Deep hole inner surface defect detection method based on endoscopic multiplication imaging

Abstract

The invention belongs to the technical field of geometric quantity precision measurement of optical measurement and machine vision, and discloses a deep hole inner surface defect detection method based on endoscopic multiplication imaging. Firstly, deep hole inner wall defects are analyzed to cause change of an endoscopic imaging result. Secondly, a defect characterization model of defect height and angle change on emergent light offset is constructed by analyzing the path change of a single light in combination with a geometrical optical principle and space coordinate system transformation; then, Trace Pro optical simulation software is adopted to simulate height defects and angle defects to verify the accuracy of the model; and finally, through multi-section point cloud splicing, realizing axial three-dimensional reconstruction of different defects on the inner wall of the deep hole, representing an emergent light deformation result caused by different defects, and carrying out reverse analysis according to a light offset result to obtain the inner surface defects of the deep hole.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of geometric precision measurement of optical measurement and machine vision, and relates to a deep hole inner surface defect detection method based on endoscopic multiplication imaging. BACKGROUND

[0002] In the field of national major projects such as nuclear power, energy power, aerospace, the comprehensive manufacturing capability of core equipment has become an important symbol to measure the industrial foundation and scientific and technological level of a country. Among them, deep hole parts are widely used as key structural parts, and the machining quality is directly related to the safety and stability of the equipment. Deep hole machining is a core technology in high-end manufacturing, which has high technical content, great manufacturing difficulty, strict reliability requirements, and usually involves multidisciplinary intersection and complex working environment, and has become an important technical support indispensable in the field of equipment manufacturing. Especially in typical small-bore deep hole parts such as nuclear power steam generator tube plate hole and internal combustion engine crankshaft oil hole, they often operate in extreme working environments such as high temperature, high pressure and strong impact load during service. Once the problems such as hole diameter error, roundness deviation or insufficient straightness occur during machining, it is easy to cause the performance of the part to decline, and then affect the safety and reliability of the whole machine operation. In severe cases, it may even cause the failure of key components or the scrapping of the whole machine, causing significant engineering risks and economic losses.

[0003] Taking the tube plate of AP1000 type nuclear power steam generator as an example, the machining of the tube plate is crucial in the manufacturing process of the nuclear power steam generator. There are 20050 holes on the tube plate, the hole diameter is 17.73 mm, the hole depth reaches 800 mm, the depth-diameter ratio is as high as 45:1, and the size tolerance, form and position tolerance and surface quality of each hole are extremely strict, becoming one of the most difficult deep holes to machine in the world. In the machining process of the steam generator tube plate, due to the uncontrollability of tool wear and random damage, the precision and consistency of deep hole machining are often difficult to guarantee. Therefore, developing an efficient deep hole inner surface defect measurement technology is crucial for improving deep hole machining process and optimizing machining quality. Although the current measurement technology of external size and plane features can achieve nanometer-level precision, there are still significant deficiencies in the measurement of deep hole internal geometric parameters. Especially when facing deep small hole structures with hole diameter less than 20 mm and depth-diameter ratio more than 40, the measurement capability is still insufficient.

[0004] Therefore, developing a detection technology for the inner surface of deep-hole parts, and conducting systematic detection and analysis of inner surface defects in deep-hole parts with an inner diameter of 16 mm, is of great significance for product quality control and fault diagnosis. This also makes the measurement of surface defect characteristics of deep-hole inner diameter a critical problem that urgently needs to be solved. This invention is based on an experimental setup already established in a method for precise measurement of the geometric characteristics of the inner surface of small holes with high aspect ratios, and further studies the detection method for inner surface defects of deep-hole parts. Summary of the Invention

[0005] Based on the aforementioned problems, this invention proposes a method for detecting defects on the inner surface of deep holes based on endoscopic magnification imaging.

