Safe body sheet metal forming processing method and system based on internet of things

Abstract

The application discloses a safe body sheet metal forming processing method and system based on the Internet of Things, and relates to the field of metal plate processing. The method comprises the following steps: obtaining the displacement-pressure curve of the single forming process of the numerical control bending machine; determining the stress response data by processing the curve in space-time truncation to analyze the bending residual stress field distribution domain; in the multi-pass welding process, the high-order current harmonic mode length of the welding power inverter is extracted through the arc sensor; the bending residual stress field distribution domain is taken as the prior physical constraint boundary, the welding seam area is dynamically locked at both ends according to the harmonic mode length change, and the upper and lower end indication areas are identified; the indication areas are traversed to analyze the deviation control scheme and obtain the target scheme; and the ultrasonic stress relief mechanism is controlled to implement local directional excitation stress relief control. The application solves the problem that the existing sheet metal cabinet stress control cannot accurately position the local high stress risk area of the welding seam and realize differentiated directional stress relief processing.

Description

Technical Field

[0001] This application relates to the field of metal sheet processing, and in particular to a method and system for forming and processing sheet metal for safes based on the Internet of Things. Background Technology

[0002] Sheet metal forming and welding are core processes in the mass production of safes. Residual stress defects within the weld seams after forming directly reduce the structural strength of the safe and shorten its lifespan. Managing these stress defects is a critical manufacturing challenge that urgently needs to be addressed in the sheet metal safe manufacturing field. Currently, the mainstream methods for managing residual stress generated by sheet metal bending and welding in the industry involve two independent processes. The bending process relies on fixed standard mold parameters to form the sheet metal, while the welding process uses constant welding current and travel speed to fuse the weld seams. After forming and welding, residual stress within the sheet metal is removed uniformly through whole-machine ultrasonic vibration. However, the existing phased, independent control and uniform stress removal processing model cannot correlate the initial residual stress generated by bending with the additional stress caused by fluctuations in welding heat input. This easily leads to multiple manufacturing defects such as excessive localized stress accumulation in weld seams, ineffective vibration in non-defective areas causing equipment damage, and insufficient stress removal in high-risk stress areas.

[0003] Currently, in related technologies, the stress coordination and management of the entire process of sheet metal cabinet bending and welding faces the technical problem of not being able to accurately locate local high-stress risk areas in the weld and achieve differentiated directional stress relief treatment. Summary of the Invention

[0004] This application provides a method and system for sheet metal forming of safes based on the Internet of Things (IoT). It collects the displacement-pressure curve of a single forming operation by a CNC bending machine through an industrial IoT sensing network, then performs spatiotemporal truncation on the curve to obtain stress response data and analyzes the residual stress field distribution domain. Subsequently, during multi-pass robotic welding of the safe, an arc sensor is used to extract the high-order harmonic modes of the power inverter in real time. Then, using the residual stress field distribution domain as a priori constraint combined with the harmonic mode length, the weld seam is dynamically locked, and two high-risk indication areas (upper and lower) are identified. The two areas are then analyzed to generate a target flexible processing scheme. Finally... The remote-controlled ultrasonic stress relief mechanism uses this solution to perform localized directional vibration stress relief treatment on two indicated areas. This solves the technical problem of the existing sheet metal cabinet bending-welding process stress coordination control, which cannot accurately locate the local high-stress risk area of ​​the weld and achieve differentiated directional stress relief treatment. It achieves the technical effects of accurately locating the local high-stress area of ​​the weld, differentiating the elimination of superimposed residual stress in defect areas, reducing equipment wear caused by indiscriminate full-area vibration, fully releasing the internal stress of high-risk points, and improving the overall structural strength of the safe and the consistency of finished product processing.

[0005] This application provides a method for sheet metal forming of safes based on the Internet of Things (IoT), comprising: using an industrial IoT sensing network at a preset acquisition frequency to acquire in real time the displacement-pressure curve of a CNC bending machine during a single forming process; performing spatiotemporal truncation processing on the displacement-pressure curve to determine stress response data, and combining the stress response data to analyze the residual stress during bending and determine the distribution domain of the residual stress field; during the multi-pass welding process of the safe body on a fully automated robotic welding workstation, using an arc sensor in the industrial IoT sensing network to extract the welding power inverter in real time. The high-order current harmonic mode length is used as the prior physical constraint boundary. Based on the real-time change of the high-order current harmonic mode length, the weld area is dynamically locked at both ends to identify the upper and lower indicator areas corresponding to the high-risk pathological processing areas. The upper and lower indicator areas are analyzed to traverse the upper and lower indicator areas to obtain the target flexible processing scheme. The ultrasonic stress relief mechanism is remotely controlled to implement local directional excitation stress relief control for the upper and lower indicator areas according to the target flexible processing scheme.

[0006] In a possible implementation, the displacement-pressure curve is subjected to spatiotemporal truncation processing to determine stress response data. Combined with the stress response data, residual bending stress is analyzed to determine the distribution domain of the residual bending stress field. The following processing is performed: using a preset standard deviation after the die cuts into the sheet metal as a boundary, the stress data of the initial and unloading segments of the displacement-pressure curve are truncated and discarded, and the stress response data of the core holding pressure segment is extracted. Based on the stress response data, springback prediction is performed to obtain the predicted value of the cabinet springback angle. The pressing depth and holding time of the CNC bending machine at the dead point position are dynamically adjusted according to the predicted value of the cabinet springback angle, mapping and constructing a distribution domain of the residual bending stress field that matches the predicted value of the cabinet springback angle.

[0007] In a possible implementation, springback prediction is performed based on the stress response data to obtain the predicted value of the cabinet springback angle, and the following processing is performed: the stress response data is traversed and nonlinear curvature fitting is performed with the standard physical stress model to obtain the material yield strength deviation value; the springback angle is calculated based on the material yield strength deviation value and the nominal thickness of the safe to obtain the predicted value of the cabinet springback angle.

[0008] In a possible implementation, the bending residual stress field distribution domain is used as a priori physical constraint boundary. Based on the real-time changes in the higher-order current harmonic mode length, the weld region is dynamically locked at both ends. The upper and lower indicator regions corresponding to the high-risk pathological processing regions are identified, and the following processing is performed: the higher-order current harmonic mode length is numerically rearranged according to the sampling time sequence to obtain a higher-order harmonic feature sequence; the weld position information covered by the bending residual stress field distribution domain is extracted, and corresponding local time-series feature segments are intercepted in the higher-order harmonic feature sequence; the local time-series feature segments are subjected to dual-end numerical filtering to obtain upper peak lattices exceeding a first preset percentile and lower trough lattices below a second preset percentile; the welding trajectory region corresponding to the upper peak lattices is locked as the upper indicator region, and the welding trajectory region corresponding to the lower trough lattices is locked as the lower indicator region.

[0009] In one possible implementation, during the multi-pass welding of the safe body by the fully automated robotic welding workstation, the high-order current harmonic mode of the welding power inverter is extracted in real time through the arc sensor in the industrial IoT sensing network. The following processing is also performed: the current spatial coordinates of the fully automated robotic welding workstation on the weld seam are tracked and locked in real time through the industrial IoT sensing network; the expected heat input variance for each weld pass during the multi-pass welding process is calculated at the current spatial coordinates along the welding time axis; when the expected heat input variance is greater than a preset consistency threshold, the pulse current of the current welding pass is adjusted according to a preset pulse current adjustment table.

[0010] In a possible implementation, the expected heat input variance for each weld pass in a multi-pass welding process is calculated at the current spatial coordinates along the welding time axis, and the following processing is performed: The multi-pass welding line energy input for each weld pass in each frame is calculated using the multi-pass welding line energy input formula, and the variance is calculated to determine the expected heat input variance for each weld pass; the multi-pass welding line energy input formula is: ;in, For the welding line energy input of the k-th pass, The welding thermal efficiency coefficient, Let be the real-time voltage of channel k. Let be the real-time current of channel k. The current welding movement speed of the fully automated robotic welding workstation when performing the k-th welding pass.

