A steel plate flatness detection and welding parameter dynamic compensation method and system

Abstract

The present application relates to the technical field of welding control measurement, and particularly relates to a steel plate flatness detection and welding parameter dynamic compensation method and system, the method comprising: constructing a steel plate flatness visual detection unit, collecting three-dimensional point cloud data of a steel plate for an electric tricycle frame; pre-processing the three-dimensional point cloud data and extracting flatness characteristic parameters, and comparing the flatness characteristic parameters with a preset flatness standard threshold to determine whether the flatness is qualified; when the flatness is unqualified, constructing a deviation-welding parameter mapping model based on a flatness deviation value to determine initial compensation parameters for welding, and collecting weld seam temperature, current and steel plate deformation in real time during the welding process, and dynamically correcting the welding parameters through a PID adaptive algorithm; after the welding is completed, the flatness is detected again to verify the compensation effect and update the mapping model. The present application identifies the flatness deviation of the steel plate in advance through visual detection, realizes dynamic optimization of the welding parameters in combination with a neural network and a PID algorithm, and improves the precision and efficiency of the welding of the steel plate for the electric tricycle frame.

Description

Technical Field

[0001] This invention relates to the field of welding control and measurement technology, specifically to a method and system for steel plate flatness detection and dynamic compensation of welding parameters. Background Technology

[0002] The frame of an electric tricycle is the core load-bearing component of the entire vehicle, mainly welded from steel plates. Its flatness directly affects the assembly accuracy, driving stability, and durability of the frame. In actual production by electric tricycle manufacturers, flatness deviations are prone to occur during the steel plate welding process. The main reasons include: Inspection is delayed and has low accuracy: Most existing enterprises use manual straightedge inspection or sampling inspection after welding. Manual straightedge inspection has low accuracy and cannot capture minute flatness deviations. In addition, the inspection is carried out after welding. When the deviation is exceeded, rework is required, resulting in material waste and reduced production efficiency. Welding parameter standardization: Welding steel plates with different flatness conditions using uniform welding parameters, including current, voltage, speed, etc., without considering the influence of the initial deviation of the steel plate on welding deformation. For example, if a steel plate with initial depression is welded according to standard parameters, the depression is easily aggravated by concentrated heat input, resulting in further deviation of the flatness of the steel plate. Lack of dynamic compensation mechanism: During the welding process of steel plates, the deformation of steel plates is affected by real-time factors such as weld temperature and current fluctuations. Fixed parameters cannot cope with dynamic changes. For example, when the current suddenly increases, it is easy to cause local overheating and deformation. Existing methods have no real-time correction capability. Poor adaptability: Electric tricycle frame steel plates come in various specifications, such as main beam steel plates of 1200mm×300mm and connecting plates of 800mm×200mm. The welding deformation patterns of steel plates of different specifications are different. The existing general compensation methods have not been optimized for the steel plate characteristics of electric tricycle manufacturers, and the compensation accuracy is only about 80%, which is difficult to meet the assembly requirements.

[0003] To address the aforementioned issues, there is an urgent need to develop a flatness control method that adapts to the characteristics of steel plates used in electric tricycles and integrates advance detection, dynamic compensation, and closed-loop verification, in order to improve welding quality and production efficiency. Summary of the Invention

[0004] To address the aforementioned technical shortcomings, this invention provides a method and system for detecting the flatness of steel plates and dynamically compensating welding parameters, thereby solving the problems of poor flatness and low welding efficiency of steel plates after welding in the prior art.

[0005] This invention is achieved through the following technical solution: A method for steel plate flatness detection and dynamic compensation of welding parameters is provided, the method comprising the following steps: Step S10: Construct a steel plate flatness visual inspection unit, collect three-dimensional point cloud data of the steel plate used for the electric tricycle frame, collect three-dimensional point cloud data of the steel plate surface for the steel plate used for the electric tricycle frame, the steel plate flatness visual inspection unit includes 2 line laser scanners, 1 industrial camera and motion platform, the scanning range covers the entire area of ​​the steel plate, and the maximum size is determined according to the specific steel plate size; Step S20: Perform data preprocessing on the three-dimensional point cloud data and extract the flatness feature parameters of the steel plate, compare them with the preset steel plate flatness standard threshold, and determine whether the flatness of the steel plate is qualified. Step S30: When the flatness of the steel plate is qualified, the inspection is passed. When the flatness of the steel plate is unqualified, a flatness deviation-welding parameter mapping model is constructed based on historical steel plate flatness inspection data and steel plate welding process data. The flatness deviation value of the current unqualified steel plate is input into the model to obtain the compensation welding parameters. Step S40: Weld steel plates with qualified flatness according to preset welding parameters, and weld steel plates with unqualified flatness according to the obtained compensation welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensation welding parameters obtained in step S30 are dynamically corrected by PID adaptive algorithm to complete the welding. Step S50: After welding is completed, the three-dimensional point cloud data of the steel plate is collected again by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, the flatness error value of the steel plate after welding is calculated. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified. The deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, return to steps S30 and S40 to recalculate the welding parameters and perform a second welding until the flatness is qualified.

