A method and system for judging the bending accuracy of sheet metal parts based on three-dimensional model

Abstract

The present invention relates to the technical field of sheet metal bending detection, specifically a method and system for judging the bending accuracy of sheet metal parts based on a three-dimensional model, including: determining error compensation parameters through historical data of a target production line, calculating bending simulation adjustment parameters in combination with sheet metal design drawings, expected processes and bending simulation parameters; generating a three-dimensional model of a sheet metal finished product using three-dimensional modeling software according to the design drawings, expected processes and bending simulation adjustment parameters, extracting its target unfolding diagram and target panoramic reference diagram; performing bending accuracy detection on them respectively, obtaining the deviation results of the unfolding diagram and the deviation results of the panoramic reference diagram; if the deviation results are all qualified, the bending accuracy is qualified; otherwise, adjusting the bending simulation parameters and the design drawings, and repeating the bending simulation adjustment parameter calculation, three-dimensional modeling and bending accuracy detection until the deviation results are all qualified. Thus, the actual production conditions are met and the bending accuracy of sheet metal parts is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sheet metal part bending detection, and in particular to a method and system for judging the bending accuracy of a sheet metal part based on a three-dimensional model. Background Art

[0002] The bending process is a key step in the sheet metal manufacturing process, and its accuracy directly affects the quality, function and assembly effect of the final product. Since sheet metal parts are widely used in aerospace, automobile manufacturing, home appliances and electronic equipment, these fields have extremely strict requirements for high precision and high consistency of products, so the requirements for sheet metal bending accuracy are also increasing. However, due to the complexity of the bending process and the involvement of multiple variables such as material properties, equipment status, process parameters and environmental factors, errors are prone to occur in the actual processing process.

[0003] Existing 3D modeling and bending simulation technologies are usually based on idealized assumptions, such as assuming that the equipment is not aged, the mold alignment is completely accurate, and the material properties are uniform and stable. However, in actual production, there are often many unavoidable factors that lead to processing deviations. For example, the aging of the production line, the rebound effect of the material, the pressure fluctuations during machine tool operation, the wear of the mold, and the interference problems that may arise under complex geometric designs may all affect the bending accuracy. Especially in multiple bending operations, the cumulative error can easily be further amplified, causing the size and shape of the finished product to deviate from the design requirements or even exceed the tolerance range. These problems have caused the existing bending simulation technology based on 3D modeling to be limited in the judgment of the actual bending accuracy of sheet metal parts, and it is difficult to meet the requirements of high-precision bending.

[0004] Therefore, a method and system for judging the bending accuracy of sheet metal parts based on three-dimensional models are proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for judging the bending accuracy of sheet metal parts based on a three-dimensional model. First, the target production line is determined and its historical production data is obtained. The error compensation parameters of the production line are determined based on the historical data; the bending simulation adjustment parameters are calculated in combination with the design drawings, expected processes and bending simulation parameters of the sheet metal parts to be produced; the sheet metal design and bending simulation are completed through three-dimensional modeling software, and a three-dimensional model of the sheet metal finished product and its unfolding diagram and panoramic reference diagram are generated; the bending accuracy is detected according to the target unfolding diagram and the target panoramic reference diagram. If the detection result is unqualified, the bending simulation parameters and design drawings are adjusted and re-simulated until the accuracy requirements are met. This method can improve the bending accuracy of sheet metal parts and reduce production errors through simulation verification and deviation detection before mass production.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for judging the bending accuracy of a sheet metal part based on a three-dimensional model, comprising:

[0008] Step 1: Determine the production line for producing the sheet metal parts to be produced and obtain the target production line;

[0009] Step 2: Obtain historical sheet metal production data of the target production line;

[0010] Step 3: Determine production line error compensation parameters based on the historical sheet metal production data;

[0011] Step 4: Obtain the design drawings, expected processes and bending simulation parameters of the sheet metal part to be produced;

[0012] Step 5: Calculating bending simulation adjustment parameters according to the bending simulation parameters and the production line error compensation parameters;

[0013] Step 6: According to the design drawing, the expected process and the bending simulation adjustment parameters, the sheet metal design and bending simulation are completed using 3D modeling software to obtain a 3D model of the finished sheet metal product; an unfolded view and a panoramic reference view of the 3D model of the finished sheet metal product are obtained, which are recorded as a target unfolded view and a target panoramic reference view, respectively;

[0014] Step seven: According to the target unfolded view and the target panoramic reference view, respectively perform bending accuracy detection on the three-dimensional model of the sheet metal product to obtain the unfolded view deviation result and the panoramic reference view deviation result; if both the unfolded view deviation result and the panoramic reference view deviation result are qualified, then the bending accuracy of the sheet metal to be produced is judged to be qualified; otherwise, adjust the bending simulation parameters and the design drawing, and repeat steps four to six until both the unfolded view deviation result and the panoramic reference view deviation result are qualified.

[0015] Furthermore, the historical sheet metal production data is specifically: the historical sheet metal production data includes historical process parameters, historical environmental data, historical production line equipment operation data, historical material performance data, historical actual detection data and historical simulation target data.

[0016] Furthermore, the production line error compensation parameters include rebound compensation parameters, size compensation parameters, pressure compensation parameters, environment correction parameters and equipment compensation parameters.