[0011] In a possible implementation, the upper and lower indicator areas are traversed to analyze the correction control scheme, obtain the target flexible manufacturing process scheme, and perform the following processing: obtain a preset local fine-grained disambiguation strategy; based on the defect types of the upper and lower indicator areas, generate multiple independent correction schemes respectively; extract the three-dimensional spatial coordinates of the upper and lower indicator areas in the safe assembly; calculate the physical stress transmission coupling matrix between adjacent welding areas based on the three-dimensional spatial coordinates to obtain the adjacent stress influence transmission parameters; use the adjacent stress influence transmission parameters to correct the excitation frequency parameters and excitation duration parameters in the multiple independent correction schemes respectively, and perform local disambiguation iteration with the optimization goal of maximizing the collaborative anti-damage strength to obtain the target flexible manufacturing process scheme that conforms to the current component spatial constraints of the safe.

[0012] In a possible implementation, the following processing is performed: For the upper indication area, the independent correction scheme includes issuing a local freeze command to freeze the current mechanical travel speed and welding power of the fully automated robotic welding workstation, and introducing the stress normal prior derived from the bending residual stress field distribution domain for local constraint fine-tuning; For the lower indication area, the independent correction scheme includes increasing the pulse thermal input current of the current welding layer.

[0013] In a possible implementation, the remotely controlled ultrasonic stress relief mechanism performs localized directional excitation stress relief control on the upper and lower indicator areas according to the target flexible processing scheme. The following processing is performed: the target flexible processing scheme is converted into control current and excitation frequency, and sent to the servo actuator of the ultrasonic stress relief mechanism; while driving the ultrasonic stress relief mechanism to perform localized directional excitation stress relief control, a parameter locking command is sent to keep the current process parameters of the fully automatic robotic welding workstation stable.

[0014] This application also provides an IoT-based sheet metal forming system for safes, including: a displacement-pressure curve acquisition module, used to acquire the displacement-pressure curve of a CNC bending machine in a single forming process in real time using an industrial IoT sensing network at a preset acquisition frequency; a bending residual stress analysis module, used to perform spatiotemporal truncation processing on the displacement-pressure curve to determine stress response data, and combine the stress response data to perform bending residual stress analysis to determine the distribution domain of the bending residual stress field; and a high-order current harmonic mode length extraction module, used to extract welding current in real time through an arc sensor in the industrial IoT sensing network during multi-pass welding of the safe body in a fully automated robotic welding workstation. The system includes: a high-order current harmonic mode of the source inverter; an indicator area identification module, used to take the bending residual stress field distribution domain as a priori physical constraint boundary, and dynamically lock the weld area at both ends according to the real-time change of the high-order current harmonic mode, respectively identifying the upper and lower indicator areas corresponding to the high-risk pathological processing area; a correction control scheme analysis module, used to traverse the upper and lower indicator areas to analyze the correction control scheme and obtain the target flexible processing scheme; and a stress relief control module, used to remotely control the ultrasonic stress relief mechanism to implement local directional excitation stress relief control for the upper and lower indicator areas according to the target flexible processing scheme.

[0015] The proposed method and system for forming sheet metal safes based on the Internet of Things (IoT) first utilizes an industrial IoT sensing network at a preset acquisition frequency to acquire the displacement-pressure curve of a CNC bending machine during a single forming process in real time. Then, the displacement-pressure curve is subjected to spatiotemporal truncation processing to determine stress response data. This stress response data is then combined with analysis of residual bending stress to determine the distribution domain of the residual bending stress field. Finally, during the multi-pass welding of the safe body at a fully automated robotic welding workstation, the arc sensor in the industrial IoT sensing network extracts the welding power inverse signal in real time. The method and system proposed in this application achieve the following technical effects: accurately locking the high-order current harmonic mode of the inverter, using the bending residual stress field distribution domain as a priori physical constraint boundary, and dynamically locking the weld area at both ends according to the real-time changes of the high-order current harmonic mode. The upper and lower indicator areas corresponding to high-risk defective processing areas are identified respectively. Then, the upper and lower indicator areas are traversed to analyze the correction control scheme and obtain the target flexible processing scheme. Finally, the ultrasonic stress relief mechanism is remotely controlled to implement local directional excitation stress relief control for the upper and lower indicator areas according to the target flexible processing scheme. Through the above process, the method and system proposed in this application achieve the following technical effects: accurately locking the local high-stress area of ​​the weld, differentially eliminating superimposed residual stress in defect areas, reducing equipment losses caused by indiscriminate global excitation, fully releasing internal stress at high-risk points, and improving the overall structural strength and finished product processing consistency of the safe. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating the IoT-based sheet metal forming process for safes provided in this application embodiment.

[0018] Figure 2 This is a schematic diagram showing the fitting comparison of the material yield strength deviation provided in the embodiments of this application.

[0019] Figure 3 This is a schematic diagram of the dual-end screening of high-order current harmonic mode length provided in the embodiments of this application.

[0020] Figure 4 A schematic diagram of the structure of the IoT-based safe sheet metal forming and processing system provided in the embodiments of this application.

[0021] Figure labeling: Displacement-pressure curve acquisition module 10, Bending residual stress analysis module 20, High-order current harmonic mode extraction module 30, Indicator area identification module 40, Correction control scheme analysis module 50, Stress relief control module 60. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0023] This application provides a method for forming and processing sheet metal for safes based on the Internet of Things, such as... Figure 1 As shown, the method includes: Step S100: Using an industrial Internet of Things (IoT) sensing network at a preset acquisition frequency, the displacement-pressure curve of the CNC bending machine during a single forming process is acquired in real time.

[0024] Specifically, this step is executed by an industrial IoT sensing and acquisition terminal deployed locally on the CNC bending machine. The communication link uses an industrial 5G private network as the carrier of the industrial IoT sensing network. The preset acquisition frequency is a configurable parameter. The general configuration rule is that the acquisition frequency value is greater than or equal to 20 times the reciprocal of the total duration of a single bending forming action. For example, an acquisition frequency of 200 times per second can be set. The execution process is as follows: First, the high-precision grating displacement sensor installed on the slider side of the CNC bending machine and the piezoelectric pressure sensor at the bottom of the mold simultaneously acquire the analog signals of slider displacement and plate pressure during the entire bending process. Then, the analog-to-digital converter chip built into the sensing and acquisition terminal converts the two analog signals into discrete digital sampling points with a unified time stamp. Each set of sampling points is bound to the displacement value and pressure value at the same acquisition time, and the combination forms the original time-series dataset of displacement-pressure curve. Finally, the original time-series dataset of the curve is pushed to the local edge computing server in real time through the industrial 5G private network for storage and forwarding.

[0025] Step S200: The displacement-pressure curve is subjected to spatiotemporal truncation processing to determine the stress response data, and the bending residual stress is analyzed in combination with the stress response data to determine the distribution domain of the bending residual stress field.

[0026] Specifically, this step is executed by an edge computing server, with the input data being the original time-series dataset of displacement-pressure curves output by the S100. The process is divided into three sequential actions. The first step involves calling a time-series truncation processing program to segment and filter the original curve dataset, removing invalid stress sampling data and outputting stress response data containing only the effective stress range. The second step uses a material nonlinear stress analysis algorithm, taking the stress response data as input, to solve for the microscopic stress distribution within the plate material, outputting a global bending residual stress numerical matrix. The third step uses the residual stress numerical matrix as a basis to match the parameters of the CNC bending machine, mapping and generating a spatially divided bending residual stress field distribution domain. This distribution domain is a three-dimensional mesh data table labeling the magnitude of residual stress corresponding to each spatial coordinate.