[0006] Preferably, the step of constructing the steel plate flatness visual inspection unit in step S10 includes: Equipment layout: Two line laser scanners are symmetrically installed above the motion platform, with their scanning directions overlapping to cover the steel plate surface, ensuring no blind spots. An industrial camera is installed between the scanners, with its lens axis perpendicular to the steel plate surface, to assist in positioning the reference holes on the steel plate. The motion platform is placed on a horizontal foundation, with anti-slip rubber pads laid on the surface to prevent the steel plate from slipping when moving. Coordinate calibration: A standard calibration board is used to unify the coordinates of the scanner and the camera. The coordinate data of the calibration board in the field of view of the scanner and the camera are collected respectively. The coordinates of the two are unified through the coordinate transformation algorithm to establish a world coordinate system, so that the coordinate deviation of the data collected by different devices is ≤0.02mm. Motion control: The PLC controller controls the motion platform to move the steel plate. When the steel plate moves with the platform, the scanner and industrial camera are triggered to collect three-dimensional data of the steel plate surface. During the acquisition process, the vibration amplitude of the steel plate is controlled within ±0.01mm to avoid vibration affecting the data accuracy.

[0007] Preferably, the step S20 of preprocessing the three-dimensional point cloud data and extracting the flatness feature parameters of the steel plate includes: Noise removal: A statistical filtering algorithm is used to select 15-20 neighboring points for each point cloud data, calculate the mean and standard deviation of the distance from the neighboring points to the point, and remove noise points such as dust and scanning interference points whose distance is greater than 1.5 to 2.0 times the mean standard deviation. Point cloud registration: Using the reference hole of the steel plate of the electric tricycle as the positioning reference, the coordinates of the center of the reference hole are identified by the camera image, and then the point cloud data is aligned with the center coordinates to achieve point cloud registration; Data downsampling: A voxel grid downsampling algorithm is adopted to set the voxel size of the point cloud data to 0.05mm, which reduces the amount of data while ensuring accuracy and improves the efficiency of subsequent feature extraction. Reference plane fitting: Select reference points on the edge of the steel plate from the 3D point cloud data after point cloud registration, and fit the reference plane using the least squares method; Deviation calculation: Calculate the vertical distance from all point cloud data to the reference plane. A negative distance indicates a depression, denoted as d_neg, and a positive distance indicates a convexity, denoted as d_pos; Characteristic parameters are determined as follows: the maximum indentation depth, the maximum protrusion height, and the flatness error value are determined, where the maximum indentation depth D_neg=|min(d_neg)|, the maximum protrusion height D_pos=max(d_pos), and the flatness error value F=D_neg+D_pos; Acceptance criteria: When the flatness error value F is less than or equal to the preset standard threshold for flatness of steel plate, the flatness of the steel plate is deemed acceptable; when the flatness error value F is greater than the preset standard threshold for flatness of steel plate, the flatness of the steel plate is deemed unacceptable.

[0008] Preferably, the step of constructing the flatness deviation-welding parameter mapping model in step S30 includes: Dataset Construction: Collect historical welding data of steel plates from electric tricycles, including flatness feature parameters, namely maximum indentation depth D_neg, maximum protrusion height D_pos, and flatness error value F, and corresponding welding parameters, including current, voltage, speed, and preheating temperature, and divide them into training set and validation set in a 7:3 ratio; Model structure design: The flatness deviation-welding parameter mapping model includes an input layer, a hidden layer and an output layer. The number of nodes in the input layer is set to 3, and each node corresponds to the maximum indentation depth D_neg, the maximum protrusion height D_pos and the flatness error value F, respectively. The number of hidden layers is set to 2, and the number of nodes in the output layer is set to 4. Each node corresponds to the current, voltage, speed and preheating temperature, respectively. The activation function is the Sigmoid function. Model training optimization: A BP neural network is used to train the flatness deviation-welding parameter mapping model. Gradient descent is used to optimize the network weights. The learning rate and training rounds are set. The training set is divided in the dataset construction step for training. After each training round, the validation set is divided in the dataset construction step to test the model output accuracy. When the validation set error is ≤5%, training is stopped and the model parameters are saved.

