forging process and improve the overall quality and consistency of the forging by real-time monitoring of the extension area and the degree of imbalance of the forging.

Description

Technical Field

[0001] The present invention relates to the field of spindle forging positioning, and more specifically, to a fan spindle forging positioning method based on machine vision. Background Art

[0002] Wind turbine main shaft forgings refer to forged parts used in the main shaft of wind turbine generator sets. The main shaft is one of the key components in a wind turbine. It connects the wind turbine and the generator, and transmits the rotation of the turbine to the generator to generate electricity. Since wind turbines need to withstand huge rotational forces and periodically changing loads during operation, there are high requirements for the material, processing accuracy and strength of the main shaft forgings.

[0003] The existing Chinese patent application document with publication number CN116091728A discloses a method for locating and eliminating forging defects, which relates to the field of forging process technology, including: S1, obtaining a three-dimensional model of a pre-forging, performing forging simulation on the three-dimensional model of the pre-forging and obtaining defect areas and defect types on the forging after the pre-forging is simulated and formed, wherein the defect types include metal accumulation and metal folding; S2, performing reverse tracking based on the defect area using a finite element analysis method to obtain an area to be optimized of the pre-forging; S3, adjusting the shape and size of the pre-forging according to the defect type and the position of the area to be optimized to obtain an optimized pre-forging.

[0004] The application document obtains the position of the pre-forged part corresponding to the defective area of ​​the forging through finite element analysis. At present, during the transmission process of forging production, due to the large weight and volume of the fan main shaft forging, the robot arm may not be able to accurately place the forging under long-term operation, resulting in deviation in the position of the forging on the forging table, or due to the uneven distribution of the internal metal of the forging itself, which may be caused by defects in the material itself or insufficient pre-processing. During the hydraulic hammer hydraulic pressure, due to the uneven distribution of metal, the ductility of the forging is different in different areas, resulting in deviation in forging, and the hydraulic hammer needs to be repositioned. Summary of the invention

[0005] In order to solve the problem that the ductility of the forging is different in different areas due to uneven metal distribution during hydraulic pressure of the hydraulic hammer, resulting in forging deviation, the present invention provides solutions in the following aspects.

[0006] A method for positioning a fan main shaft forging based on machine vision includes: obtaining three views and depth data of the main shaft forging and performing semantic segmentation to obtain the main shaft forging area; obtaining the extension area of ​​the main shaft forging according to the front view from the initial moment to the final moment of the hydraulic hammer pressing down, calculating the imbalance degree of the extension area after each hydraulic hammer pressing down, obtaining an imbalance degree sequence, performing second-order difference on the imbalance degree sequence to obtain a second-order difference sequence, and judging whether the hydraulic hammer stops running, and adjusting the hydraulic position of the hydraulic hammer; performing a difference based on the depth values ​​of the main shaft forging area in the left view and the right view of the main shaft forging after stopping running, obtaining an adjustment matrix and performing binary classification, calculating the classification coefficient after classification, and adjusting the movement of the hydraulic hammer according to the classification coefficient Direction, to complete the positioning of the next spindle forging forging; wherein, the moving direction includes: the direction of axial movement is based on the binary classification, from the maximum value position point of the class with a larger depth data superposition value to the minimum value position point of the class with a smaller depth data superposition value, and the ratio of the position points between the minimum value and the maximum value multiplied by the horizontal distance between the two position points is used as the axial movement distance; the direction of vertical axial movement is: in response to the adjustment matrix, if the cumulative sum of the depth data is greater than 0, it moves from right to left in the vertical axis, if it is less than 0, it moves in the opposite direction, and if it is equal to 0, it does not move, and the ratio of the depth values ​​of the spindle forging area in the left view and the right view is multiplied by the length of the top edge of the spindle forging area in the front view at the initial moment as the vertical axial movement distance.

[0007] Through semantic segmentation and deep data analysis, the system can accurately identify and locate the spindle forging area, adjust the operating status of the hydraulic hammer by calculating the imbalance degree sequence of the extended area and performing second-order differences, automatically adjust the hydraulic position of the hydraulic hammer, and promptly detect deviations in the forging process, thereby improving the quality and consistency of the spindle forgings, reducing adjustment time, and improving production efficiency.