[0027] In one possible implementation, the displacement-pressure curve is subjected to spatiotemporal truncation processing to determine stress response data. Combined with the stress response data, residual bending stress is analyzed to determine the distribution domain of the residual bending stress field. Step S200 further includes step S210, where the stress data of the initial segment and unloading segment of the displacement-pressure curve are truncated and discarded, using a preset standard deviation after the mold cuts into the sheet metal as the boundary, and the stress response data of the core holding pressure segment is extracted. Specifically, this step is a refined execution flow of the time-series truncation processing in step S200. The input is the original time-series dataset of the displacement-pressure curve, and the preset standard deviation is a manually configured stress fluctuation judgment threshold. The general configuration rule is to take 1.5 times the standard deviation of the stress sample in the holding pressure stage of a standard qualified sheet metal. For example, the preset standard deviation value is 0.8 MPa. The execution flow first divides the original curve into three time-series intervals according to the processing sequence: the initial segment before the mold cuts into the sheet metal, the core holding pressure segment where the mold fully presses the sheet metal, and the unloading segment after the mold is lifted and unloaded. Next, the standard deviations of all stress sampling points in the initial segment and the unloading segment are calculated separately. If the standard deviation of stress in any time interval is greater than the preset standard deviation, the data in that segment is determined to be invalid interference data caused by the equipment being unloaded or the sheet not being fully attached to the mold. All sampling points in that segment are discarded. Finally, only all time sampling points in the core pressure holding segment are retained, and the interference-free stress response data is integrated and output. This step improves the calculation accuracy of residual stress analysis and springback prediction by eliminating interference data that does not actually involve sheet stress during bending.

[0028] Step S220 involves predicting springback based on the stress response data to obtain the predicted springback angle value of the cabinet. Specifically, this step is a detailed execution of the springback prediction process in step S200. The input is the stress response data output in S210, and the process is divided into two sequential execution actions. The first step calls a nonlinear curvature fitting program to numerically match and solve the stress response data with a standard physical stress model, outputting the material yield strength deviation value. The second step uses the yield strength deviation value and the nominal thickness of the safe's sheet metal as dual input parameters, substitutes them into the springback angle calculation formula to complete the quantitative calculation, and outputs the predicted value of the cabinet's springback angle.

[0029] Step S230 involves dynamically adjusting the indentation depth and holding time of the CNC bending machine at the dead point position based on the predicted springback angle of the cabinet, mapping and constructing a bending residual stress field distribution domain that matches the predicted springback angle of the cabinet. Specifically, this step is a refined execution flow of the stress field mapping in step S200. The inputs are the predicted springback angle of the cabinet output from S220 and the adjustable range data table of the CNC bending machine equipment parameters. The adjustable parameters of the equipment include two sets of parameters: dead point indentation depth and holding time. The execution flow is as follows: first, the built-in parameter mapping table of the bending machine is read. This table stores the optimal indentation depth compensation and holding time extension corresponding to different springback angle intervals. Then, the corresponding compensation parameters are matched according to the predicted springback angle and superimposed on the original baseline process parameters of the bending machine to generate dynamically adjusted indentation depth and holding time values. Subsequently, parameter adjustment commands are sent to the lower-level CNC system, and the internal stress distribution of the plate corresponding to the adjusted bending parameters is recorded simultaneously. Finally, the residual stress values ​​corresponding to the adjusted global spatial coordinates are integrated to output a complete three-dimensional mesh data table of the bending residual stress field distribution domain. This step compensates for the springback deformation of the sheet metal by bending equipment parameter compensation, and simultaneously constructs an initial stress field of sheet metal that matches the actual forming state, providing accurate prior constraints for the welding process.

[0030] In one possible implementation, springback prediction is performed based on the stress response data to obtain the predicted value of the cabinet springback angle. Step S220 further includes step S221, which involves traversing the stress response data and performing nonlinear curvature fitting with the standard physical stress model to obtain the material yield strength deviation value. Specifically, this step is a refined execution flow of the first step of S220, nonlinear curvature fitting. The input is the stress response data output from S210. The standard physical stress model adopts the standard constitutive model of low-carbon steel elastic-plastic bending, and the model has built-in standard yield strength, stress-strain curve, and curvature change reference parameters. The execution flow is as follows: first, all time-series sampling points in the stress response data are traversed, and the real-time plate bending curvature and real-time stress values ​​corresponding to each set of sampling points are extracted. Then, the least squares nonlinear fitting algorithm is used to map all measured curvature-stress data points to the reference curve of the standard physical stress model. Finally, the longitudinal stress difference between the measured data curve and the standard model curve is solved. This difference is the material yield strength deviation value. The positive and negative values ​​represent that the actual yield strength of the plate is higher or lower than the standard nominal value, respectively. This step eliminates the rebound calculation error caused by batch fluctuations in materials by quantifying the performance differences between a single cabinet panel and standard materials.

[0031] For example, four sets of continuous pressure-holding sampling data of cold-rolled steel plates for safe bodies were selected for fitting calculation. The data are shown in Table 1 and... Figure 2 As shown.

[0032] Table 1

[0033] The measured curvature and measured stress of the four sampling points in the table were extracted as the fitting input dataset. The theoretical stress of the standard model baseline curve at the corresponding curvature is shown in the third column of the table. The stress difference of a single sampling point was calculated point by point. The calculation logic is: stress difference of a single sampling point = measured stress − stress corresponding to the same curvature in the standard model. The four sets of data yielded differences of -14MPa, -17MPa, -19MPa, and -17MPa, respectively. The least squares nonlinear fitting algorithm optimized the minimum sum of the squares of the four differences. After iteratively matching the standard constitutive curve, the arithmetic mean of the stress differences of all sampling points was taken as the final material yield strength deviation value. Substituting the numerical values: (-14) + (-17) + (-19) + (-17) = -67, the total difference was divided by the number of sampling points, 4, to obtain the final material yield strength deviation value = -16.75MPa. Output judgment logic: If the yield strength deviation value is negative, it means that the actual yield strength of this batch of cold-rolled steel sheets is lower than the nominal yield strength of the material standard. This deviation value will be directly input into S222 to complete the springback angle calculation.

[0034] Step S222: Calculate the springback angle based on the material yield strength deviation and the nominal thickness of the safe body to obtain the predicted springback angle value. Specifically, this step is a refined execution of the springback angle calculation in step S220. The inputs are the material yield strength deviation and the nominal thickness of the safe body plate output in S221. The springback angle calculation formula is: Predicted springback angle value (°) = Benchmark springback angle (°) + [Material yield strength deviation (MPa) × Nominal plate thickness (mm) ÷ Standard plate thickness (mm)] ÷ Thickness deviation weighting constant. The standard plate thickness is set to 1.2mm for example, and the thickness deviation weighting constant is set to 10 for example. The execution process is as follows: First, read the nominal plate thickness value and calculate the thickness correction coefficient: Thickness correction coefficient = Nominal plate thickness ÷ Standard plate thickness. Then, retrieve the benchmark springback angle corresponding to the bending angle, and finally substitute it into the formula to complete the four arithmetic operations, outputting the predicted springback angle value in °.

[0035] The calculations were performed using the data from Table 1 as an example, and the results are shown in Table 2.

[0036] Table 2

[0037] Taking the data in the first row as an example, substituting it into the formula: 3.2 + (-16.75 × 1.2 ÷ 1.2) ÷ 10 = 3.2 - 1.675 = 1.525°. The larger the nominal thickness of the sheet, the larger the thickness correction coefficient, the stronger the springback attenuation effect caused by the yield deviation, and the smaller the final predicted springback angle value.

[0038] In step S300, during the multi-pass welding process of the safe body at the fully automated robotic welding workstation, the high-order current harmonic mode of the welding power inverter is extracted in real time through the arc sensor in the industrial Internet of Things sensing network.

[0039] Specifically, the executing body of this step is the welding edge computing unit supporting the fully automatic robotic tailor welding workstation, the communication carrier of the industrial Internet of Things sensing network reuses the industrial 5G private network, the arc sensor adopts a non-contact arc current acquisition sensor clamped at the front end of the welding torch, and the tailor welding power inverter has a built-in high-frequency current sampling module. The execution process is as follows: first, the arc sensor collects the original waveform of the welding current output by the inverter at a sampling frequency of 1000Hz, then the built-in fast Fourier transform operation chip performs frequency domain decomposition on the current waveform to separate the fundamental component and the second-order and higher-order high-order current harmonic components, and then calculates the modulus value of each group of high-order harmonic components, wherein the calculation of the high-order current harmonic modulus is: taking the square root of the sum of the square of the real part and the square of the imaginary part of a single-order harmonic. Finally, the high-order current harmonic modulus with a time stamp is uploaded to the welding edge computing unit for storage in real time. This step captures the current harmonic characteristics caused by welding arc fluctuations in real time, quantitatively characterizes the instantaneous fluctuation state of welding heat input, and is used to identify the risk of stress defects in the weld area.

