Drawing welding seam information identification method and system

Abstract

The invention discloses a drawing welding seam information identification method and system, and relates to the technical field of welding seam information identification. A welding seam line three-dimensional coordinate set is marked as a space welding seam multi-segment line point column, a least square method is adopted to fit a main plane where the point column is located, and a main plane normal vector is calculated; performing orthogonal projection on the three-dimensional coordinate points along a normal vector to obtain a two-dimensional point column; the Euclidean distance between adjacent points of the two-dimensional point column is calculated and accumulated to obtain the total arc length of the welding seam, the endogenous scale is calculated according to the total number of the point columns, and a two-dimensional equal-arc-length point column is generated through endogenous scale interpolation; offsetting different times of endogenous scales along the normal direction of the two-dimensional equal-arc-length point array, constructing five closed parallel seams and forming a parallel seam group; and calculating a discrete curvature sequence of each parallel seam by using a second-order central difference method to obtain a curvature mean value and a standard deviation, further calculating a constant-curvature closed seam index, constructing an index quintuple to judge an invariant Boolean value so as to determine whether the seam is a constant-curvature closed circular seam, calculating an estimated radius, and summarizing a structured parameter packet.

Description

Technical Field

[0001] This invention relates to the field of weld seam information recognition technology, and specifically to a method and system for identifying weld seam information on drawings. Background Technology

[0002] In fields such as steel structure manufacturing, reliable connections between components rely heavily on weld processing. During the design phase, weld models are constructed using professional design software such as CAD, BIM, and Tekla. These models store welds as spatial weld polyline point sequences, composed of the three-dimensional coordinates of the weld seam and associated with the topological relationships of the parts to be connected. Once the design is complete, this weld information needs to be transferred from the design phase to the production-end MES system, welding process specification system, and welding equipment control system. This process requires accurate identification of constant curvature closed circumferential welds and extraction of structured parameters such as the total arc length, radius, and principal plane normal vector. These parameters are crucial for subsequent welding process planning, welding equipment allocation, and quality inspection. By using these parameters, it is ensured that the production end completes welding operations efficiently and accurately according to design requirements, avoiding production delays and quality issues caused by missing or incorrect parameters.

[0003] In the weld data transfer process from the design end to the production end, the core technical problem is that existing weld identification methods cannot meet the requirements of automation and accuracy. The weld data exported by the design software only contains the three-dimensional coordinate information of the spatial weld polyline point series, without indicating the specific type of weld. Moreover, due to the differences in design software algorithms and modeling operation habits, the spatial weld polyline point series from different sources have uneven density, with some areas having dense point series and others having sparse point series. Existing identification methods are susceptible to interference from the density of points when dealing with uneven point sequences. This can lead to misjudgments of the same type of constant curvature closed loop welds. Manual intervention is required for repeated adjustments based on different projects and component types, resulting in poor adaptability. These issues prevent existing methods and systems from achieving automated deterministic identification of constant curvature closed loop welds using the weld's own three-dimensional coordinate point sequence. It is also difficult to reliably extract its structured parameters. Ultimately, the transfer of weld data from the design end to the production end requires manual verification and supplementation. This not only reduces the automation efficiency of the production process but may also lead to welding process mismatches due to human judgment errors, affecting weld formation quality and even posing potential safety hazards to critical components such as steel structures. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a method and system for identifying weld information in drawings, solving the problem that existing technologies cannot achieve automated deterministic identification of constant curvature closed circumferential welds through the three-dimensional coordinate point array of the weld itself.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The three-dimensional coordinate set of the weld line is marked as a spatial weld polyline point sequence. The principal plane of the spatial weld polyline point sequence is fitted by the least squares method and the normal vector of the principal plane is calculated. The elements in the spatial weld polyline point sequence are orthogonally projected onto the principal plane along the direction of the normal vector of the principal plane to obtain a two-dimensional point sequence. The total arc length of the weld, which reflects the total length of the weld, is obtained by calculating the Euclidean distance between two adjacent two-dimensional coordinate points in the two-dimensional point sequence. Based on the total number of elements in the spatial weld polyline point sequence and the total arc length of the weld, the endogenous scale reflecting the unified scale reference is calculated. Then, a two-dimensional equal arc length point sequence is generated by interpolation of the endogenous scale. For each point in the two-dimensional equal arc length point sequence, the tangent direction of the curve is calculated by the adjacent point difference method, and the normal direction vector is derived. Using the endogenous scale as the offset scale, the two-dimensional equal arc length point sequence is offset point by point along the direction of the normal direction vector to construct five closed parallel seams, and the five closed parallel seams are combined into a parallel seam family. For each parallel seam in the parallel seam family, the discrete curvature sequence of each parallel seam is calculated using the second-order central difference method. The mean curvature and standard deviation of the discrete curvature sequence are calculated. Based on the mean curvature and standard deviation of curvature, the constant curvature closure seam index of each parallel seam is calculated. The constant curvature closure seam index of each parallel seam is evaluated for equivalence, and the evaluation result is marked as an invariant Boolean value. The constant curvature closure seam index is an index that measures the constancy of the curvature of the weld line along the arc length direction.