[0006] The technical solution of this invention:

[0007] A method for detecting defects on the inner surface of deep holes based on endoscopic magnification imaging, comprising the following steps:

[0008] Step 1: Analysis of the mechanism by which internal surface defects affect imaging results

[0009] When defects exist on the inner wall of a deep hole, the light path is deformed during propagation due to the defects, resulting in a distortion of the geometry of the annular light spot in the imaging result. By analyzing this distortion characteristic, the relationship between light propagation and defect characteristics is established, providing basic data for subsequent defect detection and 3D reconstruction.

[0010] Step 2: Construction of Defect Height Characterization Model

[0011] A single ray from the ring beam is selected, parallel to the axis of the deep hole, and enters the deep hole for reflection. Based on the reflection path, a three-dimensional XYZ coordinate system is established: the origin is the intersection of the axis of the deep hole and the right end face of the magnifying lens group; the X-axis is along the radial direction of the deep hole, pointing to the outer side of the inner wall of the deep hole, representing the direction of the defect height, i.e., the vertical direction of the convex or concave part of the inner wall of the deep hole; the Y-axis is along the radial direction of the deep hole and perpendicular to the X-axis, representing the direction of the distance of the outgoing ray from the axis within the cross-section of the deep hole, i.e., the radial offset direction of the light spot on the receiving screen; the Z-axis is along the axial direction of the deep hole, representing the axial position of the deep hole, i.e., the depth position of the defect within the deep hole.

[0012] The following feature points are defined: M point is the intersection point of the first cone surface of the magnification lens group and the reflected light of the deep hole inner wall; L point is the intersection point of the incident light and the first cone surface of the magnification lens group; P point is the actual reflection point of the defect area of the deep hole inner wall; S point and U point are light propagation auxiliary points on the second cone surface of the magnification lens group; T point, K point and Q point are auxiliary points for light propagation, which play an auxiliary role in the spatial propagation of light and the geometric relationship between the deep hole inner wall and the light, and are used to construct a clearer optical detection model; PT represents the incident path offset; PK and PL are auxiliary lines for geometric derivation; MQ represents the lateral offset; NS is an auxiliary line for deriving the mathematical correlation between the defect characteristics and the exit light offset; I point is a point on the second cone surface of the magnification lens group; F point is a point on the deep hole axis, which is used to define the position of the deep hole axis; IF represents the offset correction amount, and MF is a line perpendicular to the deep hole axis, which is used to derive the mathematical correlation between the defect characteristics and the exit light offset, and when the height of the inner surface defect changes to 0, MF is equal to the incident aperture R; IP is the hypotenuse of△IPT, which is used for geometric derivation; N point is the intersection point of the light reflected by the actual reflection point P and the second cone surface of the magnification lens group; Z point is the intersection point of the incident light and the reflected light of the deep hole inner wall; H point and E point are points for auxiliary geometric analysis; HE is an auxiliary line for deriving the mathematical correlation between the defect characteristics and the exit light offset; NZ and ZP are two sides of△NZP, ZE and NP are the height and bottom of△NZP respectively, which are used for geometric derivation; NE is a right angle side of△NZE, which is used for geometric derivation; ZL is an auxiliary line for geometric derivation; C point is a fixed reference point of the magnification lens group; A point is the intersection point of the extension line of the cylindrical surface generatrix of the magnification lens group and the perpendicular line of the incident light, which is a point for auxiliary geometric analysis; CA is a right angle side of△CAL, which is used for geometric derivation;

[0013] Based on the principle of geometric optics and the transformation of spatial coordinate system, combined with the constraint conditions of system parameters, a representation model of the height change of the inner surface defect of the deep hole ΔX and the distance Y of the exit light to the axis is constructed; wherein the incident aperture radius R, the light incident radius r, the lens group spacing d, the offset correction amount IF, the reference length HC, the critical margin ΔC, θ1 and θ2 are known;

[0014] The incident path offset is determined by the light incident radius and geometric deformation, and is as follows:

[0015]

[0016] The right-angled triangle△NZP is orthogonally decomposed, and the foot E satisfies:

[0017]

[0018] The length of NP segment is obtained from the geometric symmetry:

[0019]