[0008] Preferably, the semantic segmentation construction process is:

[0009] Marking the forging area of ​​the three views of the historical spindle forgings, and putting multiple three views carrying the marks into the semantic segmentation network for training, and completing the forging extraction network in response to the cross entropy loss function being less than a preset loss value or reaching a preset number of training times;

[0010] The real-time three-view image is put into the forging extraction network to obtain the spindle forging area marked with 0-1, where the pixel value of the pixel point in the spindle forging area is 1, and the pixel value of the non-spindle forging area is 0.

[0011] Preferably, the degree of imbalance includes:

[0012] The three views of the spindle forging are obtained in real time, and the pixel points at the intersection position of the spindle forging area in the front view obtained in real time and the spindle forging area in the front view at the initial moment are removed to obtain the lateral extension area of ​​the spindle forging; the vertical line of the geometric center of the spindle forging area in the initial frame image is used as the central axis, the lateral edges of the extension area are extracted, and the imbalance degree of the edges on both sides is calculated;

[0013] The ratio of the horizontal distances from the pixel points on the extended edges on both sides to the central axis is calculated respectively to obtain the horizontal distance ratio; the ratio of the depth data of the pixel points on the extended edges on both sides is calculated respectively to obtain the depth data ratio; and the absolute value of the complement of the product between the horizontal distance ratio and the depth data ratio is summed as the degree of imbalance.

[0014] By acquiring three views in real time and eliminating the intersection pixels, the lateral extension area of ​​the spindle forging can be accurately monitored. By calculating the horizontal distance ratio and the depth data ratio and summing the absolute value of their product complement as the degree of imbalance, the ductility difference on both sides of the forging can be quantified. By real-time monitoring and adjustment of the degree of imbalance, the production process can be optimized and production delays caused by repositioning and adjustment can be reduced.

[0015] Preferably, the imbalance degree further includes:

[0016] Calculate the ratio of the horizontal distances between the pixel points on the extended edges on both sides and the central axis to obtain the horizontal distance ratio; calculate the ratio of the depth data of the pixel points on the extended edges on both sides to obtain the depth data ratio; calculate the average value between the horizontal distance ratio and the depth data ratio;

[0017] The variance of the sum of squares of the differences between the product of the horizontal distance ratio and the depth data ratio and the average value is calculated, and the square root of the variance is taken as the degree of imbalance.

[0018] Preferably, the determining whether the hydraulic hammer has stopped running and adjusting the hydraulic position of the hydraulic hammer includes:

[0019] In response to three consecutive positive values ​​appearing in the second-order difference sequence, the spindle forging is subjected to pressure, resulting in unbalanced extrusion on both sides, so that the extrusion amounts on both sides of the spindle forging are inconsistent, the hydraulic hammer is stopped, and the hydraulic position of the hydraulic hammer is adjusted.

[0020] By monitoring the second-order difference sequence, the unbalanced state of the spindle forging during the forging process can be responded to in time, and the operation of the hydraulic hammer can be stopped in time, which can reduce the damage of the forging caused by excessive extrusion and protect the integrity and structural strength of the forging.

[0021] Preferably, the step of obtaining the adjustment matrix and performing binary classification further includes:

[0022] The adjustment matrix is ​​classified into two groups using the Otsu threshold method to obtain two groups of classification results, wherein the group with a larger cumulative value in the two groups of classification results is the squeezed point, and conversely, the group with a smaller cumulative value is the non-squeezed point;

[0023] The un-squeezed points in the adjustment matrix are deleted to obtain the significant adjustment matrix, and the classification coefficient is calculated based on the significant adjustment matrix.

[0024] By identifying the extrusion points, the spindle forgings can be adjusted more specifically to achieve a more uniform forging effect. By optimizing the adjustments, the overall quality of the forgings can be improved and defects caused by uneven forging can be reduced. The forging process can be effectively optimized, the quality of forgings can be improved, material waste can be reduced, production efficiency and safety can be improved, and costs can be reduced.