[0006] Furthermore, the elements in the polyline point array of the spatial weld are three-dimensional coordinate points, and the total number of elements N in the polyline point array of the spatial weld is ≥ 5.

[0007] Furthermore, assuming the equation of the principal plane is ax + by + cz + d = 0, the coefficients a, b, and c of the plane equation are obtained by minimizing the sum of the squared distances from all elements in the polyline point sequence of the spatial weld to the plane. The vector formed by the coefficients a, b, and c is the normal vector of the principal plane. The principal plane is the plane in which the weld line mainly extends in three-dimensional space.

[0008] Furthermore, each three-dimensional coordinate point in the polyline point array of the spatial weld is orthogonally projected onto the principal plane along the direction of the normal vector of the principal plane to obtain the two-dimensional coordinate point corresponding to each three-dimensional coordinate point. After all the two-dimensional coordinate points are combined into a set, a two-dimensional point array is obtained. The Euclidean distance between two adjacent two-dimensional coordinate points in the two-dimensional point column is calculated sequentially, and the total arc length of the weld corresponding to the two-dimensional point column is obtained by summing all the calculated Euclidean distances.

[0009] Furthermore, after subtracting 1 from the total number of elements N in the polyline point array of the spatial weld, the total number of intervals between adjacent elements is N-1. The value obtained by dividing the total arc length of the weld by (N-1) is marked as the endogenous scale. The first point in the two-dimensional point sequence is selected as the first point of the new point sequence. Starting from the first point of the new point sequence, new points are determined by linear interpolation with the endogenous scale as the step size. This process continues until the total number of selected points reaches N. Finally, a new point sequence is obtained in which the arc lengths of adjacent points are all at the endogenous scale. This new point sequence is marked as a two-dimensional equal arc length point sequence.

[0010] Furthermore, for each point q in the two-dimensional array of points with equal arc lengths... i If 2≤i≤N-1, the tangent direction vector t can be calculated by performing central difference through adjacent points symmetrical to the arc length. i =q i+1 -q i-1 ; If i=1, the tangent direction vector t1=q2-q1 is calculated by performing forward difference through the adjacent points on the right. If i=N, the tangent direction vector t is calculated by performing backward difference through the left adjacent points. N =q N -q N-1 ; The above tangent direction vector t i Rotate counterclockwise by 90° to convert into the normal direction vector n i Output the normal direction vector n of each point in a two-dimensional array of points with equal arc lengths. i .

[0011] Furthermore, using the endogenous scale as the offset scale, along the normal direction vector n i By offsetting each point in a two-dimensional array of points of equal arc length, five closed parallel seams are constructed, as follows: Parallel seam C -2 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of the endogenous scale to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C -1 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of one endogenous scale to obtain a new point sequence, fine-tune the endpoints to close it; Original seam C0: The two-dimensional equal arc length point column itself, with an offset of 0; Parallel seam C +1 For each point q in a two-dimensional array of points with equal arc lengths iAfter offsetting by one time the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C +2 For each point q in a two-dimensional array of points with equal arc lengths i After offsetting by twice the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. The above five closed parallel seams are combined into a parallel seam family {C} -2 C -1 C0, C +1 C +2}

[0012] Furthermore, for each parallel seam C in the parallel seam family... j j={-2, -1, 0, +1, +2}, the second-order central difference method is used to calculate C for each parallel seam. j The discrete curvature sequence is as follows: For parallel seam C j The i-th point q i (j), where 2≤i≤N-1, calculate [q i+1 (j)-2q i (j)+q i-1 The modulus length of (j) is obtained by dividing the modulus length by the square of the endogenous scale to obtain the parallel seam C. j The i-th point q i The discrete curvature of (j) is used to obtain the parallel seam C by summing the discrete curvatures. j Discrete curvature sequence k i (j); Calculate C for each parallel seam j Discrete curvature sequence k i (j) is the mean curvature and standard deviation of curvature. Based on the mean curvature and standard deviation of curvature, C is calculated for each parallel seam. j The constant curvature closure seam index is calculated as follows: When parallel seam C j When the corresponding average curvature is greater than 0, subtract the parallel seam C from the value of 1. j The difference obtained by dividing the standard deviation of curvature by the mean curvature is denoted as parallel seam C. j The constant curvature closure index is denoted as HCCI(j); When parallel seam C j When the corresponding average curvature is ≤0, the parallel seam is determined to be a non-circular feature, and HCCI(j) is recorded as an outlier. Extract the average curvature of the original seam C0 in the parallel seam family and label it as u0; The constant curvature closure seam index HCCI(j) is constructed as a constant curvature closure seam index quintuple [HCCI(-2),HCCI(-1),HCCI(0),HCCI(+1),HCCI(+2)]; If HCCI(-2)=HCCI(-1)=HCCI(0)=HCCI(+1)=HCCI(+2), define an invariant Boolean value and determine if the invariant Boolean value is true; otherwise, it is false.