[0020] Set the motion point P along the ZL axis, its vertical distance to the reference line HC is:

[0021]

[0022] The geometric chain relationship of PL segment is simplified as:

[0023]

[0024] The ZL axis length and the lateral offset MQ satisfy:

[0025]

[0026] The vertical distance of point N to point F and the vertical distance of point M to the light incident radius are expressed as:

[0027]

[0028] According to the geometric derivation, the distance Y of the outgoing light to the axis is expressed as the vertical distance of point N to point F, that is, Y = NS + IF; the inner surface defect height variation ΔX is equal to the difference between the incident aperture R and the vertical distance of point M to the light incident radius, that is, ΔX = R - MF;

[0029] According to the above equations, the relationship between the inner surface defect height variation ΔX and the outgoing light to the axis distance Y is obtained by eliminating the intermediate variables:

[0030]

[0031] Step 3: Defect angle representation model construction

[0032] Suppose there is an angle deviation θ of a point on the inner wall of the deep hole, analyze the offset of the reflection path when the incident light beam reaches the position, and combine the reflection path to establish a coordinate system here which is shared with step two; the same letter feature points also have the same meaning; in addition, points B and D are points on the outgoing light plane, which are used to assist in describing the light; BD is an auxiliary line, which is used to derive the outgoing radius; IN, DN, DI and IS are auxiliary lines on the second cone surface of the multiplier mirror group, which are related to the outgoing light and are used for geometric derivation; ZP, ZQ and ZL are located on the incident light, which are used for geometric derivation;

[0033] Based on the geometric optics principle and the space coordinate system transformation, combined with the system parameter constraint condition, a representation model of the defect angle variation θ and the outgoing light to the axis distance Y is constructed; wherein DI and MQ are known.

[0034] The incident light offset is determined by the receiving light screen distance and the inner diameter angle deviation, and its expression is:

[0035]

[0036] where NP segment length is obtained by orthogonal decomposition of isosceles triangle △NZP:

[0037]

[0038] The exit radius is decomposed into three parts: exit radius = BD + NS + IF;

[0039] BD segment: obtained by geometric projection relationship:

[0040]

[0041] NS segment: obtained by orthogonal decomposition:

[0042]

[0043] The vertical distance PL of moving point P along ZL axis satisfies:

[0044]

[0045] The length of ZL axis is determined by the geometric constraint of MQ and the change of defect angle θ:

[0046]

[0047] Combining the displacement relationship of moving point:

[0048]

[0049] Based on ∠ZMQ = 64° - 2θ and ∠MZQ = 26° + 2θ, the side-angle relationship of triangle △MZQ is constructed by using the sine theorem:

[0050]

[0051] Combining the expression of ZL, the implicit equation of MQ is obtained:

[0052]

[0053] Solving the above equations together and eliminating intermediate variables, the explicit relationship between dynamic angle θ and exit radius Y is obtained:

[0054]

[0055] Step 4: Single-section optical simulation verification

[0056] The deep hole scene containing virtual defects is constructed by using optical simulation software, different defect height and angle parameters are set, and the light propagation path and imaging process are simulated. The height defects and angle defects are simulated respectively, the corresponding parameters are adjusted, the measurement range is obtained, and the accuracy of the defect characterization model is verified by comparing the simulation data with the calculation results of the theoretical model.

[0057] Step 5: Multi-section defect point cloud splicing analysis

[0058] By extracting the defect feature points in each section simulation data and calculating the three-dimensional coordinates, the multi-section simulation data is fused to construct the three-dimensional model of the height defect and angle defect interval by combining the point cloud splicing algorithm.