[0025] Preferably, the classification coefficient includes:

[0026] ;

[0027] In the formula, represents the classification coefficient, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

[0028] Preferably, the classification coefficient further includes:

[0029] ;

[0030] In the formula, represents the classification coefficient, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

[0031] Preferably, adjusting the moving direction of the hydraulic hammer according to the classification coefficient includes:

[0032] The moving direction includes axial movement and vertical axial movement. The axial movement is the forward and backward movement of the hydraulic hammer in the axial direction of the spindle forging, and the vertical axial movement is the left and right movement of the hydraulic hammer in the axial direction perpendicular to the spindle forging.

[0033] Through axial and vertical axial movement, the hydraulic hammer can be accurately positioned in three-dimensional space, ensuring uniform distribution of forging force, improving the processing accuracy of forgings, and then improving the accuracy and flexibility of the forging process, thereby improving the quality of forgings, reducing material waste, improving production efficiency and safety, and reducing costs.

[0034] The present invention has the following effects:

[0035] 1. The present invention can timely discover the deviation in the forging process by real-time monitoring of the extension area and the degree of imbalance of the forging, thereby adjusting the hydraulic position of the hydraulic hammer and the moving direction of the hydraulic hammer, which is beneficial to reducing the difference in ductility of the forging caused by uneven metal distribution, thereby improving the overall quality and consistency of the forging.

[0036] 2. The present invention uses machine vision technology to automatically complete the positioning and adjustment of forgings, reducing the need for manual intervention. The automated positioning method can reduce downtime during the forging process, improve production efficiency, and help improve the quality of forgings, reduce scrap rates, and thus reduce production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0038] Figure 1 It is a method flow chart of steps S1 to S3 in the method for positioning a fan main shaft forging based on machine vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0040] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0041] Reference Figure 1The fan main shaft forging positioning method based on machine vision includes steps S1 to S3, which are as follows:

[0042] Further analysis shows that when the spindle forging is placed on the forging table by the robot arm, due to the density and properties of the spindle forging material, the amount of extension of the spindle forging material to both sides is inconsistent when the hydraulic pressure is pressed down. Therefore, it is necessary to monitor the extension balance of the spindle material. The specific steps are as follows:

[0043] S1: Obtain the three-view image and depth data of the spindle forging and perform semantic segmentation to obtain the spindle forging area.

[0044] It should be noted that cameras are arranged in the axial direction and horizontal positions on both sides of the spindle forging to obtain the front view, left view and right view of the spindle forging, among which the axial view is the front view, and the lateral view is the left view and the right view. The above-mentioned camera is a depth camera.

[0045] The semantic segmentation construction process is:

[0046] Marking the forging area of ​​the three views of the historical spindle forgings, and putting multiple three views carrying the marks into the semantic segmentation network for training, and completing the forging extraction network in response to the cross entropy loss function being less than a preset loss value or reaching a preset number of training times;

[0047] The real-time three-view image is put into the forging extraction network to obtain the spindle forging area marked with 0-1, where the pixel value of the pixel point in the spindle forging area is 1, and the pixel value of the non-spindle forging area is 0.

[0048] In this embodiment, semantic segmentation is a technology well known to those skilled in the art and will not be described in detail.

[0049] S2: According to the front view from the initial moment to the final moment of the hydraulic hammer pressing down, the extension area of ​​the spindle forging is obtained, the imbalance degree of the extension area after each hydraulic hammer pressing down is calculated, and the imbalance degree sequence is obtained. The second-order difference of the imbalance degree sequence is performed to obtain the second-order difference sequence, and it is judged whether the hydraulic hammer stops running, and the hydraulic position of the hydraulic hammer is adjusted.