[0013] Furthermore, the invariant Boolean value is obtained. If the invariant Boolean value is true, the weld is determined to be a closed circumferential weld with constant curvature. If the invariant Boolean value is not true, then the weld is determined to be a non-circular weld. When the weld is determined to be a closed circumferential weld with constant curvature, the estimated radius of the weld is calculated as follows: The value obtained by dividing the numerical value 1 by the average curvature u0 of the original weld C0 is the estimated radius of the weld. The estimated radius, total arc length, principal plane normal vector, and original constant curvature closure index HCCI(0) of the welds that are determined to be constant curvature closed circumferential welds are summarized into a structured parameter package.

[0014] Furthermore, a drawing weld information recognition system is proposed to implement a drawing weld information recognition method as described in any of the above claims, including: The 3D to equal arc length module exports the 3D coordinate set of the weld seam line from the drawing or model, forming a spatial weld polyline point sequence. It uses the least squares method to fit the principal plane where the point sequence is located and calculates the normal vector. It orthogonally projects the 3D points along the normal vector to obtain a 2D point sequence. It calculates the Euclidean distance between adjacent points in the 2D point sequence and accumulates them to obtain the total arc length of the weld. After calculating the intrinsic scale, it uses the intrinsic scale interpolation to generate a 2D equal arc length point sequence. The parallel seam family construction module calculates the tangent vector for a two-dimensional equal arc length point sequence using the center difference for the middle point and the front and rear differences for the first and last points. After rotating the tangent vector counterclockwise by 90°, the normal direction vector is obtained. The endogenous scale is offset by different multiples along the normal and the opposite direction to generate five closed parallel seams, which are combined into a parallel seam family. The curvature index determination module uses the second-order central difference method to calculate the discrete curvature sequence of each parallel seam, calculates the mean and standard deviation of curvature to obtain the constant curvature closed seam index, constructs a constant curvature closed seam index quintuple, determines whether the values ​​in the quintuple are equal, and defines an invariant Boolean value to record the determination result. The weld identification and determination module determines the weld type based on invariant Boolean values. When the invariant Boolean value is true, the weld is determined to be a constant curvature closed circumferential weld. The estimated radius of the constant curvature closed circumferential weld is calculated and summarized into a structured parameter package.