[0059] The deep hole inner surface defect detection method based on endoscopic multiplication imaging can realize non-destructive measurement of deep hole inner surface defects; by constructing the defect height and defect angle characterization model, the mapping relationship between the light propagation path and the inner wall geometry is accurately described; and the accuracy and feasibility of the model and the detection method are verified through experiments and simulation; the three-dimensional reconstruction of the deep hole inner surface defect model is realized by multi-section defect point cloud splicing, and the deep hole inner surface defect is accurately restored. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 (a) is a single light line return defect height change and exit light line to axis distance offset representation model graph, Figure 1 (b) is a deep hole reflection plane, Figure 1 (c) is an exit light line plane;

[0061] Figure 2 (a) is a single light line return defect angle change and exit light line to axis distance offset representation model graph, Figure 2 (b) is a deep hole reflection plane, Figure 2 (c) is an exit light line plane;

[0062] Figure 3 (a) is a deep hole inner surface defect height change schematic diagram; Figure 3 (b) is a deep hole inner wall defect height resulting in light change graph;

[0063] Figure 4 (a) is a deep hole inner surface defect height ΔX=0.08mm, and the simulation result graph of the defect resulting in single light line to axis distance change Y=3.4542mm is obtained; Figure 4 (b) is a deep hole inner surface defect height ΔX=0.18mm, and the simulation result graph of the defect resulting in single light line to axis distance change Y=3.2744mm is obtained; Figure 4(c) is the simulation result diagram of the single light ray to the axis distance change Y=3.7417mm caused by the defect with the height ΔX=-0.08mm of the inner surface defect of the deep hole; Figure 4 (d) is the simulation result diagram of the single light ray to the axis distance change Y=3.9214mm caused by the defect with the height ΔX=-0.18mm of the inner surface defect of the deep hole;

[0064] Figure 5 The schematic diagram of the angle change of the inner surface defect of the deep hole;

[0065] Figure 6 (a) is the simulation result diagram of the single light ray to the axis distance change Y=3.6787mm caused by the defect with the angle θ=0.2° of the inner surface defect of the deep hole; Figure 6 (b) is the simulation result diagram of the single light ray to the axis distance change Y=3.7774mm caused by the defect with the angle θ=0.4° of the inner surface defect of the deep hole; Figure 6 (c) is the simulation result diagram of the single light ray to the axis distance change Y=3.4858mm caused by the defect with the angle θ=-0.2° of the inner surface defect of the deep hole; Figure 6 (d) is the simulation result diagram of the single light ray to the axis distance change Y=3.3678mm caused by the defect with the angle θ=-0.4° of the inner surface defect of the deep hole;

[0066] Figure 7 (a) is the image collected by the camera; Figure 7 (b) is the image processed by MATLAB;

[0067] Figure 8 is the comparison diagram of the simulation and experimental results;

[0068] Figure 9 (a) is the point cloud splicing diagram of the light result change point caused by the height of the inner surface defect of the deep hole; Figure 9 (b) is the enlarged view of the height defect;

[0069] Figure 10 (a) is the point cloud splicing diagram of the light result change point caused by the angle of the inner surface defect of the deep hole; Figure 10 (b) is the enlarged view of the angle defect;

[0070] Figure 11 is the flowchart of the method. DETAILED DESCRIPTION

[0071] The specific embodiments of the present application are further described below in combination with the drawings and technical solutions.

[0072] A deep hole inner surface defect detection method, the steps are as follows:

[0073] Step 1: Analysis of the influence mechanism of internal surface defects on imaging results

[0074] When there are defects on the inner wall of a deep hole, the light path will be deformed due to the defects during propagation, causing the geometric shape of the annular light spot in the imaging result to be distorted. By analyzing this distortion feature, a connection between light propagation and defect characteristics is established, providing basic data for subsequent defect detection and three-dimensional reconstruction.

[0075] Step 2: Construction of defect characterization model

[0076] In the construction of the defect height characterization model, a single light ray in the annular light beam is selected as the research object, which enters the deep hole parallel to the axis and reflects in the process, combined with the reflection path, an X-Y-Z three-dimensional coordinate system is established; according to the analysis of light propagation law, based on the propagation path and combined with the geometric characteristics of the defect area, such as Figure 1 Based on the geometric optics principle and spatial coordinate system transformation, combined with the system parameter constraint condition, the characterization model between the height change ΔX of the internal surface defect of the deep hole and the distance Y of the outgoing light ray to the axis is constructed as:

[0077]

[0078] In the construction of the defect angle characterization model, the modeling method is consistent with the analysis idea of height defects, aiming to realize the quantitative characterization of angle change. Specifically, assuming that there is an angle deviation θ at a point on the inner wall of the deep hole, when the incident light beam reaches the position and reflects, the reflection path will be offset due to the change of the surface normal, and then affect the final imaging result, as shown in Figure 2 Based on the geometric optics principle and spatial coordinate system transformation, combined with the system parameter constraint condition, the characterization model between the angle change θ of the defect and the distance Y of the outgoing light ray to the axis is constructed as:

[0079]

[0080] Step 3: Single-section optical software simulation and model verification

[0081] Firstly, the deep hole scene containing virtual defects is constructed by using the optical simulation software Trace Pro; secondly, by setting different defect height and angle parameters, the propagation path and imaging process of light under various defect conditions are simulated; finally, combined with the image feature changes obtained in the simulation process, the applicability and accuracy of the constructed model are verified.

[0082] In the process of constructing the height defect simulation platform, the inner wall of the deep hole is first simplified as a plane structure, and a protruding point is set on the plane to simulate the existence of the defect. By adjusting the height parameter of the protruding point, the influence of defects of different depths on the light propagation path is simulated. The distance between the outgoing light and the hole axis is observed. As shown in Figure 3 , the figure shows the deviation of the single light reflection path under the condition of the height change of the inner wall surface of the deep hole. Subsequent simulation analysis will be based on the method shown in the schematic diagram.

[0083] With the change of the position of the inner surface of the deep hole, the height of the local protruding point of the inner wall also changes, causing the deviation of the single light reflection path, and ultimately leading to the change of the three-dimensional coordinates of the light on the receiving screen. In the simulation experiment, by gradually adjusting the height defect parameter, the measurement range of the height change of the inner surface of the deep hole is obtained as-0.18mm-0.18mm. According to the range, multiple height adjustments are made, and the light propagation simulation is completed in Trace Pro.

[0084] The defect height representation model in the technical scheme can calculate the theoretical model results, and the simulation data is compared with the theoretical model calculation results. Through comparative analysis, the average error is 10.9μm, and the overall error is small; through the results obtained by the theoretical model and simulation, the distance change data of the outgoing light to the axis under different height defect conditions are analyzed by selecting representative data points. As shown in Figure 4 , when there is a height change in the inner wall, the imaging result compared with the normal state (the black circle represents the outgoing light result of the inner surface of the deep hole without defects, ΔX represents the height change of the inner surface defect, and Y represents the distance change of the single light to the axis caused by the inner surface defect), it can be seen that the single light to the axis caused by the inner surface defect has a significant deviation. Based on this, the identification of single-point height defects can be realized. Further, according to the distance between the outgoing light and the axis, the corresponding inner wall defect height value can be deduced.

[0085] In the process of constructing the angle defect simulation platform, the inner wall of the deep hole is first simplified as a plane model, and the plane is rotated to simulate the existence of the angle defect. By adjusting the surface inclination angle parameter, the change of the light propagation path under different angle change conditions is effectively simulated. As shown in Figure 5 , is the outgoing light when simulating the defect angle.

[0086] With the change of the angle of the simulated defect on the inner surface of the deep hole, the reflection path of the single light ray will also be offset, eventually leading to the change of the three-dimensional coordinates of the light ray on the receiving screen. In the simulation experiment, by gradually adjusting the angle defect parameters, the measurement range of the angle change of the inner surface of the deep hole is obtained as-0.4° to 0.5°. According to the range, multiple angle adjustments are made, and the corresponding light propagation simulation is completed in TracePro.