[0040] In a possible implementation, during the multi-pass tailor welding process of the insurance cabinet by the fully automatic robotic tailor welding workstation, the high-order current harmonic modulus of the tailor welding power inverter is extracted in real time through the arc sensor in the industrial Internet of Things sensing network. Step S300 further includes step S310, which tracks and locks the current spatial coordinate of the fully automatic robotic tailor welding workstation on the weld seam in real time through the industrial Internet of Things sensing network. Specifically, this step is a coordinate tracking refinement process supporting S300, the executing body is the robot motion controller, and the input is the real-time position pulse signal of the robot six-axis encoder. The execution process is as follows: first read the pulse value of each axis encoder of the robot, convert it into the three-dimensional spatial coordinate of the end of the welding torch through the robot kinematics positive solution algorithm, with the coordinate unit being mm, then bind the real-time coordinate with the standard path coordinate of the weld seam in time sequence, and each frame of current harmonic sampling data is synchronously bound with the spatial coordinate of the welding torch at the same moment to complete coordinate locking. This step establishes a one-to-one mapping relationship among welding time sequence, spatial position and current harmonic, and realizes fixed-point calculation of heat input fluctuation at the local position of the weld seam.

[0041] Step S320: Calculate the expected heat input variance of each pass of welding in the multi-pass welding process at the current spatial coordinate along the welding time axis. Specifically, this step is the core process for judging the consistency of heat input in S300, the input is the time-sequence harmonic data packet bound with spatial coordinates output by S310, and the whole process is divided into two sequential execution actions. The first step is to call the multi-pass welding heat input formula to solve the heat input value of each welding process frame by frame, complete the variance solution based on all the heat input values of a single pass, and output the expected heat input variance corresponding to each welding pass; the second step is to compare the variance value with the preset consistency threshold to output the judgment result. This step quantifies the fluctuation degree of multi-pass welding heat input at the same weld position, so as to judge the occurrence risk of internal stress overlapping defects in the weld seam.

[0042] Step S330: When the expected thermal input variance is greater than a preset consistency threshold, adjust the pulse current of the current welding layer according to a preset pulse current adjustment table. Specifically, this step is the current adjustment execution flow S300, where the inputs are the expected thermal input variance value output in S320 and the preset consistency threshold. The preset consistency threshold is a configurable upper limit for fluctuation; for example, the threshold is set to 300 (J / mm). 2 The preset pulse current adjustment table is a two-dimensional data table. Rows correspond to the range of variance exceeding a threshold, columns correspond to the welding layer sequence number, and cells store the corresponding pulse current increase / decrease adjustment amount. The execution process first compares the expected heat input variance with a preset consistency threshold. If the variance value is less than or equal to the threshold, the heat input fluctuation is considered controllable, and no current adjustment is performed. If the variance value is greater than the threshold, the range of variance exceeding the threshold is read, and the corresponding pulse current adjustment amount is obtained by matching it with the preset pulse current adjustment table. Subsequently, a current adjustment command is sent to the downstream welding power inverter to correct the peak and base values ​​of the pulse current for the current welding layer in real time. By reducing the difference in heat input across multiple welding layers, the internal stress overlap defect caused by the superposition of residual stress in the sheet metal above and below the weld is suppressed.

[0043] In one possible implementation, calculating the expected heat input variance for each weld pass during the multi-pass welding process at the current spatial coordinates along the welding time axis, step S320 further includes step S321, calculating the multi-pass welding line energy input for each weld pass in each frame during the multi-pass welding process using the multi-pass welding line energy input formula, and performing variance calculation to determine the expected heat input variance for each weld pass; the multi-pass welding line energy input formula is: ; in, For the welding line energy input of the k-th pass, The welding thermal efficiency coefficient, Let be the real-time voltage of channel k. Let be the real-time current of channel k. The current welding movement speed when the fully automated robotic welding workstation performs the k-th welding pass.

[0044] Specifically, this step is a refinement of the S320 variance calculation execution process. The input is all welding time-series frame data bound to the current weld space coordinates. The multi-pass welding line energy input formula is: Single-pass welding line energy input (J / mm) = Welding thermal efficiency coefficient × Single-pass real-time voltage (V) × Single-pass real-time current (A) ÷ Single-pass current welding movement speed (mm / s). Where: Single-pass welding line energy input refers to the heat input per unit length of the k-th weld, in J / mm; Welding thermal efficiency coefficient is an inherent constant of the welding equipment, exemplarily 0.85 for argon arc welding; Single-pass real-time voltage refers to the welding machine output voltage collected in the corresponding frame, in V; Single-pass real-time current refers to the welding machine output current collected in the corresponding frame, in A; Single-pass current welding movement speed refers to the real-time walking speed of the robot welding torch in this frame, in mm / s. The execution process is as follows: First, the time-series frame data is grouped according to the welding pass. All sampled frames of the same pass are divided into the same calculation group. Then, the line energy input value of each frame is calculated by substituting into the formula, and the line energy values ​​of all frames of the same pass are stored. Then, the overall variance calculation formula is applied to calculate the expected heat input variance using the line energy values ​​of all frames in a single pass as samples. The variance formula is: the sum of the squares of the differences between the line energy of each frame and the average line energy of that pass, divided by the total number of sampling frames for that pass. Finally, the expected heat input variance value corresponding to each welding process is output. This step quantifies the fluctuation range of the welding heat input of a single weld seam through quantitative calculation of line energy and variance statistics, providing a quantitative basis for current adjustment.

[0045] For example, five consecutive frames of sampled data from a single weld were selected for calculation, and the results are shown in Table 3.

[0046] Table 3

[0047] The welding thermal efficiency coefficient is taken as 0.85. Substituting the data from frame 1 into the formula: 0.85 × 22.0 × 130 ÷ 4.0 = 607.75 J / mm. The average line energy across five frames is 610.75 J / mm. Summing the squares of the five differences and dividing by 5 yields the expected heat input variance of 132.26 (J / mm). 2 This value represents the degree of heat input fluctuation in this weld pass.

[0048] Step S400: Using the bending residual stress field distribution domain as a priori physical constraint boundary, the weld area is dynamically locked at both ends according to the real-time change of the higher-order current harmonic mode, and the upper and lower indicator areas corresponding to the high-risk pathological processing areas are identified respectively.

[0049] Specifically, this step is executed by the welding edge calculation unit, and the input includes two types of data: a three-dimensional mesh data table of the bending residual stress field distribution domain output by S200, and a time series dataset of high-order current harmonic mode lengths output by S300. The overall process is divided into four sequential execution actions. The first step rearranges the harmonic modes according to time sequence to generate a continuous time-series feature sequence. The second step matches the weld space location corresponding to the residual stress field distribution domain and extracts the corresponding time-series segment from the feature sequence. The third step performs double-percentile value filtering on the extracted time-series segment to separate the peak and trough lattices. The fourth step maps the two types of lattices to the welding trajectory space, locking the upper and lower indicator areas. This step combines the initial residual stress constraints of the sheet metal with the harmonic fluctuation characteristics of the welding arc to locate high-risk areas of internal stress defects in the weld, thus locking the target location for correction and stress relief processing.

[0050] In one possible implementation, the bending residual stress field distribution domain is used as a priori physical constraint boundary. Based on the real-time changes in the higher-order current harmonic modes, the weld area is dynamically locked at both ends. The upper and lower indicator areas corresponding to the high-risk pathological processing areas are identified respectively. Step S400 further includes step S410, which rearranges the higher-order current harmonic modes according to the sampling time sequence to obtain a higher-order harmonic feature sequence. Specifically, this step is the time-series data preprocessing flow S400, with the input being the discrete sampling points of the higher-order current harmonic modes with time stamps output from S300. The execution flow is as follows: first, all harmonic mode values ​​are rearranged in ascending order according to the sampling timestamps from smallest to largest to eliminate data time sequence errors caused by transmission delays; then, each rearranged group of harmonic modes is assigned a continuously increasing sequence index, with each index bound to a harmonic mode, forming a one-dimensional array of higher-order harmonic feature sequences without time stamp errors. This step unifies the timing logic of harmonic data, eliminates data out-of-order interference caused by network transmission jitter, and ensures the timing accuracy of subsequent region identification.