[0009] Preferably, the step of dynamically correcting the compensated welding parameters obtained in step S30 using a PID adaptive algorithm in step S40 includes the following steps: Deviation definition: Assume the target deformation of the steel plate is 0, that is, the steel plate has no deformation, and the real-time deformation is ΔX. Define the PID input deviation e = ΔX - 0 = ΔX. PID parameter calculation: Set the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd, with a value range of [0.05-0.2]. Calculate the PID output control quantity ΔU based on the PID input deviation e, as shown in equation (1): (1) in The integral of the deviation represents the sum of all e over a period of time, and de / dt is the differential of the deviation, representing the rate of change of the deformation deviation of the steel plate. Parameter correction: Welding parameters are corrected according to the control quantity ΔU. When the real-time deformation ΔX > 0, i.e. the steel plate bulges, the welding current is reduced and the welding speed is increased. The current reduction is ΔI = -0.2 × ΔU, and the welding speed change is ΔV = 0.5 × ΔU. When ΔX < 0, i.e. the steel plate is concave, the preheating temperature is increased and the welding speed is reduced. The preheating temperature increase is ΔT = 0.3 × ΔU, and the welding speed change is ΔV = -0.5 × ΔU. At the same time, the corrected parameters must be within the parameter range set in step S30. Adaptive adjustment: The average deviation is calculated every 10 correction cycles. When the average deviation is greater than 0.1 mm, Kp and Ki are increased by 10%. When the average deviation is less than 0.05 mm, Kp and Kd are decreased by 5%. When 0.05 mm ≤ average deviation ≤ 0.1 mm, Kp and Ki are kept unchanged to ensure a balance between the dynamic response speed and stability of the algorithm.

[0010] Preferably, step S50, which involves entering the deviation data and compensation parameters into the database to update the model, includes: Data filtering: Filter eligible compensation data from the database within the past 3 months and remove abnormal data; Model fine-tuning: Add the selected qualified compensation data as new data to the training set, keep the model network structure unchanged, set the training rounds and retrain, and update the network weights; Accuracy verification: The model with updated network weights is tested on the validation set data. When the test error is ≤5%, the new model is saved. When the error is >5%, the number of training rounds is increased until the error is acceptable, ensuring that the model adapts to the changes in the welding process of steel plates for electric tricycles.

[0011] Furthermore, to achieve the above objectives, the present invention also proposes a steel plate flatness detection and welding parameter dynamic compensation system, wherein the steel plate flatness detection and welding parameter dynamic compensation system comprises: Visual inspection module: used to construct a visual inspection unit for the flatness of steel plates and to collect three-dimensional point cloud data of steel plates used for electric tricycle frames; Data preprocessing module: used to preprocess 3D point cloud data and extract steel plate flatness feature parameters, compare them with preset steel plate flatness standard thresholds, and determine whether the steel plate flatness is qualified; Flatness Deviation-Welding Parameter Mapping Model Construction Module: When the flatness of the steel plate is qualified, it is inspected; when the flatness of the steel plate is unqualified, it constructs a flatness deviation-welding parameter mapping model based on historical steel plate flatness inspection data and steel plate welding process data, and inputs the flatness deviation value of the current unqualified steel plate into the model to obtain the compensation welding parameters. Welding parameter dynamic compensation module: It is used to weld steel plates with qualified flatness according to preset welding parameters, and to weld steel plates with unqualified flatness according to the obtained compensated welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensated welding parameters obtained in step S30 are dynamically corrected through PID adaptive algorithm to complete the welding. The steel plate flatness secondary inspection module is used to collect three-dimensional point cloud data of the steel plate again after welding by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, it calculates the flatness error value of the steel plate after welding. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified, and the deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, it returns to the flatness deviation-welding parameter mapping model construction module and the welding parameter dynamic compensation module to recalculate the welding parameters and perform secondary welding until the flatness is qualified.

[0012] Furthermore, to achieve the above objectives, the present invention also proposes a steel plate flatness detection and welding parameter dynamic compensation device. The device includes: a memory, a processor, and programs such as a steel plate flatness detection and welding parameter dynamic compensation algorithm stored in the memory and executable on the processor. The steel plate flatness detection and welding parameter dynamic compensation algorithm and other programs are steps for implementing the steel plate flatness detection and welding parameter dynamic compensation method described above.

[0013] In addition, to achieve the above objectives, the present invention also provides a computer program product, which includes programs such as a steel plate flatness detection and welding parameter dynamic compensation algorithm. When the steel plate flatness detection and welding parameter dynamic compensation algorithm is executed by a processor, it implements the steel plate flatness detection and welding parameter dynamic compensation method as described above.