[0050] The specific steps to obtain the degree of imbalance are:

[0051] The three views of the spindle forging are obtained in real time, and the pixel points at the intersection position of the spindle forging area in the front view obtained in real time and the spindle forging area in the front view at the initial moment are removed to obtain the lateral extension area of ​​the spindle forging; the vertical line of the geometric center of the spindle forging area in the initial frame image is used as the central axis, the lateral edges of the extension area are extracted, and the imbalance degree of the edges on both sides is calculated;

[0052] The ratio of the horizontal distances from the pixel points on the extended edges on both sides to the central axis is calculated to obtain the horizontal distance ratio; the ratio of the depth data of the pixel points on the extended edges on both sides is calculated to obtain the depth data ratio; the absolute value of the complement of the product between the horizontal distance ratio and the depth data ratio is summed as the degree of imbalance.

[0053] Specifically, the degree of imbalance satisfies the following relationship:

[0054] ;

[0055] In the formula, Indicates The degree of imbalance on both sides of the spindle forging, Indicates the number of pixels on the vertical axis. Represents the index of the vertical pixel point, The vertical axis on one side edge is The horizontal distance from the pixel to the central axis, Indicates that the longitudinal axis on the other lateral edge is The horizontal distance from the pixel to the central axis, The vertical axis on one side edge is The depth data of the pixel points, Indicates that the longitudinal axis on the other lateral edge is The depth data of the pixel.

[0056] That is to say, Indicates the relative distance between the two side edges in the horizontal direction. If the ratio is greater than 1, it means that one side is farther away from the central axis than the other side. If it is less than 1, the opposite is true. Indicates the relative depth of the two side edges in the vertical direction. If the ratio is greater than 1, it means one side is deeper than the other side, and if it is less than 1, it is the opposite. The closer it is to 0, the more balanced the two sides are.

[0057] In addition, another embodiment further includes:

[0058] Calculate the ratio of the horizontal distances between the pixel points on the extended edges on both sides and the central axis to obtain the horizontal distance ratio; calculate the ratio of the depth data of the pixel points on the extended edges on both sides to obtain the depth data ratio; calculate the average value between the horizontal distance ratio and the depth data ratio;

[0059] The variance of the sum of the squares of the differences between the product of the horizontal distance ratio and the depth data ratio and the average value is calculated, and the square root of the variance is taken as the degree of imbalance.

[0060] Specifically, the degree of imbalance satisfies the following relationship:

[0061] ;

[0062] In the formula, Indicates The degree of imbalance on both sides of the spindle forging, Indicates the number of pixels on the vertical axis. Represents the index of the vertical pixel point, The vertical axis on one side edge is The horizontal distance from the pixel to the central axis, Indicates that the longitudinal axis on the other lateral edge is The horizontal distance from the pixel to the central axis, The vertical axis on one side edge is The depth data of the pixel points, Indicates that the longitudinal axis on the other lateral edge is The depth data of the pixel points, Indicates that the vertical axes on both sides are The average value of the horizontal distance ratio and the depth data ratio.

[0063] In response to three consecutive positive values ​​appearing in the second-order difference sequence, the spindle forging is subjected to pressure, resulting in unbalanced extrusion on both sides, so that the extrusion amounts on both sides of the spindle forging are inconsistent, the hydraulic hammer is stopped, and the hydraulic position of the hydraulic hammer is adjusted.

[0064] It should be noted that when an aggravating trend occurs, it means that the position of the spindle forging is misaligned. The downward pressure should be stopped immediately to determine the position of the hydraulic hammer for the next adjustment to avoid excessive extrusion of the spindle forging, which may cause damage to the forging and destroy the strength of the spindle forging.

[0065] S3: The depth values ​​of the spindle forging area in the left view and the right view of the spindle forging after stopping are subtracted to obtain an adjustment matrix and perform binary classification, calculate the classification coefficient after classification, adjust the moving direction of the hydraulic hammer according to the classification coefficient, and complete the positioning of the next spindle forging.

[0066] In this embodiment, in addition to directly performing binary classification according to the adjustment matrix and calculating the classification coefficient after classification, in the second embodiment, the adjustment matrix can also be pre-processed after binary classification. The specific steps are as follows:

[0067] The adjustment matrix is ​​classified into two groups using the Otsu threshold method to obtain two groups of classification results. Among them, the group with a larger cumulative value is the squeezed point, and vice versa, the group with a smaller cumulative value is the non-squeezed point.