[0017] Further, determining the error compensation parameters of the production line according to the historical sheet metal production data includes: associating and matching the historical actual detection data and the historical simulation target data one by one, and calculating the historical actual error value, wherein the historical actual error value includes a rebound parameter error, a dimensional parameter error, a pressure parameter error, an environmental parameter error, and an equipment parameter error;

[0018] Associating and aligning the historical actual error values, historical process parameters, historical environmental data, historical production line equipment operation data, and historical material performance data according to timestamps and production batches to construct an aligned data set;

[0019] Performing feature engineering on the aligned data set to obtain a feature data set;

[0020] The feature data set is used as the input state of the reinforcement learning model, and a reward feedback signal is constructed according to the historical actual error value, wherein at time step Define instant rewards for:

[0021] ;

[0022] in, represents the natural exponential function, is a marker variable used to index different error types, is the springback parameter error, is the size parameter error, is the pressure parameter error, is the environmental parameter error, is the device parameter error; is the time step Next The absolute error value of each error type; For the A weight coefficient of the error type;

[0023] The policy network is trained offline using the proximal policy optimization algorithm, with maximizing the expected discounted cumulative reward as the optimization goal; the formula for maximizing the expected discounted cumulative reward is:

[0024] ;

[0025] in, is the policy function, is the discount factor, is the time step Length of rounds; is the optimization objective;

[0026] Through multiple rounds of iterative updates and hyperparameter tuning, when the preset training termination conditions are met, a trained PPO model is obtained;

[0027] Obtaining expected standard process parameters, expected standard environmental data, expected standard production line equipment operation data, and expected standard material performance data of the sheet metal parts to be produced, and obtaining a test set after preprocessing and feature engineering processing;

[0028] The test set is input into the trained PPO model to obtain the production line error compensation parameters.

[0029] Furthermore, the bending simulation parameters include bending coefficient, K factor, bending deduction, bending radius, sheet metal thickness, automatic release groove cutting and reverse parameters.

[0030] Furthermore, the target panoramic reference image includes a front view, a top view and a side view of the three-dimensional model of the sheet metal finished product.

[0031] Furthermore, the bending accuracy test is performed on the three-dimensional model of the sheet metal product to obtain the deviation result of the unfolded view, including:

[0032] Extract the geometric features of key geometric points, contour boundaries and bending lines in the standard unfolded diagram to obtain standard geometric feature data; define the allowable error range of each of the key geometric points, the contour boundaries and the bending lines respectively; construct an error-allowed unfolded diagram according to a geometric extension method, the error-allowed unfolded diagram is generated by extending the standard geometric feature data and is used to limit the allowable error range of the target unfolded diagram; compare the target unfolded diagram with the error-allowed unfolded diagram to determine whether the target unfolded diagram is completely within the range of the error-allowed unfolded diagram; if the judgment result is yes, the unfolded diagram deviation result is qualified; if the judgment result is no, the unfolded diagram deviation result is unqualified.

[0033] Furthermore, the bending accuracy test is performed on the three-dimensional model of the sheet metal product to obtain the deviation result of the panoramic reference map, including:

[0034] Extracting geometric features of the standard panoramic reference image, including geometric features of key points in the front view, top view and side view, geometric features of boundary lines and geometric features of characteristic surfaces, to obtain standard panoramic geometric feature data;

[0035] Respectively define a coordinate offset range of each key point in the standard panoramic reference image, an offset distance range of each boundary line, and a shape error range of the feature surface, and construct an error-allowed panoramic reference image according to the standard panoramic reference image, the coordinate offset range, the offset distance range, and the shape error range;

[0036] The target panoramic reference image is compared with the corresponding error-allowed panoramic reference image; whether the front view, the top view and the side view of the target panoramic reference image are all within the range of the corresponding error-allowed panoramic reference image; if the judgment result is yes, the panoramic reference image deviation result is qualified; if the judgment result is no, the panoramic reference image deviation result is unqualified.

[0037] Furthermore, the step six also includes: after obtaining the three-dimensional model of the sheet metal part product, performing interference detection on the three-dimensional model of the sheet metal part product, if the interference detection passes, obtaining the target unfolded view and the target panoramic reference view; if the interference detection fails, adjusting the order of the expected processes and re-executing step six until the interference detection passes.

[0038] A system for judging the bending accuracy of sheet metal parts based on a three-dimensional model, comprising: a production line data acquisition module, for determining a production line for producing sheet metal parts to be produced, and obtaining a target production line; obtaining historical sheet metal production data of the target production line;

[0039] A production line error compensation parameter acquisition module, used to determine the production line error compensation parameters according to the historical sheet metal production data;

[0040] A parameter acquisition module, used to acquire the design drawings, expected processes and bending simulation parameters of the sheet metal part to be produced;

[0041] A bending simulation adjustment parameter calculation module, used to calculate the bending simulation adjustment parameter according to the bending simulation parameter and the production line error compensation parameter;

[0042] A three-dimensional modeling and bending simulation module is used to complete the sheet metal design and bending simulation using three-dimensional modeling software according to the design drawings, the expected process and the bending simulation adjustment parameters, and obtain a three-dimensional model of the finished sheet metal part;