[0014] The advantages and effects of this invention are: This invention proposes a method and system for steel plate flatness detection and dynamic compensation of welding parameters. It adopts line laser scanning technology to improve the flatness detection accuracy. At the same time, it constructs a flatness deviation-welding parameter mapping model based on BP neural network to output customized compensation parameters for steel plates with different flatness deviations. In addition, the welding parameters are corrected in real time through PID adaptive algorithm, which can cope with dynamic interference such as current fluctuations and temperature changes, improve the welding quality and production efficiency of the frame, and reduce rework costs. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for detecting the flatness of steel plates and dynamically compensating welding parameters according to the present invention.

[0017] Figure 2 This is a schematic diagram of a steel plate flatness detection and welding parameter dynamic compensation system according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, in one embodiment of the present invention, a method for detecting the flatness of steel plates and dynamically compensating welding parameters includes the following steps: Step S10: Construct a visual inspection unit for the flatness of the steel plate, and collect 3D point cloud data of the steel plate used for the electric tricycle frame. For example, the steel plate thickness is 2-5mm and the steel plate model is Q235 / Q345. Collect 3D point cloud data of the steel plate surface. The visual inspection unit for the flatness of the steel plate includes 2 line laser scanners with a scanning accuracy of ±0.05mm, 1 industrial camera with a resolution of 2048×1536, and a motion platform. The scanning range covers the entire area of ​​the steel plate, and the maximum size is determined according to the specific steel plate size. The acquisition frequency is set to 10-20Hz.

[0020] Specifically, the step of constructing the steel plate flatness visual inspection unit in step S10 includes: Equipment layout: Two line laser scanners are symmetrically installed above the motion platform at a height of 800-1000mm. The scanning directions cross and cover the surface of the steel plate at an angle of 60-90° to ensure that there are no blind spots in the cross-coverage of the steel plate surface. An industrial camera is installed between the scanners with the lens axis perpendicular to the surface of the steel plate to assist in positioning the reference holes of the steel plate. The motion platform is placed on a horizontal foundation and the surface is covered with anti-slip rubber pads to prevent the steel plate from slipping when moving. Coordinate calibration: A standard calibration board is used to unify the coordinates of the scanner and the camera. The standard calibration board has an accuracy of ±0.01mm. The coordinate data of the calibration board in the field of view of the scanner and the camera are collected respectively. The coordinates of the two are unified through the coordinate transformation algorithm to establish a world coordinate system, so that the coordinate deviation of the data collected by different devices is ≤0.02mm. Motion control: The motion platform is controlled by a PLC controller to move the steel plate. For example, the S7-1200 can move the platform at a speed of 50-100 mm / s. When the steel plate moves with the platform, the scanner and industrial camera are triggered to collect three-dimensional data of the steel plate surface. During the acquisition process, the vibration amplitude of the steel plate is controlled within ±0.01 mm to avoid vibration affecting the accuracy of the data.

[0021] In addition, after scanning, the point cloud data from the two scanners and the image data from the camera are transmitted to an industrial computer. The TCP / IP protocol is used to achieve real-time data transmission with a transmission delay of ≤100ms. The raw data is stored in a local MySQL database and named in the format of "plate number-collection time-device number" for easy traceability later.

[0022] Step S20: Perform data preprocessing on the three-dimensional point cloud data and extract the flatness feature parameters of the steel plate. Compare the parameters with the preset flatness standard threshold of the steel plate to determine whether the flatness of the steel plate is qualified.

[0023] Specifically, step S20, which involves preprocessing the 3D point cloud data and extracting the flatness feature parameters of the steel plate, includes: Noise removal: A statistical filtering algorithm is used to select 15-20 neighboring points for each point cloud data, calculate the mean and standard deviation of the distance from the neighboring points to the point, and remove noise points such as dust and scanning interference points whose distance is greater than 1.5 to 2.0 times the mean standard deviation. Point cloud registration: Using the reference hole of the steel plate of the electric tricycle as the positioning reference, with a diameter of 10-15mm and a position accuracy of ±0.1mm, the coordinates of the center of the reference hole are identified by the camera image, and then the point cloud data is aligned with the center coordinates to achieve point cloud registration. The registration error is ≤0.03mm. Data downsampling: A voxel grid downsampling algorithm is adopted to set the voxel size of the point cloud data to 0.05mm, which reduces the amount of data while ensuring accuracy and improves the efficiency of subsequent feature extraction. Reference plane fitting: Select reference points on the edge of the steel plate from the 3D point cloud data after point cloud registration, with a number of ≥20, and use the least squares method to fit the reference plane with a fitting error ≤0.02mm; Deviation calculation: Calculate the vertical distance from all point cloud data to the reference plane. A negative distance indicates a depression, denoted as d_neg, and a positive distance indicates a convexity, denoted as d_pos; Characteristic parameters are determined as follows: the maximum indentation depth, the maximum protrusion height, and the flatness error value are determined, where the maximum indentation depth D_neg=|min(d_neg)|, the maximum protrusion height D_pos=max(d_pos), and the flatness error value F=D_neg+D_pos; Acceptance criteria: When the flatness error value F is less than or equal to the preset standard threshold for flatness of the steel plate, for example, the preset threshold for the main beam steel plate of the electric tricycle frame is 0.8 mm and the threshold for the connecting plate is 1.0 mm, the flatness of the steel plate is deemed acceptable. When the flatness error value F is greater than the preset standard threshold for flatness of the steel plate, the flatness of the steel plate is deemed unacceptable.