[0087] The theoretical model result can be calculated by the defect angle representation model in the technical solution, and the simulation data is compared with the theoretical model calculation result. The average error is 24.88 μm, and the overall error is small; by comparing the results obtained by the theoretical model and simulation, under the condition of ensuring the height unchanged, the distance change data of the outgoing light ray to the axis under different angle defect conditions, and by selecting the representative data points for analysis, it can be found that when there is an angle change on the inner wall, the imaging result compared with the normal state is as shown in Figure 6 It can be seen that the single light ray to the axis distance caused by the inner surface defect is obviously offset. Based on this, the identification of single point angle defect can be realized. Further, according to the distance between the outgoing light ray and the axis, the corresponding inner wall defect angle value can be deduced.

[0088] Step 4: Multi-section defect point cloud splicing

[0089] Firstly, in order to verify the actual accuracy of the deep hole inner surface endoscopic magnification imaging scheme, a defect-free circular hole part with an inner diameter of 16 mm is detected, based on the experimental table which has been built, the ring-shaped structured light containing the inner wall topography information formed on the receiving screen is collected by the industrial camera, as shown in Figure 7 (a), and the original image is processed by using the MATLAB image processing method, as shown in Figure 7 (b), the structured light fringe center point coordinate data is obtained; secondly, the ring-shaped structured light is obtained by Trace Pro simulation, and the structured light fringe center point coordinate data is obtained; finally, the actual extracted structured light fringe center point coordinate data is compared with the theoretically calculated coordinate points generated by simulation, as shown in Figure 8 The experimental data show that the root mean square error between the actual measured coordinates and the simulation prediction value is less than 10 μm, which proves the actual accuracy of the experimental method, and the accurate imaging of the deep hole inner surface can be realized.

[0090] In the defect height measurement experiment, on the basis of the original experimental table, a high-precision Z-axis micro-motion platform is installed below the measured part to controllably adjust the height parameter. The high-precision Z-axis micro-motion platform is used to move the platform position step by step at a step length of 10 μm, simulating the inner wall defect situation at different heights; and by controlling the displacement table, the measured part is uniformly moved at a step length of 10 μm, realizing the overall detection of the partial area of the inner surface of the deep hole. After each displacement adjustment, the imaging image on the receiving screen is collected by the industrial camera, the imaging coordinates of each section are extracted through image processing, the coordinates of each section are spliced along the axial direction, and three-dimensional point cloud data are generated, as shown in Figs. Figure 9 (a) and (b). Based on the height defect mathematical model proposed in the technical solution, the three-dimensional point cloud data are processed and inverted by combining the optical path reverse analysis method, and the accurate measurement of the height defect of the inner surface of the deep hole is realized.

[0091] In the defect angle measurement experiment, on the basis of the original experimental table, two high-precision micro-motion platforms are used to control the height changes of the two ends of the measured part, the height difference between the two platforms is adjusted, specifically, the height difference is gradually increased at a step length of 10 μm, so that a controllable angular deflection is generated, thereby realizing the continuous adjustment of the dynamic deflection angle θ (the range is -0.2° to 0.3°), simulating the defect situation at different inclination angles; and by controlling the displacement table, the measured part is uniformly moved at a step length of 10 μm, realizing the overall detection of the partial area of the inner surface of the deep hole. After each displacement adjustment, the imaging image on the receiving screen is collected by the industrial camera, the imaging coordinates of each section are extracted through image processing, the imaging coordinates of each section are extracted, the coordinates of each section are spliced along the axial direction, and three-dimensional point cloud data are generated, as shown in Figs. Figure 10 (a) and (b). Based on the angle defect mathematical model proposed in the technical solution, the three-dimensional point cloud data are processed and inverted by combining the optical path reverse analysis method, and the accurate measurement of the angle defect of the inner surface of the deep hole is realized.

[0092] The description presented in the above exemplary embodiments is only used to illustrate the technical solutions of the present application, and is not intended to be exhaustive, nor is it intended to limit the present application to the precise forms described. Obviously, many changes and variations are possible for those skilled in the art based on the above teachings. The exemplary embodiments are selected and described in order to explain the specific principles of the present application and its practical application, so that other skilled persons in the art can easily understand, implement and utilize various exemplary embodiments of the present application and various selected forms and modifications thereof. The scope of protection of the present application is intended to be defined by the appended claims and their equivalent forms.