[0051] Step S420 involves extracting the weld location information covered by the bending residual stress field distribution domain and intercepting the corresponding local temporal feature segments in the higher-order harmonic feature sequence. Specifically, this step is the feature segment extraction process in S400, with the input being the higher-order harmonic feature sequence output in S410 and the bending residual stress field distribution domain output in S200. The execution process involves first traversing all three-dimensional grid coordinates within the bending residual stress field distribution domain and extracting the weld spatial trajectory intervals covered by all grids. Then, the coordinate-temporal index mapping relationship stored in S310 is retrieved to match the feature sequence index intervals corresponding to the weld spatial trajectory intervals. Finally, from the complete higher-order harmonic feature sequence, all harmonic modulus data within the index intervals are intercepted, and the local temporal feature segments corresponding to the weld locations are output. This step removes harmonic data of welds irrelevant to the residual stress field, reduces the computational load of subsequent feature screening, and focuses on defect identification in sheet metal weld intervals with initial residual stress.

[0052] Step S430 involves performing dual-end numerical filtering on the local time-series feature segment to obtain upper peak lattices exceeding the first preset percentile and lower trough lattices below the second preset percentile. Specifically, this step is the dual-end numerical filtering process S400, with the local time-series feature segment output from S420 as the input. The first and second preset percentiles are configurable parameters; for example, the first preset percentile is set to the 90th percentile and the second preset percentile to the 10th percentile. The execution flow involves first sorting all harmonic mode length values ​​within the local time-series feature segment in ascending order, and then calling the percentile calculation algorithm to solve for the 90th percentile critical value and the 10th percentile critical value, respectively. All harmonic mode length sampling points within a segment with values ​​greater than or equal to the 90th percentile critical value are grouped into an upper peak matrix, and all harmonic mode length sampling points within a segment with values ​​less than or equal to the 10th percentile critical value are grouped into a lower trough matrix. The time-series indexes and bound weld space coordinates corresponding to these two types of matrices are stored respectively. This step separates the extreme value ranges of arc fluctuations through quantile filtering. These extreme value ranges correspond to the locations of extreme fluctuations in weld heat input, which are the high-incidence points of stress defects.

[0053] An example of selecting a local temporal feature segment of data is shown in Table 4. Figure 3 As shown.

[0054] Table 4

[0055] The five sets of data, arranged in ascending order, are 0.42A, 0.68A, 0.95A, 1.21A, and 1.57A. Using a linear interpolation percentile algorithm, the 10th percentile critical value is 0.524A, and the 90th percentile critical value is 1.426A. Since the harmonic mode length 0.42A is less than 0.524A, index 1 is assigned to the lower trough lattice; since the harmonic mode length 1.57A is greater than 1.426A, index 5 is assigned to the upper peak lattice.

[0056] Step S440: Lock the welding trajectory area corresponding to the upper peak dot matrix as the upper indicator area, and lock the welding trajectory area corresponding to the lower trough dot matrix as the lower indicator area. Specifically, this step is the defect area locking process S400. The input is the upper peak dot matrix and lower trough dot matrix output by S430. The dot matrix stores the three-dimensional spatial coordinates of the weld corresponding to each set of sampling points. The execution process is as follows: First, extract the spatial coordinates of all continuous welds bound to the upper peak dot matrix, mark and lock the spatial range of this continuous welding trajectory as the upper indicator area. This area corresponds to the high-risk area of ​​excessive heat input and tensile defects caused by residual stress superposition. Then, extract the spatial coordinates of all continuous welds bound to the lower trough dot matrix, mark and lock the spatial range of this continuous welding trajectory as the lower indicator area. This area corresponds to the high-risk area of ​​insufficient heat input and shrinkage compressive stress defects inside the weld. Finally, output a data table of the spatial ranges of the upper and lower indicator areas with complete three-dimensional coordinate boundaries. This step involves spatially locating high-risk areas of stress defects, providing a precise coordinate range for process correction and directional vibration stress relief.

[0057] Step S500: Traverse the upper indicator area and the lower indicator area to analyze the correction control scheme and obtain the target flexible processing scheme.

[0058] Specifically, this step is executed by a cloud-based process optimization server, with the input being the spatial data tables of the upper and lower indicator areas output by the S400. The process is divided into four sequential actions. The first step retrieves a preset local fine-grained disambiguation strategy and generates independent correction schemes for the two types of defect areas. The second step extracts the three-dimensional spatial coordinates of the cabinet corresponding to the two types of indicator areas. The third step solves the stress transmission coupling matrix of adjacent welding areas based on the spatial coordinates and calculates the transmission parameters of adjacent stress influence. The fourth step uses the transmission parameters to correct the excitation-related parameters within the correction scheme, completing the disambiguation optimization with the maximum collaborative anti-damage strength as the iterative goal, and outputting a target flexible processing scheme adapted to the cabinet's spatial constraints. This step integrates independent area correction measures with the stress coupling influence between areas, optimizing a complete processing scheme that balances local defect elimination with no deterioration of stress in adjacent areas.

[0059] In one possible implementation, the upper and lower indicator areas are traversed to analyze the correction control scheme and obtain the target flexible processing scheme. Step S500 further includes step S510, which involves obtaining a preset local fine-tuning disambiguation strategy and generating multiple independent correction schemes based on the defect types of the upper and lower indicator areas. For the upper indicator area, the independent correction scheme includes issuing a local freeze command to freeze the current mechanical travel speed and welding power of the fully automated robot welding workstation, and introducing the stress normal prior derived from the bending residual stress field distribution domain for local constraint fine-tuning. For the lower indicator area, the independent correction scheme includes increasing the pulse thermal input current of the current welding layer. Specifically, this step is the S500 independent correction scheme generation process, with the input being the defect types of the upper and lower indicator areas output by S400, and the preset local fine-tuning disambiguation strategy being a database storing correction actions corresponding to different defects. The execution process is as follows: First, the defect type in the upper indicator area is determined to be an excessive heat input tensile stress overlap defect. The corresponding independent correction scheme is then retrieved: a local freeze command is sent to the lower-level robot controller to lock the welding torch mechanical travel speed and welding machine welding power to a constant value. Simultaneously, the stress normal direction parameters are extracted from the S200 bending residual stress field distribution domain, and a slight downward adjustment constraint is made to the current welding heat input based on the normal direction. Next, the defect type in the lower indicator area is determined to be an insufficient heat input shrinkage compressive stress defect. The corresponding independent correction scheme is then retrieved: an adjustment command is sent to the upper-level welding power supply to increase the current pulse heat input current reference value for the current welding layer. Complete independent correction schemes for both types of areas are stored separately. These schemes include basic parameters such as mechanical motion parameters, welding electrical parameters, initial excitation frequency, and initial excitation duration. This step provides targeted basic correction measures for two different types of stress defects, achieving independent suppression of defects in a single area.

[0060] Step S520: Extract the three-dimensional spatial coordinates of the upper and lower indicator areas during the safe assembly. Specifically, this step is the S500 spatial coordinate extraction process, with the input being the spatial range data table of the upper and lower indicator areas output from S400. The execution flow is as follows: first, traverse all boundary grid points of the upper indicator area to extract the three-dimensional spatial coordinates of the area's center point; then, traverse all boundary grid points of the lower indicator area to extract the three-dimensional spatial coordinates of the area's center point. Simultaneously, extract the three-dimensional coordinates of the centers of the remaining adjacent weld seams of the safe body. After unifying all coordinate data into mm units and coordinate systems, package and output the data to the coupling matrix calculation module. This step obtains the spatial positional relationships of all weld seam areas, providing spatial geometric input for the stress coupling calculation between areas.