[0024] Step S30: When the flatness of the steel plate is qualified, the inspection is passed. When the flatness of the steel plate is unqualified, a flatness deviation-welding parameter mapping model is constructed based on historical steel plate flatness inspection data and steel plate welding process data. The flatness deviation value of the current unqualified steel plate is input into the model to obtain the compensation welding parameters.

[0025] Specifically, step S30, which involves constructing the flatness deviation-welding parameter mapping model, includes: Dataset construction: Collect historical welding data of steel plates of electric tricycles, including 500-1000 sets of flatness feature parameters, namely maximum indentation depth D_neg, maximum protrusion height D_pos and flatness error value F, and corresponding welding parameters, including current, voltage, speed and preheating temperature, and divide them into training set and validation set in a 7:3 ratio; Model structure design: The flatness deviation-welding parameter mapping model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is set to 3, and each node corresponds to the maximum indentation depth D_neg, the maximum protrusion height D_pos, and the flatness error value F, respectively. The hidden layer is set to 2 layers, with 12 nodes in the first layer and 8 nodes in the second layer. The number of nodes in the output layer is set to 4, and each node corresponds to the current, voltage, speed, and preheating temperature, respectively. The activation function is the Sigmoid function. Model training optimization: A backpropagation (BP) neural network is used to train the flatness deviation-welding parameter mapping model. Gradient descent is used to optimize the network weights. The learning rate and training epochs are set as follows: the learning rate is set to 0.01-0.05 and the training epochs are set to 1000-2000. The training set is divided in the dataset construction step for training. After each training epoch, the validation set is divided in the dataset construction step to test the model output accuracy. When the validation set error is ≤5%, training is stopped and the model parameters are saved.

[0026] Step S40: Weld steel plates with qualified flatness according to preset welding parameters, and weld steel plates with unqualified flatness according to the obtained compensation welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensation welding parameters obtained in step S30 are dynamically corrected by the PID adaptive algorithm to complete the welding.

[0027] Specifically, step S40, which involves dynamically correcting the compensated welding parameters obtained in step S30 using a PID adaptive algorithm, includes the following steps: Deviation definition: Assume the target deformation of the steel plate is 0, that is, the steel plate has no deformation, and the real-time deformation is ΔX. Define the PID input deviation e = ΔX - 0 = ΔX. PID parameter calculation: Set the proportional coefficient Kp, with a value range of [5,10], the integral coefficient Ki, with a value range of [0.1,0.5], and the derivative coefficient Kd, with a value range of [0.05-0.2]. Calculate the PID output control quantity ΔU based on the PID input deviation e. The calculation formula is shown in equation (1). (1) in The deviation integral represents the sum of all values ​​of 'e' over a period of time. For example, in the welding process, it represents the cumulative deformation deviation of the steel plate during this period, avoiding the neglect of persistent small deviations. de / dt is the deviation derivative, which represents the rate of change of the deformation deviation of the steel plate. For example, if 'e' changes from 0.1mm to 0.2mm in 1 second, de / dt is 0.1mm / s, reflecting whether the deformation is slowing down or speeding up. Parameter correction: Welding parameters are corrected according to the control quantity ΔU. When the real-time deformation ΔX > 0, i.e. the steel plate bulges, the welding current is reduced and the welding speed is increased. The current reduction is ΔI = -0.2 × ΔU, and the welding speed change is ΔV = 0.5 × ΔU. When ΔX < 0, i.e. the steel plate is concave, the preheating temperature is increased and the welding speed is reduced. The preheating temperature increase is ΔT = 0.3 × ΔU, and the welding speed change is ΔV = -0.5 × ΔU. At the same time, the corrected parameters must be within the parameter range set in step S30. Adaptive adjustment: The average deviation is calculated every 10 correction cycles. When the average deviation is greater than 0.1 mm, Kp and Ki are increased by 10%. When the average deviation is less than 0.05 mm, Kp and Kd are decreased by 5%. When 0.05 mm ≤ average deviation ≤ 0.1 mm, Kp and Ki are kept unchanged to ensure a balance between the dynamic response speed and stability of the algorithm.