[0043] The bending accuracy detection module is used to obtain the unfolded view and the panoramic reference view of the three-dimensional model of the sheet metal product, which are recorded as the target unfolded view and the target panoramic reference view respectively; according to the target unfolded view and the target panoramic reference view, the bending accuracy detection is performed on the three-dimensional model of the sheet metal product respectively to obtain the unfolded view deviation result and the panoramic reference view deviation result; if the unfolded view deviation result and the panoramic reference view deviation result are both qualified, the bending accuracy of the sheet metal to be produced is judged to be qualified; otherwise, the bending simulation parameters and the design drawing are adjusted, and the process of the bending simulation adjustment parameter calculation module, the three-dimensional modeling and bending simulation module and the bending accuracy detection module is repeated until the unfolded view deviation result and the panoramic reference view deviation result are both qualified.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention breaks through the limitations of traditional reliance on theoretical simulation ideal values ​​by introducing an error compensation parameter optimization method based on reinforcement learning. Compared with static compensation based only on theoretical models, the present invention dynamically generates accurate error compensation parameters by combining the historical production data of the target production line, which fully reflects the characteristics and deviations of the production line in the actual processing process. Through the reinforcement learning model, multi-dimensional data such as historical actual error values, process parameters, and equipment operation data are integrated into the training process to optimize the weight distribution of different error types and achieve personalized compensation adjustments for the target production line. This method not only improves the matching degree between the simulation and the actual processing results, solves the problem that theoretical simulation ignores the actual equipment, environment and process differences, but also greatly reduces the trial and error links and error accumulation, significantly improves the accuracy and production efficiency of sheet metal processing, and is suitable for high-precision sheet metal manufacturing needs under complex production line conditions.

[0046] 2. The present invention realizes a comprehensive detection of the bending accuracy of sheet metal parts by constructing an error-allowed expansion diagram and combining it with the geometric feature comparison of the target expansion diagram. Compared with a single dimension measurement or local detection method, the present invention dynamically constructs an error-allowed expansion diagram that reflects the error range of the finished product by extracting the geometric features of key geometric points, contour boundaries and bending lines, thereby achieving global coverage and avoiding detection blind spots; the error range is accurately defined using a geometric expansion method, greatly improving detection accuracy; at the same time, the method can dynamically adjust the error range according to standard geometric features, adapt to different processes and design requirements, and overcome the limitations of traditional static detection; relying on automated geometric feature comparison technology, it can efficiently determine whether the deviation results are qualified, significantly reduce trial and error costs, and improve production efficiency. Through a comprehensive, accurate and efficient detection method, the bending accuracy of sheet metal parts can be effectively improved.

[0047] 3. The present invention constructs a set of accurate evaluation systems from local geometric features to overall structural features by performing bending accuracy detection on the three-dimensional model of the finished sheet metal part, integrating the deviation detection of the unfolded view and the deviation detection of the panoramic reference view. By extracting the geometric features of the key geometric points, boundary lines and feature surfaces of the standard panoramic reference view and the unfolded view, and combining the coordinate offset range, boundary line offset range and feature surface shape error range, the error-allowed unfolded view and the error-allowed panoramic reference view are dynamically constructed, and the target unfolded view and the target panoramic reference view are compared and comprehensively evaluated one by one. This method ensures the accuracy of local geometric deviation detection, while providing comprehensive coverage of overall structural errors, effectively improving the reliability and pertinence of detection. By coupling the dual detection process of the unfolded view and the panoramic reference view, the present invention significantly reduces the deviation problem missed by a single detection method, which helps to comprehensively improve the geometric accuracy and overall precision control capabilities of sheet metal bending processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1A flow chart of a method for determining the bending accuracy of a sheet metal part based on a three-dimensional model provided by an embodiment of the present invention;

[0049] Figure 2 A schematic example of a standard development diagram provided for an embodiment of the present invention;

[0050] Figure 3 A schematic example of an error tolerance expansion diagram provided in an embodiment of the present invention;

[0051] Figure 4 A structural diagram of a system for determining the bending accuracy of a sheet metal part based on a three-dimensional model provided in an embodiment of the present invention.

[0052] In the figure: 1. Standard boundary; 2. Standard broken line; 3. First standard hole position; 4. Second standard hole position; 5. Third standard hole position; 6. Allowable error range of standard boundary; 7. Allowable error range of first standard hole position; 8. Allowable error range of second standard hole position; 9. Allowable error range of third standard hole position. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] See also Figures 1 to 4 The present invention provides a method and system for judging the bending accuracy of sheet metal parts based on a three-dimensional model. The technical solution is as follows:

[0055] Embodiment 1

[0056] A machinery manufacturing company A has several production lines with different working conditions. In order to ensure that the bending accuracy requirements of a sheet metal part can be met before mass production and reduce production errors and trial and error costs, the company adopted a method based on a three-dimensional model to judge the bending accuracy of sheet metal parts. Figure 1 , the method comprises the following steps:

[0057] See also Figure 1 S10 in step 1: determine the production line for producing the sheet metal parts to be produced and obtain the target production line; the target production line refers to a collection of specific processing equipment and process configurations used to produce the batch of sheet metal products, including machine tools, bending equipment, molds, transmission systems and set process parameters involved in the production process.

[0058] See also Figure 1S20 in step 2: obtaining historical sheet metal production data of the target production line; the historical sheet metal production data includes historical process parameters, historical environmental data, historical production line equipment operation data, historical material performance data, historical actual detection data and historical simulation target data; wherein the historical actual detection data includes: historical actual rebound parameters, historical actual size parameters, historical actual pressure parameters, historical actual environmental parameters and historical actual equipment parameters; the historical simulation target data includes historical simulation target rebound parameters, historical simulation target size parameters, historical simulation target pressure parameters, historical simulation target environmental parameters and historical simulation target equipment parameters.