[0015] Compared with existing technologies, it has the following advantages: This solution proposes a method and system for identifying weld information on drawings. It fits the principal plane containing the polyline point array of the spatial weld using the least squares method, orthogonally projects the three-dimensional coordinate points along the normal vector of the principal plane to obtain a two-dimensional point array. Based on the total number of elements in the polyline point array and the total arc length of the weld, an intrinsic scale is calculated. Then, a two-dimensional equal-arc-length point array is generated through interpolation using the intrinsic scale. This effectively solves the identification error problem caused by uneven point array density in existing methods. Existing methods are easily affected by curvature calculations when dealing with differences in point array density, and require manual adjustment to adapt to different scenarios. In this solution, the intrinsic scale is determined only by the geometric length of the weld itself and the total number of points, requiring no manual intervention. The two-dimensional equal-arc-length point array uses linear interpolation to ensure that the arc length of adjacent points is equal to the intrinsic scale, completely eliminating the influence of point array density on subsequent calculations. This ensures the accuracy of discrete curvature sequence calculations and avoids misjudgments of the same type of constant curvature closed loop weld due to differences in point array density, significantly improving the recognition adaptability. This solution proposes a method and system for identifying weld information on drawings. It constructs a family of parallel seams, calculates the discrete curvature sequence of each parallel seam using the second-order central difference method, and obtains the constant curvature closed seam index by combining the average curvature and the standard deviation of curvature. Then, it uses a five-tuple of the constant curvature closed seam index to determine the invariant Boolean value, thus solving the problem that existing methods cannot achieve deterministic identification of constant curvature closed circumferential seams. The constant curvature closed seam index, as an indicator of the constant curvature of the weld, is calculated solely based on the weld's geometric information, without relying on external text labels, and its calculation logic is clear and traceable. The invariant Boolean value automates the determination of constant curvature closed circumferential seams by judging whether the constant curvature closed seam indices of the five parallel seams are equal. The determination criteria are clear and unique, giving the identification results strong determinism and avoiding subjective errors caused by human experience. This solution proposes a method and system for identifying weld information on drawings. It determines the weld type by judging invariant Boolean values, and calculates the estimated radius for constant curvature closed circumferential welds using the average curvature of the original weld. The estimated radius, total weld arc length, principal plane normal vector, and constant curvature closed circumferential weld index are summarized into a structured parameter package. Combined with four main functional modules—a 3D-to-equal-arc-length conversion module and a parallel weld family construction module—this solution solves the problem of manual intervention in weld data transfer from design to production. The structured parameter package directly provides key parameters required for process planning and equipment assignment at the production end, eliminating the need for manual input. The system's modules achieve full automation from 3D coordinate point input to parameter package output, eliminating the need for manual intervention in data processing. This effectively avoids the risk of welding process mismatch, ensures weld formation quality, improves data transfer efficiency from design to production, and eliminates safety hazards in critical components caused by errors in manual data processing. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This application provides a method for identifying weld information in drawings; The method specifically includes the following steps: Step 1: Export the 3D coordinate set of the weld seam line from the drawing or model to obtain a spatial weld polyline point sequence. The elements in the spatial weld polyline point sequence are 3D coordinate points. The total number of elements N in the spatial weld polyline point sequence is ≥ 5 to ensure the effectiveness of subsequent curvature calculations. Use the least squares method to fit the principal plane where the spatial weld polyline point sequence is located, and then calculate the normal vector of the principal plane. The specific calculation process is as follows: Assuming the equation of the principal plane is ax + by + cz + d = 0, the coefficients a, b, and c of the plane equation are obtained by minimizing the sum of squared distances from all elements in the polyline point sequence of the spatial weld to the plane. The vector formed by the coefficients a, b, and c is the normal vector of the principal plane. Specifically, the principal plane refers to the plane that can contain the polyline points of the spatial weld to the greatest extent. It is the plane on which the weld line mainly extends in three-dimensional space. It is used to project the spatial weld onto this plane and transform it into a two-dimensional curve to simplify geometric analysis. The principal extension plane of the weld line is automatically determined by an unsupervised geometric fitting method. The objectivity and consistency of the plane selection can be guaranteed without manual intervention. It provides a spatial reference for the subsequent transformation of the spatial weld into a planar curve and eliminates the interference of spatial dimension on the subsequent curvature analysis. Each three-dimensional coordinate point in the polyline point array of the spatial weld is orthogonally projected onto the principal plane along the direction of the normal vector of the principal plane to obtain the two-dimensional coordinate point corresponding to each three-dimensional coordinate point. After all the two-dimensional coordinate points are combined into a set, a two-dimensional point array is obtained. Specifically, the three-dimensional geometric shape of the weld in space is transformed into a curve representation in a two-dimensional plane, eliminating the influence of spatial dimension differences on subsequent geometric analysis. This allows subsequent operations such as curvature calculation and parallel seam family construction to be performed only in a two-dimensional plane, greatly simplifying the complexity of technical implementation. For a two-dimensional point sequence, calculate the Euclidean distance between adjacent two-dimensional coordinate points in the two-dimensional point sequence in sequence, and accumulate all the calculated Euclidean distances to obtain the total arc length of the weld corresponding to the two-dimensional point sequence. The total arc length of the weld reflects the total length of the weld; Since the total number of elements in the spatial weld polyline point sequence is N, the total number of intervals between adjacent elements is N - 1. The value obtained by dividing the total arc length of the weld by (N - 1) is marked as the endogenous scale. The endogenous scale is only related to the geometric length and the number of points of the weld itself, and is a unified scale benchmark for subsequent parallel seam family construction and curvature calculation, ensuring that the offset matches the geometric scale of the weld itself; Select the first point in the two-dimensional point sequence as the first point of the new point sequence. Starting from the first point of the new point sequence, determine new points by linear interpolation with the endogenous scale as the step size until the total number of selected points reaches N and then stop. Finally, a new point sequence with the arc length between adjacent points being the endogenous scale is obtained. Mark this new point sequence as the two-dimensional equal-arc-length point sequence. Specifically, in the original two-dimensional point sequence, the arc lengths between adjacent points may be uneven. In some places, the points are dense and the arc lengths are short, while in some places, the points are sparse and the arc lengths are long. With the endogenous scale as the fixed arc length step size, reselect points for the two-dimensional point sequence. Finally, a two-dimensional equal-arc-length point sequence with the number of points still being N is obtained. Since the curvature is directly related to the arc length change rate, only when the arc length is uniform can the calculated curvature truly reflect the curvature constancy of the weld. If the arc length is uneven, the curvature calculation will be interfered by the density of points. Therefore, based on the two-dimensional equal-arc-length point sequence, the accuracy of subsequent discrete curvature calculation can be guaranteed.