[0061] Step S530: Calculate the physical stress transmission coupling matrix between adjacent welded areas based on the three-dimensional spatial coordinates to obtain the adjacent stress influence transmission parameters. Specifically, this step is the stress coupling parameter calculation process in S500. The input is the three-dimensional center coordinates of all weld areas output in S520. The physical stress transmission coupling matrix adopts the standard coupling matrix of sheet metal planar thermal stress transmission finite element method. The matrix rows and columns correspond to different weld area numbers, and the values ​​in the matrix cells represent the stress transmission coupling coefficient between two areas. The execution process is as follows: First, calculate the three-dimensional spatial straight-line distance between the center points of any two weld areas, i.e., the center-to-center distance, in mm; then, substitute the center-to-center distance into the stress attenuation calculation formula to solve for the stress transmission coupling coefficient between areas. The formula is: Coupling coefficient = Reference coupling constant ÷ Center-to-center distance. 2 The reference coupling constant is set to 1200 mm for example. 2 All pairwise coupling coefficients are filled into a two-dimensional coupling matrix. Finally, the matrix is ​​normalized, and the normalized coupling coefficients corresponding to the upper and lower indicator regions and their adjacent weld regions are extracted as the output parameters for the transmission of adjacent stress influence. This step quantifies the degree of influence of stress transmission and superposition between different weld regions, avoiding the exacerbation of stress defects in adjacent welds by single-region correction measures.

[0062] Two adjacent weld areas were selected for calculation as an example, and the results are shown in Table 5.

[0063] Table 5

[0064] Taking the first row of data as an example, 1200 ÷ 30 2 The maximum value of all coupling coefficients in the entire domain is 1.85. The normalized transfer parameter is 0.72 when the original coefficients are divided by the maximum value. The larger the value, the stronger the stress interaction between regions.

[0065] Step S540 involves using the adjacent stress influence transfer parameters to correct the excitation frequency and excitation duration parameters in the multiple independent correction schemes, performing local disambiguation iterations with the optimization objective of maximizing collaborative anti-damage strength, to obtain a target flexible processing scheme that conforms to the current component spatial constraints of the safe. Specifically, this step is the process scheme iterative optimization flow of S500, with the inputs being the multiple independent correction schemes output by S510 and the adjacent stress influence transfer parameters output by S530. The objective function expression for maximizing collaborative anti-damage strength is: Target strength = Target area stress elimination amplitude − Adjacent area stress increment × Adjacent stress influence transfer parameters. The execution flow involves first reading the initial excitation frequency and initial excitation duration parameters within the independent correction schemes, and then using the adjacent stress influence transfer parameters to perform weighted correction on the two sets of parameters. The larger the transfer parameter value, the greater the reduction in excitation frequency and the reduction in excitation duration. Subsequently, the optimized objective function is substituted to solve for the cooperative anti-damage strength value corresponding to the corrected parameters. A grid search iterative algorithm is used to slightly adjust the excitation frequency and duration parameters, repeatedly solving for the target strength until the parameter combination corresponding to the maximum target strength is found. The optimal excitation parameters are then replaced in the independent correction scheme, integrating the complete correction logic after correction for both types of regions. This outputs a target flexible manufacturing process scheme adapted to the three-dimensional spatial constraints of the safe body component. The scheme includes welding current adjustment commands, robot speed locking commands, and a complete set of ultrasonic excitation control parameters. This step considers both local defect elimination and stress protection of adjacent welds, outputting a complete flexible manufacturing process with optimal global stress.

[0066] Step S600: The remotely controlled ultrasonic stress relief mechanism performs localized directional excitation stress relief control on the upper and lower indicator areas according to the target flexible processing scheme.

[0067] Specifically, this step is executed by a cloud-based process optimization server and a local servo driver for the ultrasonic stress relief mechanism. The input is the target flexible manufacturing process scheme output by the S500, and the entire process is divided into two sequential actions. The first step analyzes the excitation-related parameters within the process scheme, converts them into digital commands for control current and excitation frequency that the servo driver can recognize, and then issues them. The second step simultaneously issues robot parameter locking commands to stabilize the welding process parameters and simultaneously drives the ultrasonic mechanism to perform directional local excitation on two indicated areas, eliminating residual stress superimposed within the sheet metal weld. This step executes a complete optimization process, directionally eliminating stress defects in high-risk areas of the weld while simultaneously ensuring that the welding process is not interfered with by stress relief operations.

[0068] In one possible implementation, the remotely controlled ultrasonic stress relief mechanism performs localized directional excitation stress relief control on the upper and lower indicating areas according to the target flexible processing technology scheme. Step S600 further includes step S610, which converts the target flexible processing technology scheme into control current and excitation frequency, and sends them to the servo actuator of the ultrasonic stress relief mechanism. Specifically, this step is the process parameter conversion and sending process of S600, with the target flexible processing technology scheme output by S500 as the input. The execution process is as follows: first, the optimal excitation frequency, excitation duration, and local excitation power upper limit values ​​stored in the scheme are extracted; then, the ultrasonic servo driver parameter conversion lookup table is retrieved to map the excitation frequency to the drive coil control current frequency and the excitation power to the drive current amplitude, generating a servo execution digital instruction package with timing control. The instruction package contains the excitation start and end times, current amplitude, and frequency parameters corresponding to the upper and lower indicating areas, respectively. Finally, the instruction package is sent to the servo actuator of the ultrasonic stress relief mechanism via the industrial 5G private network for buffering and execution. This step completes the standardized conversion of process parameters into hardware driving instructions, enabling precise and controllable execution of the ultrasonic excitation equipment.

[0069] Step S620: While driving the ultrasonic stress-relieving mechanism to perform localized directional excitation stress-relieving control, a parameter locking command is issued to maintain the stability of the current process parameters of the fully automated robotic welding workstation. Specifically, this step is the synchronous collaborative control process of S600. The execution process is as follows: First, a parameter locking command is issued to the motion controller and welding power inverter of the fully automated robotic welding workstation. The locked content includes all real-time process parameters such as welding torch travel speed, welding pulse current, and welding voltage. The locking period is completely matched with the ultrasonic excitation duration. Then, the ultrasonic servo actuator is triggered to read the cached command packet, drive the ultrasonic excitation probe to move to the weld space coordinates corresponding to the upper and lower indicator areas, and perform localized directional high-frequency excitation according to the command parameters. During the excitation process, the data of the arc sensor and robot encoder are continuously read to confirm that the welding process parameters do not drift. After the excitation duration is exhausted, the robot process parameter locking is automatically released, completing the entire process of single-section cabinet sheet metal forming. This step realizes the synchronous collaborative operation of the welding process and the ultrasonic stress-relieving process, avoids the stress-relieving operation from interfering with the stability of welding heat input, and simultaneously releases the superimposed residual stress inside the weld in a directional manner, eliminating stress overlap defects.

[0070] This application embodiment collects the displacement-pressure curve of a single forming process of a CNC bending machine through an industrial Internet of Things (IoT) sensing network. The curve is then spatiotemporally truncated to obtain stress response data, and the residual stress field distribution domain of the bending process is analyzed. Subsequently, during the multi-stage welding of the cabinet by the robot, an arc sensor is used to extract the high-order harmonic mode length of the power inverter in real time. Then, using the residual stress field distribution domain as a priori constraint combined with the harmonic mode length, the weld seam is dynamically locked, and two high-risk indicator areas are identified. The two areas are then analyzed to generate a target flexible processing scheme. Finally, an ultrasonic stress-relieving mechanism is remotely controlled to perform localized directional vibration stress relief treatment on the two indicator areas according to this scheme. These technical means solve the technical problem of existing sheet metal cabinet bending-welding stress collaborative management processes, which cannot accurately locate local high-stress risk areas in the weld seam and achieve differentiated directional stress relief treatment. This achieves the technical effects of accurately locking local high-stress areas in the weld seam, differentially eliminating superimposed residual stress in defective areas, reducing equipment wear caused by indiscriminate full-domain vibration, fully releasing internal stress at high-risk points, and improving the overall structural strength and finished product processing consistency of the safe.

[0071] In the above text, refer to Figures 1-3 This paper describes in detail a sheet metal forming method for a safe body based on the Internet of Things (IoT) according to an embodiment of the present invention. Next, we will refer to... Figure 4 This invention describes an Internet of Things-based sheet metal forming and processing system for safe bodies according to an embodiment of the present invention.