Claims

1. A method for detecting defects on the inner surface of deep holes based on endoscopic magnification imaging, characterized in that, The steps are as follows: Step 1: Analysis of the mechanism by which internal surface defects affect imaging results; When there are defects on the inner surface of a deep hole, the light path will be deformed due to the defects during propagation, which will distort the geometry of the ring-shaped light spot in the imaging result; by analyzing the distortion characteristics, the relationship between light propagation and defect characteristics can be established. Step 2: Construction of a defect height characterization model; Step 3: Construction of a defect-perspective characterization model; Step 4: Single-section optical simulation verification; A deep hole scene containing virtual defects was constructed using optical simulation software. Different defect height and angle parameters were set to simulate the light propagation path and imaging process. Simulations were performed on height defects and angle defects respectively. The corresponding parameters were adjusted to obtain the measurement range. The simulation data was compared with the calculation results of the theoretical model to verify the accuracy of the defect characterization model. Step 5: Multi-section defect point cloud splicing analysis; By extracting the defect feature points from the simulation data of each section and calculating their three-dimensional coordinates, and combining them with a point cloud stitching algorithm, the simulation data of multiple sections are fused to construct a three-dimensional model of the height defect and angle defect range.

2. The method for detecting deep hole internal surface defects based on endoscopic magnification imaging according to claim 1, characterized in that, The specific steps for constructing the defect height characterization model in step 2 are as follows: A single ray from the ring beam is selected, parallel to the axis of the deep hole, and enters the deep hole for reflection. Based on the reflection path, a three-dimensional XYZ coordinate system is established: the origin is the intersection of the axis of the deep hole and the right end face of the magnifying lens group; the X-axis is along the radial direction of the deep hole, pointing to the outer side of the inner wall of the deep hole, representing the direction of the defect height, i.e., the vertical direction of the convex or concave part of the inner wall of the deep hole; the Y-axis is along the radial direction of the deep hole and perpendicular to the X-axis, representing the direction of the distance of the outgoing ray from the axis within the cross-section of the deep hole, i.e., the radial offset direction of the light spot on the receiving screen; the Z-axis is along the axial direction of the deep hole, representing the axial position of the deep hole, i.e., the depth position of the defect within the deep hole. Define the following feature points: Point M is the intersection of the reflected light rays from the first conical surface of the magnifying lens group and the reflected light rays from the inner wall of the deep hole; Point L is the intersection of the incident ray and the ray reflected from the first cone surface of the magnifying lens group; Point P is the actual reflection point of the defect region on the inner wall of the deep hole; points S and U are auxiliary points for light propagation on the second conical surface of the magnifying lens group; points T, K, and Q are auxiliary points for light propagation, which play a supporting role in explaining the spatial propagation of light and the geometric relationship between the inner wall of the deep hole and the light, and are used to construct a clearer optical detection model; PT represents the incident path offset; PK and PL are auxiliary lines for geometric derivation; MQ represents the lateral offset; NS is an auxiliary line used to derive the mathematical relationship between defect features and the offset of the outgoing light rays; Point I is a point on the second conical surface of the magnifying lens assembly; point F is a point on the deep hole axis, used to define the position of the deep hole axis; IF represents the offset correction amount; MF serves as an auxiliary line perpendicular to the deep hole axis, used to derive the mathematical relationship between defect characteristics and the offset of the outgoing light ray. When the change in the height of the inner surface defect is 0, MF is equal to the incident aperture R; IP serves as the hypotenuse in △IPT, used for geometric derivation; point N is the intersection of the light ray after reflection at the actual reflection point P and the second conical surface of the magnifying lens assembly; point Z is the intersection of the incident light ray and the light ray reflected from the inner wall of the deep hole. Points H and E are auxiliary points for geometric analysis; HE serves as an auxiliary line used to derive the mathematical