[0019] Step 2: For each point q in the two-dimensional equal-arc-length point sequence i , calculate the curve tangent direction through adjacent point difference and deduce the normal direction as follows: If 2 ≤ i ≤ N - 1, perform central difference through adjacent points symmetric about the arc length to calculate the tangent direction vector t i = q i+1 - q i-1 ; If i = 1, perform forward difference through the adjacent point on the right to calculate the tangent direction vector t1 = q2 - q1; If i = N, perform backward difference through the adjacent point on the left to calculate the tangent direction vector t N = q N - q N-1 ; Rotate the above tangent direction vector t i counterclockwise by 90° to convert it into the normal direction vector n i , for example, if the tangent vector is (a, b), then the normal direction vector is (-b, a), and output the normal direction vector n of each point in the two-dimensional equal-arc-length point sequence iSpecifically, the normal direction vector is derived from the difference between adjacent points of the equal arc length point column, without the need for additional curve fitting, which ensures the objectivity and geometric correlation of the normal calculation. At the same time, the normal direction vector clarifies the direction of the normal of the curve at each discrete point, providing a unified direction reference for the subsequent construction of parallel seams along the normal, and ensuring the consistency of multiple sets of parallel seams in geometric offset. Using the endogenous scale as the sole offset scale, along the normal direction vector n i By offsetting each point in a two-dimensional array of points of equal arc length, five closed parallel seams are constructed, as follows: Parallel seam C -2 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of the endogenous scale to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C -1 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of one endogenous scale to obtain a new point sequence, fine-tune the endpoints to close it; Original seam C0: This is the two-dimensional equal arc length point sequence itself, with an offset of 0; Parallel seam C +1 For each point q in a two-dimensional array of points with equal arc lengths i After offsetting by one time the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C +2 For each point q in a two-dimensional array of points with equal arc lengths i After offsetting by twice the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. The above five closed parallel seams are combined into a parallel seam family {C} -2 C -1 C0, C +1 C +2 The parallel seam family is used for subsequent calculation of the constant curvature closed seam index and verification of normal invariance; Specifically, a closed parallel seam refers to a closed curve formed by offsetting a specific endogenous scale along the normal direction of the original weld seam based on a two-dimensional equal arc length point array. Five closed parallel seams are constructed to form a family of normal gradient parallel seams. The normal invariance of the constant curvature closed seam index is verified by applying offsets of -2, -1, 0, +1, and +2 times the endogenous scale to the original seam in the normal direction. Since the curvature of annular welds is approximately constant, their constant curvature closed seam index should remain stable under small normal perturbations. However, the constant curvature closed seam index of non-annular welds will fluctuate significantly with normal offset. When constructing closed parallel seams, the normal direction of each point is calculated based on the two-dimensional equal arc length point array and the endogenous scale. The direction vector is used to offset each point by a specific multiple of the intrinsic scale along the normal direction and the opposite direction of the normal direction. After retaining the original seam, five sets of points are obtained. Since the offset operation may cause the endpoints to deviate from the closed state, the endpoints need to be finely adjusted to form a closed curve. If the weld is annular, the constant curvature closed seam index of the five parallel seams should be consistent. If it is non-annular, the constant curvature closed seam index will differ due to curvature fluctuations. By offsetting the intrinsic scale by a specific multiple, it is possible to ensure that the disturbance scale is closely related to the arc length characteristics of the weld itself, avoid the deviation introduced by human intervention, and thus ensure the objectivity and reliability of the verification of the normal invariance of the constant curvature closed seam index. Finally, stable identification of weld information is achieved through the constant curvature closed seam index.