[0072] The IoT-based safe body sheet metal forming and processing system according to embodiments of the present invention addresses the technical problem of existing sheet metal safe bending-welding process stress collaborative management failing to accurately locate local high-stress risk areas in welds and achieve differentiated directional stress relief treatment. It achieves the technical effects of accurately locating local high-stress areas in welds, differentially eliminating superimposed residual stress in defective areas, reducing equipment wear caused by indiscriminate global vibration, fully releasing internal stress at high-risk points, and improving the overall structural strength and finished product consistency of the safe body. The IoT-based safe body sheet metal forming and processing system includes: a displacement-pressure curve acquisition module 10, a bending residual stress analysis module 20, a high-order current harmonic mode extraction module 30, an indicator area identification module 40, a correction control scheme analysis module 50, and a stress relief control module 60.

[0073] The displacement-pressure curve acquisition module 10 is used to acquire the displacement-pressure curve of the CNC bending machine in a single forming process in real time using an industrial Internet of Things (IoT) sensing network at a preset acquisition frequency; the bending residual stress analysis module 20 is used to perform spatiotemporal truncation processing on the displacement-pressure curve to determine the stress response data, and combine the stress response data to perform bending residual stress analysis to determine the bending residual stress field distribution domain; the high-order current harmonic mode length extraction module 30 is used to extract the high-order current harmonic mode length of the welding power inverter in real time through the arc sensor in the industrial IoT sensing network during the multi-pass welding process of the safe body in the fully automatic robot welding workstation. The indicator area identification module 40 is used to use the bending residual stress field distribution domain as a priori physical constraint boundary, and to perform dual-end dynamic locking of the weld area according to the real-time change of the higher-order current harmonic mode, respectively identifying the upper and lower indicator areas corresponding to the high-risk pathological processing area; the deviation correction control scheme analysis module 50 is used to traverse the upper and lower indicator areas to analyze the deviation correction control scheme and obtain the target flexible processing scheme; the stress relief control module 60 is used to remotely control the ultrasonic stress relief mechanism to implement local directional excitation stress relief control for the upper and lower indicator areas according to the target flexible processing scheme.

[0074] The detailed description of the specific configuration of the bending residual stress analysis module 20 is as follows: As mentioned above, the displacement-pressure curve is processed by spatiotemporal truncation to determine stress response data. Combined with the stress response data, bending residual stress analysis is performed to determine the bending residual stress field distribution domain. The bending residual stress analysis module 20 may further include: a stress response data extraction unit used to truncate and discard the stress data of the initial segment and unloading segment of the displacement-pressure curve, using a preset standard deviation after the mold cuts into the plate as the boundary, and extract the stress response data of the core holding pressure segment; a springback prediction unit used to perform springback prediction based on the stress response data to obtain the cabinet springback angle prediction value; and a bending residual stress field distribution domain construction unit used to dynamically adjust the pressing depth and holding pressure time of the CNC bending machine at the dead point position according to the cabinet springback angle prediction value, mapping and constructing a bending residual stress field distribution domain that matches the cabinet springback angle prediction value.

[0075] Specifically, the rebound prediction unit can further include: a nonlinear curvature fitting subunit used to traverse the stress response data and perform nonlinear curvature fitting with the standard physical stress model to obtain the material yield strength deviation value; and a rebound angle calculation subunit used to calculate the rebound angle based on the material yield strength deviation value and the nominal thickness of the safe to obtain the predicted value of the rebound angle.

[0076] The detailed description of the specific configuration of the indicator region identification module 40 is explained as follows: As mentioned above, the bending residual stress field distribution domain is used as the a priori physical constraint boundary. Based on the real-time change of the higher-order current harmonic mode length, the weld area is dynamically locked at both ends, identifying the upper and lower indicator regions corresponding to the high-risk pathological processing area. The indicator region identification module 40 may further include: a numerical rearrangement unit for numerically rearranging the higher-order current harmonic mode length according to the sampling time sequence to obtain a higher-order harmonic feature sequence; and a local time sequence feature fragment interception unit. The unit is used to extract the weld location information covered by the bending residual stress field distribution domain and intercept the corresponding local time-series feature segments in the higher-order harmonic feature sequence; the dual-end numerical filtering unit is used to perform dual-end numerical filtering on the local time-series feature segments to obtain the upper peak lattice exceeding the first preset percentile and the lower trough lattice below the second preset percentile respectively; the welding trajectory region locking unit is used to lock the welding trajectory region corresponding to the upper peak lattice as the upper indication region and lock the welding trajectory region corresponding to the lower trough lattice as the lower indication region.

[0077] The detailed description of the high-order current harmonic mode length extraction module 30 is as follows: As mentioned above, during the multi-pass welding process of the safe body by the fully automatic robot welding workstation, the high-order current harmonic mode length of the welding power inverter is extracted in real time through the arc sensor in the industrial Internet of Things sensing network. The high-order current harmonic mode length extraction module 30 may further include: a current spatial coordinate locking unit for tracking and locking the current spatial coordinates of the fully automatic robot welding workstation on the weld seam in real time through the industrial Internet of Things sensing network; a desired heat input variance calculation unit for calculating the desired heat input variance of each welding pass in the multi-pass welding process at the current spatial coordinates along the welding time axis; and a pulse current adjustment unit for adjusting the pulse current of the current welding layer according to a preset pulse current adjustment table when the desired heat input variance is greater than a preset consistency threshold.

[0078] Specifically, the expected heat input variance is calculated for each weld pass during the multi-pass welding process at the current spatial coordinates along the welding time axis. The expected heat input variance calculation unit may further include: a variance calculation subunit used to calculate the multi-pass welding line energy input for each weld pass in each frame during the multi-pass welding process using the multi-pass welding line energy input formula, and to perform variance calculation to determine the expected heat input variance for each weld pass; the multi-pass welding line energy input formula is: ;in, For the welding line energy input of the k-th pass, The welding thermal efficiency coefficient, Let be the real-time voltage of channel k. Let be the real-time current of channel k. The current welding movement speed of the fully automated robotic welding workstation when performing the k-th welding pass.

[0079] The detailed description of the specific configuration of the correction control scheme analysis module 50 is as follows: As mentioned above, the correction control scheme analysis is performed by traversing the upper and lower indicator areas to obtain the target flexible processing scheme. The correction control scheme analysis module 50 may further include: an independent correction scheme generation unit for obtaining a preset local fine-grained disambiguation strategy, and mapping and generating multiple independent correction schemes according to the defect types of the upper and lower indicator areas; a three-dimensional spatial coordinate extraction unit for extracting the three-dimensional spatial coordinates of the upper and lower indicator areas in the safe assembly; a physical stress transmission coupling matrix calculation unit for calculating the physical stress transmission coupling matrix between adjacent welding areas according to the three-dimensional spatial coordinates to obtain the adjacent stress influence transmission parameters; and a local disambiguation iteration unit for using the adjacent stress influence transmission parameters to correct the excitation frequency parameters and excitation duration parameters in the multiple independent correction schemes, and performing local disambiguation iteration with the optimization goal of maximizing the collaborative anti-damage strength to obtain the target flexible processing scheme that conforms to the current component spatial constraints of the safe.

[0080] The independent correction scheme generation unit may further include: for the upper indication area, the independent correction scheme includes issuing a local freeze command to freeze the current mechanical walking speed and welding power of the fully automatic robot welding workstation, and introducing the stress normal prior derived from the bending residual stress field distribution domain for local constraint fine-tuning; for the lower indication area, the independent correction scheme includes increasing the pulse thermal input current of the current welding layer.

[0081] The detailed description of the specific configuration of the stress relief control module 60 is explained as follows: As mentioned above, the remotely controlled ultrasonic stress relief mechanism implements local directional excitation stress relief control for the upper and lower indicator areas according to the target flexible processing technology scheme. The stress relief control module 60 may further include: a control parameter sending unit for converting the target flexible processing technology scheme into control current and excitation frequency, and sending them to the servo actuator of the ultrasonic stress relief mechanism; and a parameter locking unit for sending a parameter locking command while driving the ultrasonic stress relief mechanism to perform local directional excitation stress relief control, so as to keep the current process parameters of the fully automatic robot welding workstation stable.