[0059] By comprehensively acquiring the historical sheet metal production data of the target production line, this method can systematically cover the key factors that affect the bending accuracy in the production process, including process, environment, equipment, material properties, and detection and simulation target data. Through the detailed classification and one-to-one correspondence of historical actual detection data and historical simulation target data, the sources and distribution patterns of errors in actual production can be accurately identified, laying the foundation for generating high-precision error compensation parameters. This data-driven approach avoids the limitations of traditional reliance on a single ideal parameter model, greatly improves the matching degree between bending simulation and actual production, effectively reduces trial and error costs and rework risks, and enhances the adaptability of the target production line under complex working conditions, thereby achieving high-precision and high-consistency control of sheet metal processing.

[0060] See also Figure 1 S30 in step three: determine the error compensation parameters of the production line according to the historical sheet metal production data; the error compensation parameters of the production line include springback compensation parameters, size compensation parameters, pressure compensation parameters, environmental correction parameters and equipment compensation parameters. By introducing springback compensation parameters, size compensation parameters, pressure compensation parameters, environmental correction parameters and equipment compensation parameters, the main sources of errors affecting the bending accuracy of sheet metal parts can be fully covered, thereby achieving accurate compensation for the error characteristics of the target production line. This multi-dimensional compensation design makes the bending simulation parameters more in line with the actual processing conditions, makes up for the deficiency of ignoring the actual working conditions of the production line in the traditional method, and effectively improves the matching degree between the bending simulation and the actual processing. Through the precise adjustment of the compensation parameters, the trial and error and rework links are significantly reduced, which not only improves the bending accuracy and consistency control ability, but also enhances the adaptability of the production line to complex working conditions, providing a reliable technical guarantee for high-precision sheet metal processing.

[0061] Further, determining the error compensation parameters of the production line according to the historical sheet metal production data includes: associating and matching the historical actual detection data and the historical simulation target data one by one, and calculating the historical actual error value, wherein the historical actual error value includes a rebound parameter error, a dimensional parameter error, a pressure parameter error, an environmental parameter error, and an equipment parameter error;

[0062] Associating and aligning the historical actual error values, historical process parameters, historical environmental data, historical production line equipment operation data, and historical material performance data according to timestamps and production batches to construct an aligned data set;

[0063] Performing feature engineering on the aligned data set, including feature selection, feature extraction and data normalization, to obtain a feature data set;

[0064] The feature data set is used as the input state of the reinforcement learning model, and a reward feedback signal is constructed according to the historical actual error value, wherein at time step Define instant rewards for:

[0065] ;

[0066] in, represents the natural exponential function, is a marker variable used to index different error types, is the springback parameter error, is the size parameter error, is the pressure parameter error, is the environmental parameter error, is the device parameter error; is the time step Next The absolute error value of each error type; For the A weight coefficient of the error type;

[0067] The policy network is trained offline using the proximal policy optimization algorithm, with maximizing the expected discounted cumulative reward as the optimization goal; the formula for maximizing the expected discounted cumulative reward is:

[0068] ;

[0069] in, is the policy function, is the discount factor, is the time step Length of rounds; The optimization goal of accumulating rewards for the expected discount;

[0070] Through multiple rounds of iterative updates and hyperparameter tuning, when the preset training termination conditions are met, a trained PPO model is obtained;

[0071] The expected standard process parameters, expected standard environmental data, expected standard production line equipment operation data and expected standard material performance data of the sheet metal parts to be produced are obtained, and the test set is obtained through the same preprocessing and feature engineering process as the training set; specifically, these expected standard data are the real-time planned input parameters of the target production line. For example, the expected standard environmental data include the planned workshop temperature, humidity and other environmental control parameters. To ensure the consistency of the training set and the test set, the preprocessing and feature engineering of the test set include feature selection, feature extraction and data normalization, which are consistent with the feature processing methods and parameters used in the training set to ensure the adaptability of the model to the test set input and the accuracy of the prediction results. The test set is input into the trained PPO model to obtain the error compensation parameters of the production line.

[0072] The present invention adopts a reinforcement learning model, especially based on the proximal policy optimization (PPO) algorithm, and introduces an immediate reward formula and an optimization goal of expected discounted cumulative reward, which can effectively deal with the dynamic trade-off problem of multi-dimensional errors. The immediate reward formula prioritizes the optimization of the impact of key errors by weighting and nonlinear exponential processing of various error types such as rebound error, dimensional error, pressure error, etc., while compressing the cumulative impact of errors on reward signals, thereby improving the training convergence speed and optimization efficiency. The expected discounted cumulative reward goal avoids the local optimal problem by balancing short-term and long-term effects, and realizes the global optimization of bending accuracy. In addition, the model dynamically adapts to the multi-condition characteristics of the production line according to the input feature data set, generates accurate error compensation parameters, and thus greatly improves the fit between bending simulation and actual production. Overall, this method significantly reduces the errors used for subsequent bending simulation and three-dimensional modeling, and improves the accuracy and stability of sheet metal bending processing simulation.

[0073] See also Figure 1S40 in step four: obtain the design drawings, expected processes and bending simulation parameters of the sheet metal to be produced; the bending simulation parameters include bending coefficient, K factor, bending deduction, bending radius, sheet metal thickness, automatic release groove cutting and reverse parameters. The bending simulation parameters can accurately describe the geometric and physical properties involved in the sheet metal bending process, so that the bending simulation results are closer to the actual processing conditions. If these parameters are not defined, the bending simulation may deviate from the actual production situation, resulting in amplified errors in key geometric data such as unfolded length and bending angle, thereby affecting subsequent error compensation and processing decisions. By introducing these parameters, the deviation between theory and practice can be systematically reduced, the accuracy and reliability of bending simulation can be improved, and a reliable basis can be provided for subsequent simulation optimization, while reducing the risk of trial and error and rework.