[0068] The un-squeezed points in the adjustment matrix are deleted to obtain the significant adjustment matrix, and the classification coefficient is calculated based on the significant adjustment matrix.

[0069] That is to say, when the second embodiment is adopted, the significant adjustment matrix is ​​obtained by deleting the un-extruded points, wherein the significant adjustment matrix only contains the extrusion point data that is useful for analysis. This can reduce noise and interference in subsequent analysis and improve the efficiency of data processing and the reliability of results.

[0070] In the first embodiment, the method for calculating the classification coefficient is as follows:

[0071] Specifically, the classification coefficient satisfies the following relationship:

[0072] ;

[0073] In the formula, represents the classification coefficient, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

[0074] That is to say, calculating the mean difference of depth data in different columns in the adjustment matrix helps to identify key change points, optimize the adjustment process, and improve the overall quality and efficiency of forging processing.

[0075] In addition, in the second embodiment, the method for calculating the classification coefficient is as follows:

[0076] ;

[0077] In the formula, represents the classification coefficient, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

[0078] That is to say, Indicates The mean of the column depth data is The ratio of the mean values ​​of the column depth data reflects the numerical variation trend of the two column depth data. By identifying the columns with significant differences and stability, the position of the hydraulic hammer can be adjusted more accurately to correct the unbalanced state of the forging.

[0079] Adjust the direction of hydraulic hammer movement according to the classification coefficient, including:

[0080] The moving directions include axial movement and vertical axial movement. The axial movement is the forward and backward movement of the hydraulic hammer in the axial direction of the spindle forging, and the vertical axial movement is the left and right movement of the hydraulic hammer in the axial direction perpendicular to the spindle forging.

[0081] Among them, the moving directions include: the direction of axial movement is to move from the maximum value position point of the class with a larger depth data superposition value to the minimum value position point of the class with a smaller depth data superposition value based on binary classification, and the ratio of the position points between the minimum value and the maximum value multiplied by the horizontal distance between the two position points is used as the axial movement distance; the direction of vertical axial movement is: in response to the adjustment of the depth data cumulative sum of the matrix is ​​greater than 0, then move from right to left in the vertical axis, if it is less than 0, then move in the opposite direction, and if it is equal to 0, do not move, and the ratio of the depth values ​​of the spindle forging area in the left view and the right view multiplied by the top edge length of the spindle forging area in the front view at the initial moment is used as the vertical axial movement distance.

[0082] That is to say, the direction of axial movement is from the maximum value position point of the category with larger depth data superposition value to the minimum value position point of the category with smaller depth data superposition value. The selection of this direction is based on the distribution of depth data, and the purpose is to adjust the position of the forging so that the forging force is more evenly distributed; while the vertical axial movement is based on the overall distribution of the depth data of the adjustment matrix, and the purpose is to adjust the position of the hydraulic hammer according to the overall imbalance state of the forging. Through precise axial and vertical axial movement, it can be ensured that the forging is subjected to more uniform force during the forging process, thereby improving the overall quality of the forging.