[0028] Step S50: After welding is completed, the three-dimensional point cloud data of the steel plate is collected again by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, the flatness error value of the steel plate after welding is calculated. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified. The deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, return to steps S30 and S40 to recalculate the welding parameters and perform a second welding until the flatness is qualified.

[0029] Specifically, step S50, which involves entering the deviation data and compensation parameters into the database to update the model, includes: Data screening: Qualified compensation data within the past 3 months are screened from the database, with a sample size of ≥100 groups, and outlier data, i.e., data with a deviation value >3 times the standard deviation, are removed; Model fine-tuning: Add the selected qualified compensation data as new data to the training set, keep the model network structure unchanged, set the training rounds and retrain, such as 50-100 rounds, and update the network weights; Accuracy verification: The model with updated network weights is tested on the validation set data. When the test error is ≤5%, the new model is saved. When the error is >5%, the number of training rounds is increased until the error is acceptable, ensuring that the model adapts to the changes in the welding process of steel plates for electric tricycles.

[0030] In addition, such as Figure 2As shown, in one embodiment of the present invention, a steel plate flatness detection and welding parameter dynamic compensation system is proposed. The system includes: Visual inspection module: used to construct a visual inspection unit for the flatness of steel plates and to collect three-dimensional point cloud data of steel plates used for electric tricycle frames; Data preprocessing module: used to preprocess 3D point cloud data and extract steel plate flatness feature parameters, compare them with preset steel plate flatness standard thresholds, and determine whether the steel plate flatness is qualified; Flatness Deviation-Welding Parameter Mapping Model Construction Module: When the flatness of the steel plate is qualified, it is inspected; when the flatness of the steel plate is unqualified, it constructs a flatness deviation-welding parameter mapping model based on historical steel plate flatness inspection data and steel plate welding process data, and inputs the flatness deviation value of the current unqualified steel plate into the model to obtain the compensation welding parameters. Welding parameter dynamic compensation module: It is used to weld steel plates with qualified flatness according to preset welding parameters, and to weld steel plates with unqualified flatness according to the obtained compensated welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensated welding parameters obtained in step S30 are dynamically corrected through PID adaptive algorithm to complete the welding. The steel plate flatness secondary inspection module is used to collect three-dimensional point cloud data of the steel plate again after welding by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, it calculates the flatness error value of the steel plate after welding. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified, and the deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, it returns to the flatness deviation-welding parameter mapping model construction module and the welding parameter dynamic compensation module to recalculate the welding parameters and perform secondary welding until the flatness is qualified.

[0031] This application provides a steel plate flatness detection and welding parameter dynamic compensation system, which employs a steel plate flatness detection and welding parameter dynamic compensation method as described in the above embodiments. This system solves the technical problems of low accuracy and low efficiency in traditional steel plate flatness detection methods. Compared with the prior art, the beneficial effects of the steel plate flatness detection and welding parameter dynamic compensation system provided in this application are the same as those of the steel plate flatness detection and welding parameter dynamic compensation method provided in the above embodiments. Furthermore, other technical features of the steel plate flatness detection and welding parameter dynamic compensation system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0032] This application provides a steel plate flatness detection and welding parameter dynamic compensation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steel plate flatness detection and welding parameter dynamic compensation method in the above embodiment 1.

[0033] In one embodiment of the present invention, a steel plate flatness detection and welding parameter dynamic compensation device may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), etc., and fixed terminals such as digital TVs, desktop computers, etc. The described steel plate flatness detection and welding parameter dynamic compensation device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0034] The aforementioned steel plate flatness detection and welding parameter dynamic compensation device may include a processing system (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) or a program loaded from a storage system into random access memory (RAM). The RAM also stores various programs and data required for the operation of the steel plate flatness detection and welding parameter dynamic compensation device. The processing system, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus. Typically, the following systems can be connected to the I / O interface: input systems including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output systems including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage systems including, for example, magnetic tapes, hard disks, etc.; and communication systems. The communication system allows the steel plate flatness detection and welding parameter dynamic compensation device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a steel plate flatness inspection and dynamic compensation device for welding parameters with various systems, it should be understood that it is not required to implement or have all the systems shown. Alternatively, more or fewer systems may be implemented.

[0035] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication system, or installed from a storage system, or installed from a ROM. When the computer program is executed by a processing system, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0036] This application provides a steel plate flatness detection and welding parameter dynamic compensation device, which employs a steel plate flatness detection and welding parameter dynamic compensation method as described in the above embodiments. This solves the technical problems of low accuracy and low efficiency in traditional steel plate flatness detection methods. Compared with the prior art, the beneficial effects of the steel plate flatness detection and welding parameter dynamic compensation device provided in this application are the same as those of the steel plate flatness detection and welding parameter dynamic compensation method provided in the above embodiments. Furthermore, other technical features of this steel plate flatness detection and welding parameter dynamic compensation device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0037] The various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0038] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for detecting the flatness of a steel plate and dynamically compensating welding parameters.