[0074] See also Figure 1 S50, step five: calculating bending simulation adjustment parameters according to the bending simulation parameters and the production line error compensation parameters;

[0075] See also Figure 1 S60 in step six: according to the design drawings, the expected process and the bending simulation adjustment parameters, use SolidWorks 3D modeling software to complete the sheet metal design and bending simulation, and obtain a three-dimensional model of the finished sheet metal product; perform interference detection on the three-dimensional model of the finished sheet metal product. If the interference detection is passed, in the SolidWorks environment, the three-dimensional model of the finished sheet metal product is unfolded to generate a target unfolded view, and the target panoramic reference map is extracted, including a front view, a top view and a side view; these views can fully display the geometric characteristics, bending results and key details of the three-dimensional model of the finished sheet metal product from different angles, and provide a multi-dimensional reference for subsequent bending accuracy detection. Through these views, shape deviations, bending angle errors and overall contour offsets can be effectively identified to ensure the comprehensiveness and accuracy of the detection results, thereby providing a reliable basis for error compensation and processing adjustments. If the interference detection is not passed, the order of the expected process is adjusted and step six is ​​re-executed until the interference detection is passed. By performing interference detection after generating a 3D model of the finished sheet metal part, possible process conflicts or structural interference problems in the design model can be fully checked to ensure that the generation of the unfolded view and panoramic reference view is based on a non-interference machinable model. If the detection fails, dynamically adjusting the expected process sequence and remodeling will help optimize the process flow and improve the manufacturability of the design. Compared with the traditional trial and error method, this method can accurately identify and correct potential problems in the design stage, avoiding processing failures, precision deviations or material waste due to interference.

[0076] Specifically, if the number of times the expected process sequence is adjusted exceeds a preset threshold and still fails the interference detection, the process returns to the design stage and the design drawings and bending simulation parameters are optimized and modified again until the interference detection is passed.

[0077] See also Figure 1 S70, step seven: performing bending accuracy detection on the three-dimensional model of the sheet metal finished product according to the target unfolded view and the target panoramic reference view, respectively, to obtain an unfolded view deviation result and a panoramic reference view deviation result;

[0078] See also Figure 1 In S80, it is determined that both the developed image deviation result and the panoramic reference image deviation result are qualified. Figure 1 In S81, if the deviation result of the unfolded view and the deviation result of the panoramic reference view are both qualified, it is determined that the bending accuracy of the sheet metal part to be produced is qualified; refer to Figure 1 In S82, otherwise, adjust the bending simulation parameters and the design drawing, and repeat steps 4 to 6 until the deviation result of the unfolded view and the deviation result of the panoramic reference view are both qualified.

[0079] Furthermore, the bending accuracy test is performed on the three-dimensional model of the sheet metal product to obtain the deviation result of the unfolded view, including:

[0080] Extracting geometric features of key geometric points, geometric features of contour boundaries and geometric features of bending lines in the standard unfolded diagram to obtain standard geometric feature data, including a key point coordinate set, a boundary line set and a bending line set;

[0081] The error tolerance range of each of the key geometric points, the contour boundary and the bending line is defined respectively; the error tolerance range includes the coordinate offset range of the key geometric points, the offset distance range of the contour boundary and the offset position range of the bending line;

[0082] An error-allowed expansion diagram is constructed according to a geometric expansion method. The error-allowed expansion diagram is generated by expanding the standard geometric feature data and is used to define the error-allowed range of the target expansion diagram.

[0083] See also Figure 2 and Figure 3 , Figure 2 The standard expansion diagram provided for this embodiment, Figure 3 Allows for expanded graphs for generated errors. Figure 2 and Figure 3The rectangular solid line in the figure represents the standard boundary 1 of the standard unfolded view, that is, the standard outer contour boundary of the sheet metal part; the dotted line in the rectangular solid line area represents the standard fold line 2, and the standard fold line 2 in this embodiment does not have an error tolerance range; the circular solid line on the left is the first standard hole position 3; the circular solid line in the middle is the second standard hole position 4; the circular solid line on the right is the third standard hole position 5; the rectangular dotted line represents the standard boundary error tolerance range 6, which represents the dimensional deviation range allowed in actual processing. Figure 3 The circular dotted line on the left side is the first standard hole position error allowable range 7, indicating the error range of the first aperture; Figure 3 The circular dotted line in the middle is the second standard hole position error allowable range 8, indicating the error range of the second aperture; Figure 3 The circular dotted line on the right is the third standard hole position error allowable range 9, indicating the error range of the third aperture. In this embodiment, for the convenience of display, the error range in the figure has been appropriately enlarged, and the actual error range is small, which meets the processing accuracy requirements.

[0084] The target expansion diagram is compared with the error-allowed expansion diagram to determine whether the target expansion diagram is completely within the range of the error-allowed expansion diagram; if the judgment result is yes, the expansion diagram deviation result is qualified; if the judgment result is no, the expansion diagram deviation result is unqualified.

[0085] Specifically, the geometric feature data of the target unfolded diagram is extracted, and the target unfolded diagram is superimposed on the error-allowed unfolded diagram; a point-by-point comparison method is used to detect whether the key geometric points, contour boundaries and bending lines of the target unfolded diagram are completely within the range of the error-allowed unfolded diagram: the point-by-point comparison method specifically includes:

[0086] Key point comparison: determine whether the key points of the target expansion diagram are located within the key point expansion area of ​​the error-allowed expansion diagram;

[0087] Boundary line comparison: determine whether the contour boundary of the target expansion diagram is completely within the boundary extension area of ​​the error-allowed expansion diagram;

[0088] Bending line comparison: determine whether the bending line of the target unfolded drawing falls within the bending line allowable range of the error allowable unfolded drawing;

[0089] If all the geometric features of the target unfolded view are within the allowable error range of the unfolded view, the unfolded view deviation result is judged to be qualified; if there are geometric features exceeding the allowable error range, the unfolded view deviation result is judged to be unqualified, and the area exceeding the error is marked and the deviation data is recorded for subsequent optimization and adjustment.