[0083] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0084] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A fan main shaft forging positioning method based on machine vision, characterized in that: include: Obtain the three-view and depth data of the spindle forging and perform semantic segmentation to obtain the spindle forging area; According to the front view from the initial moment to the final moment of the hydraulic hammer pressing down, the extension area of ​​the spindle forging is obtained, the imbalance degree of the extension area after each hydraulic hammer pressing down is calculated, and the imbalance degree sequence is obtained. The imbalance degree sequence is subjected to second-order difference to obtain a second-order difference sequence. In response to three consecutive positive values ​​in the second-order difference sequence, the spindle forging is subjected to pressure, resulting in unbalanced extrusion on both sides, so that the extrusion amount on both sides of the spindle forging is inconsistent, the hydraulic hammer is stopped, and the hydraulic position of the hydraulic hammer is adjusted; The depth values ​​of the spindle forging area in the left view and the right view of the spindle forging after stopping operation are subtracted to obtain an adjustment matrix and perform binary classification, calculate the classification coefficient after classification, adjust the moving direction of the hydraulic hammer according to the classification coefficient, and complete the positioning of the next spindle forging; Among them, the moving directions include: the direction of axial movement is to move from the maximum value position point of the class with a larger depth data superposition value to the minimum value position point of the class with a smaller depth data superposition value based on binary classification, and the ratio of the position points between the minimum value and the maximum value multiplied by the horizontal distance between the two position points is used as the axial movement distance; the direction of vertical axial movement is: in response to the adjustment of the depth data cumulative sum of the matrix is ​​greater than 0, then move from right to left in the vertical axis, if it is less than 0, then move in the opposite direction, and if it is equal to 0, do not move, and the ratio of the depth values ​​of the spindle forging area in the left view and the right view multiplied by the top edge length of the spindle forging area in the front view at the initial moment is used as the vertical axial movement distance.

2. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The semantic segmentation construction process is as follows: The forging area of ​​the three views of the historical spindle forging is marked, and multiple three views carrying the marks are put into the semantic segmentation network for training, and in response to the cross entropy loss function being less than a preset loss value or reaching a preset number of training times, the forging extraction network is completed; The real-time three-view image is put into the forging extraction network to obtain the spindle forging area marked with 0-1, where the pixel value of the pixel point in the spindle forging area is 1, and the pixel value of the non-spindle forging area is 0.

3. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The degree of imbalance includes: The three views of the spindle forging are obtained in real time, and the pixel points at the intersection position of the spindle forging area in the front view obtained in real time and the spindle forging area in the front view at the initial moment are removed to obtain the lateral extension area of ​​the spindle forging; the vertical line of the geometric center of the spindle forging area in the initial frame image is used as the central axis, the lateral edges of the extension area are extracted, and the imbalance degree of the edges on both sides is calculated; The ratio of the horizontal distances from the pixel points on the extended edges on both sides to the central axis is calculated respectively to obtain the horizontal distance ratio; the ratio of the depth data of the pixel points on the extended edges on both sides is calculated respectively to obtain the depth data ratio; and the absolute value of the complement of the product between the horizontal distance ratio and the depth data ratio is summed as the degree of imbalance.

4. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The degree of imbalance also includes: Calculate the ratio of the horizontal distances between the pixel points on the extended edges on both sides and the central axis to obtain the horizontal distance ratio; calculate the ratio of the depth data of the pixel points on the extended edges on both sides to obtain the depth data ratio; calculate the average value between the horizontal distance ratio and the depth data ratio; The variance of the sum of squares of the differences between the product of the horizontal distance ratio and the depth data ratio and the average value is calculated, and the square root of the variance is taken as the degree of imbalance.

5. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The obtaining of the adjustment matrix and performing binary classification also includes: The adjustment matrix is ​​classified into two groups using the Otsu threshold method to obtain two groups of classification results, wherein the group with a larger cumulative value in the two groups of classification results is the squeezed point, and conversely, the group with a smaller cumulative value is the non-squeezed point; The un-squeezed points in the adjustment matrix are deleted to obtain the significant adjustment matrix, and the classification coefficient is calculated based on the significant adjustment matrix.

6. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The classification coefficients include: ； In the formula, represents the classification coefficient, Represents the adjustment matrix Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

7. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The classification coefficient also includes: ； In the formula, represents the classification coefficient, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, Represents the first Column depth data, represents the mean function.

8. The method for positioning a fan main shaft forging based on machine vision according to claim 1, characterized in that: The hydraulic hammer movement direction is adjusted according to the classification coefficient. include: The moving direction includes axial movement and vertical axial movement. The axial movement is the forward and backward movement of the hydraulic hammer in the axial direction of the spindle forging, and the vertical axial movement is the left and right movement of the hydraulic hammer in the axial direction perpendicular to the spindle forging.

Images