[0039] The computer program product provided in this application can solve the technical problems of low accuracy and low efficiency in traditional steel plate flatness detection methods. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the steel plate flatness detection and welding parameter dynamic compensation method provided in the above embodiments, and will not be repeated here.

[0040] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting the flatness of steel plates and dynamically compensating welding parameters, characterized in that, The method includes the following steps: Step S10: Construct a visual inspection unit for the flatness of the steel plate and collect 3D point cloud data of the steel plate used for the frame of the electric tricycle; Step S20: Perform data preprocessing on the three-dimensional point cloud data and extract the flatness feature parameters of the steel plate, compare them with the preset steel plate flatness standard threshold, and determine whether the flatness of the steel plate is qualified. Step S30: When the flatness of the steel plate is qualified, the inspection is passed; when the flatness of the steel plate is unqualified, a flatness deviation-welding parameter mapping model is constructed to obtain the compensation welding parameters. Step S40: Weld steel plates with qualified flatness according to preset welding parameters, and weld steel plates with unqualified flatness according to the obtained compensation welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensation welding parameters obtained in step S30 are dynamically corrected by PID adaptive algorithm to complete the welding. Step S50: After welding is completed, the three-dimensional point cloud data of the steel plate is collected again by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, the flatness error value of the steel plate after welding is calculated. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified. The deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, return to steps S30 and S40 to recalculate the welding parameters and perform a second welding until the flatness is qualified.

2. The method for detecting the flatness of steel plates and dynamically compensating welding parameters according to claim 1, characterized in that, The step of constructing the steel plate flatness visual inspection unit in step S10 includes: Equipment layout: Two line laser scanners are symmetrically installed above the motion platform, with their scanning directions overlapping to cover the steel plate surface. An industrial camera is installed between the scanners, with its lens axis perpendicular to the steel plate surface. Coordinate calibration: A standard calibration plate is used to unify the coordinates of the scanner and camera, establishing a world coordinate system; Motion control: The PLC controller controls the motion platform to move the steel plate, and simultaneously triggers the scanner and industrial camera to collect three-dimensional data of the steel plate surface.

3. The method for detecting the flatness of steel plates and dynamically compensating welding parameters according to claim 1, characterized in that, The steps in step S20, which involve preprocessing the three-dimensional point cloud data and extracting the flatness feature parameters of the steel plate, include: Reference plane fitting: Select reference points on the edge of the steel plate in the 3D point cloud data, and fit the reference plane using the least squares method; Deviation calculation: Calculate the vertical distance from all point cloud data to the reference plane. A negative distance indicates a depression, denoted as d_neg, and a positive distance indicates a convexity, denoted as d_pos; Characteristic parameters are determined as follows: the maximum indentation depth, the maximum protrusion height, and the flatness error value are determined, where the maximum indentation depth D_neg=|min(d_neg)|, the maximum protrusion height D_pos=max(d_pos), and the flatness error value F=D_neg+D_pos; Acceptance criteria: When the flatness error value F is less than or equal to the preset standard threshold for flatness of steel plate, the flatness of the steel plate is deemed acceptable; when the flatness error value F is greater than the preset standard threshold for flatness of steel plate, the flatness of the steel plate is deemed unacceptable.

4. The method for detecting the flatness of steel plates and dynamically compensating welding parameters according to claim 1, characterized in that, The step of constructing the flatness deviation-welding parameter mapping model in step S30 includes: Dataset Construction: Collect historical welding data of steel plates from electric tricycles, including flatness feature parameters, namely maximum indentation depth D_neg, maximum protrusion height D_pos, and flatness error value F, and corresponding welding parameters, including current, voltage, speed, and preheating temperature, and divide them into training set and validation set in a 7:3 ratio; Model structure design: The flatness deviation-welding parameter mapping model includes an input layer, a hidden layer and an output layer. The number of nodes in the input layer is set to 3, and each node corresponds to the maximum indentation depth D_neg, the maximum protrusion height D_pos and the flatness error value F, respectively. The number of hidden layers is set to 2, and the number of nodes in the output layer is set to 4. Each node corresponds to the current, voltage, speed and preheating temperature, respectively. The activation function is the Sigmoid function. Model training optimization: A BP neural network is used to train the flatness deviation-welding parameter mapping model. Gradient descent is used to optimize the network weights. The learning rate and training rounds are set. The training set is divided in the dataset construction step for training. After each training round, the validation set is divided in the dataset construction step to test the model output accuracy. When the validation set error is ≤5%, training is stopped and the model parameters are saved.