[0090] By defining the error tolerance range and comparing the target unfolded diagram, this method can effectively determine whether the bending result meets expectations, avoiding the lack of accuracy and deviation omission caused by relying on a single measurement point in traditional inspection. The construction and comparison process of the error tolerance unfolded diagram significantly improves the comprehensiveness and reliability of the inspection, provides a clear basis for timely adjustment of bending simulation parameters and optimization of process design, and effectively ensures the consistency and controllability of sheet metal bending accuracy.

[0091] Furthermore, the bending accuracy test is performed on the three-dimensional model of the sheet metal product to obtain the deviation result of the panoramic reference map, including:

[0092] Extracting geometric features of the standard panoramic reference image, including geometric features of key points in the front view, top view and side view, geometric features of boundary lines and geometric features of characteristic surfaces, to obtain standard panoramic geometric feature data;

[0093] Respectively define a coordinate offset range of each key point in the standard panoramic reference image, an offset distance range of each boundary line, and a shape error range of the feature surface, and construct an error-allowed panoramic reference image according to the standard panoramic reference image, the coordinate offset range, the offset distance range, and the shape error range;

[0094] The target panoramic reference image is compared with the corresponding error-allowed panoramic reference image; whether the front view, the top view and the side view of the target panoramic reference image are all within the range of the corresponding error-allowed panoramic reference image; if the judgment result is yes, the panoramic reference image deviation result is qualified; if the judgment result is no, the panoramic reference image deviation result is unqualified.

[0095] Specifically, as a feasible method of the present invention, the geometric feature data of the standard panoramic reference image is extracted, including the coordinates of key points in the front view, the top view and the side view, the contour parameters of the boundary line and the shape data of the feature surface, so as to form the standard panoramic geometric feature data; the coordinate offset range of the key points, the offset distance range of the boundary line and the shape error range of the feature surface are respectively defined, and an error-allowed panoramic reference image is generated based on the geometric extension method to represent the error allowable range of the target panoramic reference image; the geometric feature data of the target panoramic reference image is extracted, the target panoramic reference image is superimposed with the corresponding error-allowed panoramic reference image, and it is determined one by one through geometric feature comparison whether the key points of the target panoramic reference image fall within the error allowable range, whether the boundary line is completely within the offset distance range and whether the feature surface meets the shape error range; if all the geometric features in the front view, the top view and the side view meet the error allowable range, the deviation result of the panoramic reference image is determined to be qualified; if any geometric feature exceeds the error allowable range, the deviation result of the panoramic reference image is determined to be unqualified, and the features exceeding the error and their deviation values ​​are recorded, and a visual diagram of the deviation distribution is generated at the same time for further optimization adjustment and result evaluation.

[0096] By testing the bending accuracy of the panoramic reference map of the finished three-dimensional model of the sheet metal part, extracting the geometric features of the key points, boundary lines and feature surfaces in the front view, top view and side view, and combining the error tolerance range to construct the error tolerance panoramic reference map, a comprehensive evaluation of the overall geometric accuracy of the sheet metal part can be achieved. This method ensures that each view of the target panoramic reference map is within the error tolerance range through multi-view comparison, which significantly improves the meticulousness and comprehensiveness of the detection. Compared with traditional single-view or local detection, this method can more effectively capture the overall geometric deviation and local detail error, providing a scientific basis for the accurate judgment of the bending accuracy, while reducing the accumulation of processing deviations caused by detection blind spots, laying a solid foundation for the subsequent process adjustment and optimization of compensation strategies.

[0097] In this embodiment, the standard unfolded view and the standard panoramic reference view are the two-dimensional unfolded form and three-dimensional view (including front view, top view and side view) of the ideal finished sheet metal part that meets the bending requirements and quality inspection standards generated through three-dimensional modeling and bending simulation.

[0098] In order to verify the effectiveness of the method of the present invention, four groups of experiments were conducted. The first group adopted the method of the present invention, introduced the production line error compensation parameters, performed three-dimensional modeling and bending simulation, and completed the bending accuracy detection in combination with the target expansion diagram and the target panoramic reference diagram; the second group did not introduce the production line error compensation parameters, and directly performed modeling based on the bending simulation parameters. The subsequent steps were consistent with the method of the present invention; the third group referred to Figure 1 The process shown in FIG. 1 is shown in FIG. 1 , but in step S70, only the bending accuracy is detected by the target unfolded view; the fourth group also refers to Figure 1 The process shown in FIG. 1 is the same as that shown in FIG. 1 , but in step S70 , the bending accuracy is detected only by using the target panoramic reference image.

[0099] Table 1 Comparative experimental data table

[0100]

[0101] The experimental results are shown in Table 1. Through comparative analysis of the four groups of experimental results, it can be concluded that the method of the present invention (the first group) performs best in terms of bending accuracy and unfolding size deviation control. The percentage of meeting the bending requirements reaches 98%, and the absolute values ​​of the bending angle deviation and the absolute values ​​of the key unfolding size deviation are controlled within 100%. In contrast, the percentage of the second group that did not introduce the production line error compensation parameters that met the bending requirements was significantly reduced to only 82%, and the deviations of the bending angle and unfolding size were expanded to The third and fourth groups only used the target unfolded diagram or the target panoramic reference diagram for bending detection, respectively. Although the accuracy was improved, it still could not reach the level of the first group, especially in the deviation control of bending angle and key unfolded dimensions, which was significantly lower than the first group.