[0020] Step 3: For each parallel seam C in the parallel seam family j j={-2, -1, 0, +1, +2}, the second-order central difference method is used to calculate C for each parallel seam. j The discrete curvature sequence is as follows: For parallel seam C j The i-th point q i (j), where 2≤i≤N-1, first calculate the i-th point q i The next point q of (j) i+1 (j) and the previous point q i-1 The coordinate difference of (j), i.e., q i+1 (j)-2q i (j)+q i-1 (j), then calculate the modulus of the coordinate difference, and divide the modulus by the square of the endogenous scale to obtain the parallel seam C. j The i-th point q i The discrete curvature of (j) is used to obtain the parallel seam C. j Discrete curvature sequence k i(j) Specifically, when calculating the discrete curvature sequence of each parallel seam, the first and last boundary points i=1 and i=N are excluded to reduce the error in boundary curvature calculation. The discrete curvature sequence of the parallel seam is a quantitative description of the curvature distribution of the parallel seam. It is the core basic data for calculating the constant curvature closed seam index, verifying its normal invariance, and realizing the objective identification of the circumferential weld. Calculate C for each parallel seam j Discrete curvature sequence k i (j) is the mean curvature and standard deviation of curvature. Based on the mean curvature and standard deviation of curvature, C is calculated for each parallel seam. j The constant curvature closure seam index is calculated as follows: When parallel seam C j When the corresponding average curvature is greater than 0, subtract the parallel seam C from the value of 1. j The difference obtained by dividing the standard deviation of curvature by the mean curvature is denoted as parallel seam C. j The constant curvature closure index is denoted as HCCI(j); When parallel seam C j When the corresponding average curvature is ≤0, the parallel seam is determined to be a non-circular feature, and HCCI(j) is recorded as an outlier. Extract the average curvature of the original seam C0 in the parallel seam family and label it as u0, which will be used to estimate the radius of the circumferential weld in subsequent steps; The constant curvature closure seam index HCCI(j) is constructed as a constant curvature closure seam index quintuple [HCCI(-2),HCCI(-1),HCCI(0),HCCI(+1),HCCI(+2)]; Specifically, the constant curvature closure index HCCI(j) measures the weld seam line, i.e., measures the parallel seam C. j The constant curvature closure index is an indicator of curvature constancy along the arc length. The closer the constant curvature closure index is to 1, the stronger the curvature constancy of the weld. For example, the constant curvature closure index of an ideal circumferential weld approaches 1. If the constant curvature closure index is low, it indicates that the curvature fluctuates significantly with the arc length, and the weld exhibits significant non-circular characteristics. This provides a quantifiable geometric basis for subsequent determination of circumferential welds. The calculation process of the constant curvature closure index is based on the statistical characteristics of curvature. The average curvature reflects the overall level of curvature, and the standard deviation of curvature reflects the degree of curvature fluctuation. The ratio of the two can intuitively reflect the curvature constancy. Subtracting this ratio from 1 makes the index closer to 1, representing stronger constancy, which aligns with the geometric characteristics of circumferential welds. When the parallel seam C... jWhen the average curvature is ≤0, it indicates that the curvature distribution does not conform to the characteristic that the curvature of a ring weld is positive and approximately constant. This type of weld belongs to non-ring characteristics, that is, it does not have the geometric property of constant curvature of a ring weld. At this time, the calculation logic of the constant curvature closure seam index fails, so it is recorded as an outlier, which means that the constant curvature closure seam index of the weld has no effective judgment significance. This is because the curvature of a non-ring weld may fluctuate, or even have negative curvature. Its geometric shape deviates from the ring characteristics, resulting in an average curvature of ≤0, and it is impossible to effectively measure its ring degree through the constant curvature closure seam index. Define an invariant Boolean value to record the determination result of the constant curvature closure joint index quintuple, as follows: If HCCI(-2)=HCCI(-1)=HCCI(0)=HCCI(+1)=HCCI(+2), the invariant Boolean value is true; otherwise, it is false. Specifically, the invariance Boolean value is a logical value used to record the judgment result of the constant curvature closure seam index quintuple. If these five HCCI(j) values ​​are equal, the invariance Boolean value is determined to be true; otherwise, it is not true. Since the curvature of annular welds is approximately constant, the constant curvature closure seam index should remain stable under small normal perturbations, i.e., normal offset of the five parallel seams. However, the constant curvature closure seam index of non-annular welds will show significant differences due to curvature fluctuations with normal offset. Based on this, the constant curvature closure seam index quintuple is judged. Through the consistency verification of the constant curvature closure seam index in multiple instances, the determination of annular welds without human intervention is achieved. When the invariance Boolean value is true, it indicates that the curvature of the weld is consistent under normal perturbation and can be judged as an annular weld; otherwise, it is a non-annular weld. This provides a clear logical basis for subsequent identification and ensures the objectivity and reliability of weld information identification.

[0021] Step 4: Obtain the invariant Boolean value. If the invariant Boolean value is true, then the weld is determined to be a constant curvature closed circumferential weld. If the invariant Boolean value is not true, the weld is determined to be a non-circular weld, and this method stops processing and is transferred to the existing rules. When the weld is determined to be a closed circumferential weld with constant curvature, the estimated radius of the weld is calculated as follows: The value obtained by dividing the numerical value 1 by the average curvature u0 of the original weld C0 is the estimated radius of the weld. Specifically, the estimated radius is calculated based on the geometric relationship between curvature and radius, and the reciprocal of the average curvature is the approximate radius. The estimated radius, total arc length, principal plane normal vector, and original constant curvature closure index HCCI(0) of the welds that are determined to be constant curvature closed circumferential welds are summarized into a structured parameter package; Specifically, the weld is identified and judged based on invariant Boolean values, thereby outputting weld identification results and structured parameter packages, ensuring the consistency and traceability of weld information identification. At the same time, the estimated radius of the weld provides key geometric parameters for process planning at the production end.

[0022] Furthermore, refer to Figure 2 As shown, a drawing weld information recognition system is proposed to implement a drawing weld information recognition method as described in any of the above claims, including: The 3D to equal arc length module exports the 3D coordinate set of the weld seam line from the drawing or model, forming a spatial weld polyline point sequence. It uses the least squares method to fit the principal plane where the point sequence is located and calculates the normal vector. It orthogonally projects the 3D points along the normal vector to obtain a 2D point sequence. It calculates the Euclidean distance between adjacent points in the 2D point sequence and accumulates them to obtain the total arc length of the weld. After calculating the intrinsic scale, it uses the intrinsic scale interpolation to generate a 2D equal arc length point sequence. The parallel seam family construction module calculates the tangent vector for a two-dimensional equal arc length point sequence using the center difference for the middle point and the front and rear differences for the first and last points. After rotating the tangent vector counterclockwise by 90°, the normal direction vector is obtained. The endogenous scale is offset by different multiples along the normal and the opposite direction to generate five closed parallel seams, which are combined into a parallel seam family. The curvature index determination module uses the second-order central difference method to calculate the discrete curvature sequence of each parallel seam, calculates the mean and standard deviation of curvature to obtain the constant curvature closed seam index, constructs a constant curvature closed seam index quintuple, determines whether the values ​​in the quintuple are equal, and defines an invariant Boolean value to record the determination result. The weld identification and determination module determines the weld type based on invariant Boolean values. When the invariant Boolean value is true, the weld is determined to be a constant curvature closed circumferential weld. The estimated radius of the constant curvature closed circumferential weld is calculated and summarized into a structured parameter package.