5. The method for detecting the flatness of steel plates and dynamically compensating welding parameters according to claim 1, characterized in that, The step of dynamically correcting the compensated welding parameters obtained in step S30 using a PID adaptive algorithm in step S40 includes: Deviation definition: Assume the target deformation of the steel plate is 0, that is, the steel plate has no deformation, and the real-time deformation is ΔX. Define the PID input deviation e = ΔX - 0 = ΔX. PID parameter calculation: Set the proportional coefficient Kp, integral coefficient Ki, and derivative coefficient Kd, with a value range of [0.05-0.2]. Calculate the PID output control quantity ΔU based on the PID input deviation e, as shown in equation (1): （1） in The integral of the deviation represents the sum of all e over a period of time, and de / dt is the differential of the deviation, representing the rate of change of the deformation deviation of the steel plate. Parameter correction: Welding parameters are corrected based on the control quantity ΔU. When the real-time deformation quantity ΔX > 0, the welding current is reduced and the welding speed is increased. The current reduction is ΔI = -0.2 × ΔU, and the welding speed change is ΔV = 0.5 × ΔU. When ΔX < 0, the preheating temperature is increased and the welding speed is reduced. The preheating temperature increase is ΔT = 0.3 × ΔU, and the welding speed change is ΔV = -0.5 × ΔU. Adaptive adjustment: The average deviation is calculated every 10 correction cycles. When the average deviation is greater than 0.1 mm, Kp and Ki are increased by 10%. When the average deviation is less than 0.05 mm, Kp and Kd are decreased by 5%. When 0.05 mm ≤ average deviation ≤ 0.1 mm, Kp and Ki are kept unchanged.

6. The method for detecting the flatness of steel plates and dynamically compensating welding parameters according to claim 1, characterized in that, The step S50, which involves entering the deviation data and compensation parameters into the database to update the model, includes: Data filtering: Filter eligible compensation data from the database within the past 3 months and remove abnormal data; Model fine-tuning: Add the selected qualified compensation data as new data to the training set, keep the model network structure unchanged, set the training rounds and retrain, and update the network weights; Accuracy verification: Test the model with updated network weights on the validation set data. Save the new model when the test error is ≤5%, and increase the number of training rounds when the error is >5% until the error is acceptable.

7. A steel plate flatness detection and welding parameter dynamic compensation system, characterized in that, The method for detecting the flatness of steel plates and dynamically compensating welding parameters as described in claim 1 includes: Visual inspection module: used to construct a visual inspection unit for the flatness of steel plates and to collect three-dimensional point cloud data of steel plates used for electric tricycle frames; Data preprocessing module: used to preprocess 3D point cloud data and extract steel plate flatness feature parameters, compare them with preset steel plate flatness standard thresholds, and determine whether the steel plate flatness is qualified; Flatness Deviation-Welding Parameter Mapping Model Construction Module: This module is used to construct a flatness deviation-welding parameter mapping model to obtain compensating welding parameters when the flatness of the steel plate is qualified and when the flatness of the steel plate is unqualified. Welding parameter dynamic compensation module: It is used to weld steel plates with qualified flatness according to preset welding parameters, and to weld steel plates with unqualified flatness according to the obtained compensated welding parameters. During the welding process, the weld temperature, current and steel plate deformation are collected in real time. The compensated welding parameters obtained in step S30 are dynamically corrected through PID adaptive algorithm to complete the welding. The steel plate flatness secondary inspection module is used to collect three-dimensional point cloud data of the steel plate again after welding by the steel plate flatness visual inspection unit. After completing the same preprocessing steps, it calculates the flatness error value of the steel plate after welding. When the error value is ≤ the preset steel plate flatness standard threshold, the compensation is deemed qualified, and the deviation data and compensation parameters are entered into the database to update the model. When the error value is > the preset steel plate flatness standard threshold, it returns to the flatness deviation-welding parameter mapping model construction module and the welding parameter dynamic compensation module to recalculate the welding parameters and perform secondary welding until the flatness is qualified.

8. A steel plate flatness detection and welding parameter dynamic compensation device, characterized in that, include: The system includes a memory, a processor, and a steel plate flatness detection and welding parameter dynamic compensation program stored in the memory and executable on the processor. When the processor executes the steel plate flatness detection and welding parameter dynamic compensation program, it implements a steel plate flatness detection and welding parameter dynamic compensation method as described in any one of claims 1 to 6.

9. A computer program product, characterized in that, The computer program product includes a steel plate flatness detection and welding parameter dynamic compensation program, which, when executed by a processor, implements a steel plate flatness detection and welding parameter dynamic compensation method as described in any one of claims 1 to 6.

Images