relationship between defect features and outgoing ray deflection; NZ and ZP are the two sides of △NZP, and ZE and NP are the height and base of △NZP respectively, used for geometric derivation; NE is the right-angled side of △NZE, used for geometric derivation; ZL is an auxiliary line for geometric derivation; point C is the fixed reference point of the intensifying lens group; point A is the intersection of the extension of the generatrix of the cylindrical surface of the intensifying lens group and the perpendicular line of the incident ray, serving as an auxiliary point for geometric analysis; CA is the right-angled side of △CAL, used for geometric derivation. Based on the principles of geometric optics and spatial coordinate system transformation, and combined with system parameter constraints, a characterization model is constructed for the relationship between the change in the height of the defect on the inner surface of the deep hole ΔX and the distance Y from the outgoing ray to the axis; where the incident aperture radius R, the incident ray radius r, the mirror group spacing d, the offset correction amount IF, the reference length HC, the critical edge distance ΔC, θ1, and θ2 are all known. The incident path offset is determined by both the incident radius of the ray and the geometric deformation, as shown in the following formula: ； Orthogonally decompose isosceles triangle △NZP, with the foot of the perpendicular E satisfying: ； The length of segment NP is obtained from geometric symmetry: ； Suppose that a moving point P moves along the ZL axis, and its perpendicular distance to the baseline HC is: ； The geometric chain relationship of PL segment is simplified as follows: ； The ZL axis length and the lateral offset MQ satisfy the following: ； The relationship between the perpendicular distance from point N to point F and the perpendicular distance from point M to the incident radius of the ray is expressed as follows: ； Based on geometric derivation, the distance Y from the incident ray to the axis is expressed as the perpendicular distance from point N to point F, i.e., Y = NS + IF; the change in the height of the inner surface defect ΔX is equal to the difference between the incident aperture R and the perpendicular distance from point M to the incident radius of the ray, i.e., ΔX = R - MF. In summary, by solving the above equations simultaneously and eliminating intermediate variables, the relationship between the change in inner surface defect height ΔX and the distance Y from the outgoing ray to the axis is as follows: 。 3. The method for detecting deep hole internal surface defects based on endoscopic magnification imaging according to claim 1, characterized in that, The specific implementation process of constructing the defect perspective characterization model in step 3: Assuming there is an angular deviation θ at a certain point on the inner wall of the deep hole, we analyze the offset of the reflection path when the incident beam reaches this position and is reflected. Based on the reflection path, the coordinate system established here shares a three-dimensional coordinate system with the one in step two. The same letters represent the same meaning for feature points; in addition, points B and D are points on the plane of the outgoing ray, used to help describe the ray; BD is an auxiliary line used to derive the outgoing radius; IN, DN, DI and IS are auxiliary lines on the second cone surface of the magnifying lens group, which are related to the outgoing ray and used for geometric derivation. ZP, ZQ, and ZL are located on the incident ray and are used for geometric derivation; Based on the principles of geometric optics and spatial coordinate system transformation, and combined with system parameter constraints, a characterization model is constructed for the relationship between the defect angle change θ and the distance Y from the outgoing ray to the axis; where DI and MQ are both known. The incident ray deflection is determined by the distance to the receiving screen and the inner diameter deflection angle, and its expression is: ； The length of segment NP is obtained by orthogonal decomposition of the isosceles triangle △NZP: ； The launch radius can be decomposed into three parts: Launch radius = BD + NS + IF; BD segment: From the geometric projection relationship, we get: ； NS segment: From orthogonal decomposition, we get: ； A moving point P moves along the ZL axis, and its perpendicular distance PL satisfies: ； The length of the ZL axis is determined by the geometric constraints of MQ and the change in the defect angle θ: ； Based on the displacement relationship of the moving point: ； Based on ∠ZMQ = 64° - 2θ and ∠MZQ = 26° + 2θ, the side-angle relationship of triangle △MZQ is constructed using the Law of Sines: ； Combining the expression for ZL, we obtain the implicit equation for MQ: ； Solving the above equations simultaneously and eliminating intermediate variables, we obtain the explicit relationship between the dynamic angle θ and the exit radius Y: 。

Images