[0102] The present invention combines the actual situation of a specific target production line and generates exclusive production line error compensation parameters based on its historical production data, accurately reflecting the specific characteristics of the target production line in terms of process, equipment performance and production environment. By introducing customized error compensation parameters and bending simulation adjustment parameters for the production line, the bending simulation process is highly matched with the actual processing capacity of the target production line, ensuring that the simulation results are more in line with actual production needs. On this basis, through three-dimensional modeling and deviation detection, the processing characteristics of the target production line are optimized in a targeted manner, significantly improving the accuracy and stability of sheet metal bending, reducing error accumulation and trial and error costs in production, thereby greatly improving the processing efficiency and product quality for specific production lines, and being suitable for high-precision and batch sheet metal production scenarios.

[0103] Embodiment 2

[0104] Company B is responsible for the production of sheet metal parts with complex shapes. In order to meet the requirements of high-precision bending of sheet metal parts for different parts, Company B needs to verify the feasibility and processing accuracy of the design plan before production. Due to differences in equipment, process parameters and environmental conditions of different production lines, direct production may lead to accumulation of bending errors and affect product quality. Therefore, the company adopted a system based on 3D models to judge the bending accuracy of sheet metal parts. Figure 4 , the system comprises:

[0105] The production line data acquisition module is used to determine the production line for producing the sheet metal parts to be produced, and obtain the target production line; and obtain the historical sheet metal production data of the target production line;

[0106] A production line error compensation parameter acquisition module, used to determine the production line error compensation parameters according to the historical sheet metal production data;

[0107] A parameter acquisition module, used to acquire the design drawings, expected processes and bending simulation parameters of the sheet metal part to be produced;

[0108] A bending simulation adjustment parameter calculation module, used to calculate the bending simulation adjustment parameter according to the bending simulation parameter and the production line error compensation parameter;

[0109] A three-dimensional modeling and bending simulation module is used to complete the sheet metal design and bending simulation using three-dimensional modeling software according to the design drawings, the expected process and the bending simulation adjustment parameters, and obtain a three-dimensional model of the finished sheet metal part;

[0110] The bending accuracy detection module is used to obtain the unfolded view and the panoramic reference view of the three-dimensional model of the sheet metal product, which are recorded as the target unfolded view and the target panoramic reference view respectively; according to the target unfolded view and the target panoramic reference view, the bending accuracy detection is performed on the three-dimensional model of the sheet metal product respectively to obtain the unfolded view deviation result and the panoramic reference view deviation result; if the unfolded view deviation result and the panoramic reference view deviation result are both qualified, the bending accuracy of the sheet metal to be produced is judged to be qualified; otherwise, the bending simulation parameters and the design drawing are adjusted, and the process of the bending simulation adjustment parameter calculation module, the three-dimensional modeling and bending simulation module and the bending accuracy detection module is repeated until the unfolded view deviation result and the panoramic reference view deviation result are both qualified.

[0111] This system covers all the steps mentioned in the method of embodiment 1 (such as Figure 1 As shown in the figure), the bending accuracy is gradually optimized through data transfer and feedback loops between modules to ensure that the bending accuracy of the target sheet metal part meets the expected requirements.

[0112] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for judging the bending accuracy of sheet metal parts based on a three-dimensional model, characterized in that: include: Step 1: Determine the production line for producing the sheet metal parts to be produced and obtain the target production line; Step 2: Obtain historical sheet metal production data of the target production line; Step 3: determining the error compensation parameters of the production line according to the historical sheet metal production data, including: associating and matching the historical actual detection data and the historical simulation target data one by one, and calculating the historical actual error value, wherein the historical actual error value includes the springback parameter error, the size parameter error, the pressure parameter error, the environmental parameter error and the equipment parameter error; Associating and aligning the historical actual error values, historical process parameters, historical environmental data, historical production line equipment operation data, and historical material performance data according to timestamps and production batches to construct an aligned data set; Performing feature engineering on the aligned data set to obtain a feature data set; The feature data set is used as the input state of the reinforcement learning model, and a reward feedback signal is constructed according to the historical actual error value, wherein at time step Define instant rewards for: in, represents the natural exponential function, is a marker variable used to index different error types, is the springback parameter error, is the size parameter error, is the pressure parameter error, is the environmental parameter error is the device parameter error; is the time step Next Absolute error value of each error type; For the A weight coefficient of the error type; The policy network is trained offline using the proximal policy optimization algorithm, with maximizing the expected discounted cumulative reward as the optimization goal; the formula for maximizing the expected discounted cumulative reward is: in, is the policy function, is the discount factor, is the time step Length of rounds; is the optimization objective; Through multiple rounds of iterative updates and hyperparameter tuning, when the preset training termination conditions are met, a trained PPO model is obtained; Obtaining expected standard process parameters, expected standard environmental data, expected standard production line equipment operation data, and expected standard material performance data of the sheet metal parts to be produced, and obtaining a test set after preprocessing and feature engineering processing; Inputting the test set into the trained PPO model to obtain the production line error compensation parameters; Step 4: Obtain the design drawings, expected processes and bending simulation parameters of the sheet metal part to be produced; Step 5: Calculating bending simulation adjustment parameters according to the bending simulation parameters and the production line error compensation parameters; Step 6: According to the design drawing, the expected process and the bending simulation adjustment parameters, the sheet metal design and bending simulation are completed using 3D modeling software to obtain a 3D model of the finished sheet metal product; an unfolded view and a panoramic reference view of the 3D model of the finished sheet metal product are obtained, which are recorded as a target unfolded view and a target panoramic reference view, respectively; Step seven: According to the target unfolded view and the target panoramic reference view, respectively perform bending accuracy detection on the three-dimensional model of the sheet metal product to obtain the unfolded view deviation result and the panoramic reference view deviation result; if both the unfolded view deviation result and the panoramic reference view deviation result are qualified, then the bending accuracy of the sheet metal to be produced is judged to be qualified; otherwise, adjust the bending simulation parameters and the design drawing, and repeat steps four to six until both the unfolded view deviation result and the panoramic reference view deviation result are qualified.

2. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The historical sheet metal production data specifically includes: the historical sheet metal production data includes historical process parameters, historical environmental data, historical production line equipment operation data, historical material performance data, historical actual detection data and historical simulation target data.

3. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The production line error compensation parameters include rebound compensation parameters, size compensation parameters, pressure compensation parameters, environment correction parameters and equipment compensation parameters.

4. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The bending simulation parameters include bending coefficient, K factor, bending deduction, bending radius, sheet metal thickness, automatic relief groove cutting and reverse parameters.

5. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The target panoramic reference image includes a front view, a top view and a side view of the three-dimensional model of the sheet metal finished product.

6. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The bending accuracy test is performed on the three-dimensional model of the sheet metal finished product to obtain the deviation result of the unfolded view, including: Extract the geometric features of key geometric points, contour boundaries and bending lines in the standard unfolded diagram to obtain standard geometric feature data; define the allowable error range of each of the key geometric points, the contour boundaries and the bending lines respectively; construct an error-allowed unfolded diagram according to a geometric extension method, the error-allowed unfolded diagram is generated by extending the standard geometric feature data and is used to limit the allowable error range of the target unfolded diagram; compare the target unfolded diagram with the error-allowed unfolded diagram to determine whether the target unfolded diagram is completely within the range of the error-allowed unfolded diagram; if the judgment result is yes, the unfolded diagram deviation result is qualified; if the judgment result is no, the unfolded diagram deviation result is unqualified.

7. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The bending accuracy test is performed on the three-dimensional model of the sheet metal finished product to obtain the deviation result of the panoramic reference map, including: Extracting geometric features of the standard panoramic reference image, including geometric features of key points in the front view, top view and side view, geometric features of boundary lines and geometric features of characteristic surfaces, to obtain standard panoramic geometric feature data; Respectively define a coordinate offset range of each key point in the standard panoramic reference image, an offset distance range of each boundary line, and a shape error range of the feature surface, and construct an error-allowed panoramic reference image according to the standard panoramic reference image, the coordinate offset range, the offset distance range, and the shape error range; The target panoramic reference image is compared with the corresponding error-allowed panoramic reference image; whether the front view, the top view and the side view of the target panoramic reference image are all within the range of the corresponding error-allowed panoramic reference image; if the judgment result is yes, the panoramic reference image deviation result is qualified; if the judgment result is no, the panoramic reference image deviation result is unqualified.

8. The method for judging the bending accuracy of sheet metal parts based on a three-dimensional model according to claim 1, characterized in that: The step six also includes: after obtaining the three-dimensional model of the sheet metal part product, performing interference detection on the three-dimensional model of the sheet metal part product, if the interference detection passes, obtaining the target unfolded view and the target panoramic reference view; if the interference detection fails, adjusting the order of the expected processes and re-executing step six until the interference detection passes.

9. A system for judging the bending accuracy of sheet metal parts based on a three-dimensional model, characterized in that: A method for determining the bending accuracy of a sheet metal part based on a three-dimensional model according to any one of claims 1 to 8, comprising: The production line data acquisition module is used to determine the production line for producing the sheet metal parts to be produced, and obtain the target production line; and obtain the historical sheet metal production data of the target production line; A production line error compensation parameter acquisition module, used to determine the production line error compensation parameters according to the historical sheet metal production data; A parameter acquisition module, used to acquire the design drawings, expected processes and bending simulation parameters of the sheet metal part to be produced; A bending simulation adjustment parameter calculation module, used to calculate the bending simulation adjustment parameter according to the bending simulation parameter and the production line error compensation parameter; A three-dimensional modeling and bending simulation module is used to complete the sheet metal design and bending simulation using three-dimensional modeling software according to the design drawings, the expected process and the bending simulation adjustment parameters, and obtain a three-dimensional model of the finished sheet metal part; The bending accuracy detection module is used to obtain the unfolded view and the panoramic reference view of the three-dimensional model of the sheet metal product, which are recorded as the target unfolded view and the target panoramic reference view respectively; according to the target unfolded view and the target panoramic reference view, the bending accuracy detection is performed on the three-dimensional model of the sheet metal product respectively to obtain the unfolded view deviation result and the panoramic reference view deviation result; if the unfolded view deviation result and the panoramic reference view deviation result are both qualified, the bending accuracy of the sheet metal to be produced is judged to be qualified; otherwise, the bending simulation parameters and the design drawing are adjusted, and the process of the bending simulation adjustment parameter calculation module, the three-dimensional modeling and bending simulation module and the bending accuracy detection module is repeated until the unfolded view deviation result and the panoramic reference view deviation result are both qualified.

Images