[0082] The IoT-based safe sheet metal forming and processing system provided in this embodiment of the invention can execute the IoT-based safe sheet metal forming and processing method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0083] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0084] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for forming and processing sheet metal for safes based on the Internet of Things, characterized in that, The method includes: By utilizing an industrial Internet of Things (IoT) sensing network at a preset sampling frequency, the displacement-pressure curve of a CNC bending machine during a single forming process can be acquired in real time. The displacement-pressure curve is subjected to spatiotemporal truncation processing to determine the stress response data, and the bending residual stress is analyzed in combination with the stress response data to determine the distribution domain of the bending residual stress field. During the multi-stage welding process of the safe body in the fully automated robotic welding workstation, the high-order current harmonic mode of the welding power inverter is extracted in real time through the arc sensor in the industrial Internet of Things sensing network. Using the bending residual stress field distribution domain as a priori physical constraint boundary, the weld area is dynamically locked at both ends according to the real-time change of the higher-order current harmonic mode, and the upper and lower indicator areas corresponding to the high-risk pathological processing areas are identified respectively. By traversing the upper and lower indicator areas, a correction control scheme analysis is performed to obtain the target flexible processing scheme. The remotely controlled ultrasonic stress relief mechanism performs localized directional excitation stress relief control on the upper and lower indicator areas according to the target flexible processing scheme.

2. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 1, characterized in that, The displacement-pressure curve is subjected to spatiotemporal truncation processing to determine stress response data. Combined with this stress response data, residual bending stress is analyzed to determine the distribution domain of the residual bending stress field, including: Using the preset standard deviation after the mold cuts into the plate as the boundary, the stress data of the initial segment and the unloading segment in the displacement-pressure curve are cut off and discarded, and the stress response data of the core pressure holding segment is extracted. Based on the stress response data, springback prediction is performed to obtain the predicted value of the cabinet springback angle. The pressing depth and holding time of the CNC bending machine at the dead point position are dynamically adjusted based on the predicted value of the cabinet springback angle, thereby mapping and constructing a bending residual stress field distribution domain that matches the predicted value of the cabinet springback angle.

3. The method for forming and processing sheet metal for a safe based on the Internet of Things as described in claim 2, characterized in that, Based on the stress response data, springback prediction is performed to obtain the predicted value of the cabinet springback angle, including: The stress response data is traversed and nonlinear curvature fitting is performed with the standard physical stress model to obtain the material yield strength deviation value. The rebound angle is calculated based on the yield strength deviation of the material and the nominal thickness of the safe to obtain the predicted value of the safe's rebound angle.

4. The method for forming and processing sheet metal for a safe based on the Internet of Things as described in claim 1, characterized in that, Using the bending residual stress field distribution domain as a priori physical constraint boundary, and based on the real-time change of the higher-order current harmonic mode, the weld area is dynamically locked at both ends, identifying the upper and lower indicator areas corresponding to the high-risk ill-processed areas, including: The higher-order current harmonic modes are numerically rearranged according to the sampling time sequence to obtain the higher-order harmonic characteristic sequence. Extract the weld location information covered by the bending residual stress field distribution domain, and intercept the corresponding local time-series feature segments in the higher-order harmonic feature sequence; The local time-series feature segments are subjected to dual-end numerical filtering to obtain upper peak lattice points exceeding the first preset percentile and lower trough lattice points below the second preset percentile. The welding trajectory area corresponding to the upper peak dot matrix is ​​locked as the upper indicator area, and the welding trajectory area corresponding to the lower trough dot matrix is ​​locked as the lower indicator area.

5. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 1, characterized in that, During the multi-stage welding process of the safe body in the fully automated robotic welding workstation, the high-order current harmonic mode of the welding power inverter is extracted in real time through the arc sensor in the industrial Internet of Things sensing network. This also includes: The industrial Internet of Things (IoT) sensing network is used to track and lock the current spatial coordinates of the fully automated robotic welding workstation on the weld seam in real time. Calculate the expected heat input variance for each welding pass in a multi-pass welding process at the current spatial coordinates along the welding time axis. When the expected thermal input variance is greater than the preset consistency threshold, the pulse current of the current welding layer is adjusted according to the preset pulse current adjustment table.

6. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 5, characterized in that, Calculate the expected heat input variance for each welding pass in a multi-pass welding process at the current spatial coordinates along the welding time axis, including: The multi-pass welding line energy input formula is used to calculate the multi-pass welding line energy input of each weld in each frame during the multi-pass welding process, and the variance is calculated to determine the expected heat input variance for each weld. The energy input formula for the multi-pass welding line is: ； in, For the welding line energy input of the k-th pass, The welding thermal efficiency coefficient, Let be the real-time voltage of channel k. Let be the real-time current of channel k. The current welding movement speed when the fully automated robotic welding workstation performs the k-th welding pass.

7. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 1, characterized in that, By traversing the upper and lower indicator areas to analyze the correction control scheme, a target flexible manufacturing process scheme is obtained, including: A preset local fine-grained disambiguation strategy is obtained, and multiple independent correction schemes are generated by mapping according to the defect types of the upper and lower indicator regions. Extract the three-dimensional spatial coordinates of the upper and lower indicator areas during the assembly of the safe; Calculate the physical stress transmission coupling matrix between adjacent welded areas based on the three-dimensional spatial coordinates to obtain the adjacent stress influence transmission parameters. The excitation frequency and excitation duration parameters in the multiple independent correction schemes are modified using the adjacent stress influence transmission parameters. Local disambiguation iteration is performed with the optimization objective of maximizing the collaborative anti-damage strength to obtain a target flexible processing scheme that meets the current component spatial constraints of the safe.

8. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 7, characterized in that, For the upper indicated area, the independent correction scheme includes issuing a local freeze command to freeze the current mechanical walking speed and welding power of the fully automatic robot welding workstation, and introducing the stress normal prior derived from the bending residual stress field distribution domain for local constraint fine-tuning. For the lower indicator area, the independent correction scheme includes increasing the pulse thermal input current of the current weld layer.

9. The method for forming and processing sheet metal of a safe based on the Internet of Things as described in claim 1, characterized in that, The remotely controlled ultrasonic stress relief mechanism performs localized directional excitation stress relief control on the upper and lower indicator areas according to the target flexible processing scheme, including: The target flexible processing scheme is converted into control current and excitation frequency, and sent to the servo actuator of the ultrasonic stress relief mechanism; While driving the ultrasonic stress relief mechanism to perform localized directional excitation stress relief control, a parameter locking command is issued to keep the current process parameters of the fully automatic robotic welding workstation stable.

10. A safe body sheet metal forming and processing system based on the Internet of Things, characterized in that, The system is used to implement the IoT-based safe body sheet metal forming and processing method according to any one of claims 1-9, the system comprising: The displacement-pressure curve acquisition module is used to acquire the displacement-pressure curve of the CNC bending machine in a single forming process in real time by using an industrial Internet of Things sensing network at a preset acquisition frequency. The bending residual stress analysis module is used to perform spatiotemporal truncation processing on the displacement-pressure curve, determine the stress response data, and combine the stress response data to perform bending residual stress analysis to determine the bending residual stress field distribution domain. The high-order current harmonic mode extraction module is used to extract the high-order current harmonic mode of the welding power inverter in real time through the arc sensor in the industrial Internet of Things sensing network during the multi-pass welding process of the safe body in the fully automated robot welding workstation. The indicator area identification module is used to take the bending residual stress field distribution domain as the a priori physical constraint boundary, and dynamically lock the weld area at both ends according to the real-time change of the higher-order current harmonic mode, and identify the upper and lower indicator areas of the corresponding high-risk pathological processing areas respectively. The correction control scheme analysis module is used to traverse the upper indicator area and the lower indicator area to analyze the correction control scheme and obtain the target flexible processing scheme. The stress relief control module is used to remotely control the ultrasonic stress relief mechanism to perform localized directional excitation stress relief control on the upper and lower indicator areas according to the target flexible processing scheme.

Images