[0023] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for identifying weld information on drawings, characterized in that, include: The three-dimensional coordinate set of the weld line is marked as a spatial weld polyline point sequence. The principal plane of the spatial weld polyline point sequence is fitted by the least squares method and the normal vector of the principal plane is calculated. The elements in the spatial weld polyline point sequence are orthogonally projected onto the principal plane along the direction of the normal vector of the principal plane to obtain a two-dimensional point sequence. The total arc length of the weld, which reflects the total length of the weld, is obtained by calculating the Euclidean distance between two adjacent two-dimensional coordinate points in the two-dimensional point sequence. Based on the total number of elements in the spatial weld polyline point sequence and the total arc length of the weld, the endogenous scale reflecting the unified scale reference is calculated. Then, a two-dimensional equal arc length point sequence is generated by interpolation of the endogenous scale. For each point in the two-dimensional equal arc length point sequence, the tangent direction of the curve is calculated by the adjacent point difference method, and the normal direction vector is derived. Using the endogenous scale as the offset scale, the two-dimensional equal arc length point sequence is offset point by point along the direction of the normal direction vector to construct five closed parallel seams, and the five closed parallel seams are combined into a parallel seam family. For each parallel seam in the parallel seam family, the discrete curvature sequence of each parallel seam is calculated using the second-order central difference method. The mean curvature and standard deviation of the discrete curvature sequence are calculated. Based on the mean curvature and standard deviation of curvature, the constant curvature closure seam index of each parallel seam is calculated. The constant curvature closure seam index of each parallel seam is evaluated for equivalence, and the evaluation result is marked as an invariant Boolean value. The constant curvature closure seam index is an index that measures the constancy of the curvature of the weld line along the arc length direction.

2. The method for identifying weld information on drawings according to claim 1, characterized in that, The elements in the polyline point sequence of the spatial weld are three-dimensional coordinate points, and the total number of elements N in the polyline point sequence of the spatial weld is greater than or equal to 5.

3. The method for identifying weld information on drawings according to claim 2, characterized in that, The methods for calculating the principal plane normal vector include: Assuming the equation of the principal plane is ax + by + cz + d = 0, the coefficients a, b, and c of the plane equation are obtained by minimizing the sum of squared distances from all elements in the polyline point sequence of the spatial weld to the plane. The vector formed by the coefficients a, b, and c is the normal vector of the principal plane. The principal plane is the plane in which the weld line mainly extends in three-dimensional space.

4. The method for identifying weld information on drawings according to claim 3, characterized in that, The calculation methods for the total arc length of the weld include: Each three-dimensional coordinate point in the polyline point sequence of the spatial weld is orthogonally projected onto the principal plane along the direction of the normal vector of the principal plane to obtain the two-dimensional coordinate point corresponding to each three-dimensional coordinate point. After all the two-dimensional coordinate points are combined into a set, a two-dimensional point sequence is obtained. The Euclidean distance between two adjacent two-dimensional coordinate points in the two-dimensional point column is calculated sequentially, and the total arc length of the weld corresponding to the two-dimensional point column is obtained by summing all the calculated Euclidean distances.

5. The method for identifying weld information on drawings according to claim 4, characterized in that, Methods for generating a two-dimensional point column of equal arc length include: Subtract 1 from the total number of elements N in the polyline point array of the spatial weld to obtain the total number of intervals between adjacent elements as N-1. Divide the total arc length of the weld by (N-1) and mark the value obtained as the endogenous scale. The first point in the two-dimensional point sequence is selected as the first point of the new point sequence. Starting from the first point of the new point sequence, new points are determined by linear interpolation with the endogenous scale as the step size. This process continues until the total number of selected points reaches N. Finally, a new point sequence is obtained in which the arc lengths of adjacent points are all at the endogenous scale. This new point sequence is marked as a two-dimensional equal arc length point sequence.

6. The method for identifying weld information on drawings according to claim 5, characterized in that, The derivation of the normal direction vector includes: For each point q in a two-dimensional array of points with equal arc lengths i If 2≤i≤N-1, the tangent direction vector t can be calculated by performing central difference through adjacent points symmetrical to the arc length. i =q i+1 -q i-1 ; If i=1, the tangent direction vector t1=q2-q1 is calculated by performing forward difference through the adjacent points on the right. If i=N, the tangent direction vector t is calculated by performing backward difference through the left adjacent points. N =q N -q N-1 ; The above tangent direction vector t i Rotate counterclockwise by 90° to convert into the normal direction vector n i Output the normal direction vector n of each point in a two-dimensional array of points with equal arc lengths. i .

7. The method for identifying weld information on drawings according to claim 6, characterized in that, Parallel seam family, including: Using the endogenous scale as the offset scale, along the normal direction vector n i By shifting each point of a two-dimensional array of points of equal arc length point by point, five closed parallel seams are constructed, as follows: Parallel seam C -2 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of the endogenous scale to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C -1 For each point q in a two-dimensional array of points with equal arc lengths i Along the opposite direction of the normal direction vector -n i After offsetting by a factor of one endogenous scale to obtain a new point sequence, fine-tune the endpoints to close it; Original seam C0: The two-dimensional equal arc length point column itself, with an offset of 0; Parallel seam C +1 For each point q in a two-dimensional array of points with equal arc lengths i After offsetting by one time the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. Parallel seam C +2 For each point q in a two-dimensional array of points with equal arc lengths i After offsetting by twice the endogenous scale along the direction of the normal direction vector to obtain a new point sequence, the endpoints are finely adjusted to close the sequence. The above five closed parallel seams are combined into a parallel seam family {C} -2 C -1 C0, C +1 C +2 } 8. The method for identifying weld information on drawings according to claim 7, characterized in that, include: For each parallel seam C in the parallel seam family j j={-2, -1, 0, +1, +2}, the second-order central difference method is used to calculate C for each parallel seam. j The discrete curvature sequence is as follows: For parallel seam C j The i-th point q i (j), where 2≤i≤N-1, calculate [q i+1 (j)-2q i (j)+q i-1 The modulus length of (j) is obtained by dividing the modulus length by the square of the endogenous scale to obtain the parallel seam C. j The i-th point q i The discrete curvature of (j) is used to obtain the parallel seam C by summing the discrete curvatures. j Discrete curvature sequence k i (j); Calculate C for each parallel seam j Discrete curvature sequence k i (j) is the mean curvature and standard deviation of curvature. Based on the mean curvature and standard deviation of curvature, C is calculated for each parallel seam. j The constant curvature closure seam index is calculated as follows: When parallel seam C j When the corresponding average curvature is greater than 0, subtract the parallel seam C from the value of 1. j The difference obtained by dividing the standard deviation of curvature by the mean curvature is denoted as parallel seam C. j The constant curvature closure index is denoted as HCCI(j); When parallel seam C j When the corresponding average curvature is ≤0, the parallel seam is determined to be a non-circular feature, and HCCI(j) is recorded as an outlier. Extract the average curvature of the original seam C0 in the parallel seam family and label it as u0; The constant curvature closure seam index HCCI(j) is constructed as a constant curvature closure seam index quintuple [HCCI(-2),HCCI(-1),HCCI(0),HCCI(+1),HCCI(+2)]; If HCCI(-2)=HCCI(-1)=HCCI(0)=HCCI(+1)=HCCI(+2), define an invariant Boolean value and determine if the invariant Boolean value is true; otherwise, it is false.

9. The method for identifying weld information on drawings according to claim 8, characterized in that, include: If the invariant Boolean value is obtained and the invariant Boolean value is true, then the weld is determined to be a closed circumferential weld with constant curvature. If the invariant Boolean value is not true, then the weld is determined to be a non-circular weld. When the weld is determined to be a closed circumferential weld with constant curvature, the estimated radius of the weld is calculated as follows: The value obtained by dividing the numerical value 1 by the average curvature u0 of the original weld C0 is the estimated radius of the weld. The estimated radius, total arc length, principal plane normal vector, and original constant curvature closure index HCCI(0) of the welds that are determined to be constant curvature closed circumferential welds are summarized into a structured parameter package.

10. A drawing weld information recognition system, used to implement the drawing weld information recognition method as described in any one of claims 1-9, characterized in that, include: The 3D to equal arc length module exports the 3D coordinate set of the weld seam line from the drawing or model, forming a spatial weld polyline point sequence. It uses the least squares method to fit the principal plane where the point sequence is located and calculates the normal vector. It orthogonally projects the 3D points along the normal vector to obtain a 2D point sequence. It calculates the Euclidean distance between adjacent points in the 2D point sequence and accumulates them to obtain the total arc length of the weld. After calculating the intrinsic scale, it uses the intrinsic scale interpolation to generate a 2D equal arc length point sequence. The parallel seam family construction module calculates the tangent vector for a two-dimensional equal arc length point sequence using the center difference for the middle point and the front and rear differences for the first and last points. After rotating the tangent vector counterclockwise by 90°, the normal direction vector is obtained. The endogenous scale is offset by different multiples along the normal and the opposite direction to generate five closed parallel seams, which are combined into a parallel seam family. The curvature index determination module uses the second-order central difference method to calculate the discrete curvature sequence of each parallel seam, calculates the mean and standard deviation of curvature to obtain the constant curvature closed seam index, constructs a constant curvature closed seam index quintuple, determines whether the values ​​in the quintuple are equal, and defines an invariant Boolean value to record the determination result. The weld identification and determination module determines the weld type based on invariant Boolean values. When the invariant Boolean value is true, the weld is determined to be a constant curvature closed circumferential weld. The estimated radius of the constant curvature closed circumferential weld is calculated and summarized into a structured parameter package.

Images