Intelligent welding method for machine tool manufacturing

Abstract

The invention discloses an intelligent welding method for machine tool manufacturing, which comprises a multi-sensor fusion module, a welding quality real-time evaluation module, a dynamic path planning module, a self-adaptive parameter adjustment module and a human-computer interaction interface, and the multi-sensor fusion module comprises a visual sensor, a laser ranging sensor and a temperature sensor; the welding quality real-time evaluation module is connected with the dynamic path planning module, the multi-sensor fusion module is integrated into a unified data set based on a defect detection algorithm of deep learning, and the dynamic path planning module is combined with 3D visual scanning to produce three-dimensional point cloud data of a workpiece. The dynamic path planning module generates a welding path through an ant colony optimization algorithm, the adaptive parameter adjustment module adjusts the welding state in combination with data collected by the multi-sensor fusion module, and the training process of the anomaly detection model is optimized in combination with the adaptive learning rate and the regularization technology. And the mechanical welding gun is adjusted and controlled through the human-computer interaction interface.

Description

Technical Field

[0001] The present invention relates to the field of welding technology, and in particular to an intelligent welding method for machine tool manufacturing. Background Art

[0002] Industrial robots are playing an increasingly important role in industrial production. Using robots for welding in production helps to improve the stability of product quality, accelerate the enterprise's response speed to orders and the overall production speed, shorten the delivery cycle. At the same time, in countries with an increasingly serious aging population, using robots to assist or replace manual labor helps to reduce production costs and provide more competitive product prices. As a main type of industrial robots, welding robots have been widely used in the welding field, significantly improving welding quality and efficiency. To solve the problems such as complex processes, poor process connectivity, and chaotic welding management during the welding process, traditional welding control systems emerged. The traditional welding control system realizes the full-process control of the welding process of a single welding machine through software control, collects welding parameters at each stage of welding, and at the same time, through the welding control system, each scattered process section can be integrated to achieve the management of the welding process. However, the traditional welding control system has deficiencies in seam tracking and post-weld quality, and in addition, the traditional welding control is not intelligent and automated enough during the welding process.

[0003] Therefore, it is necessary to provide an intelligent welding method for machine tool manufacturing to solve the above existing technical problems. The intelligent control system for welding has been developed on the basis of the traditional welding control system, mainly adopting the distribution method of a central server and on-site clients, networking the welding machines at the construction site, designing welding channels, collecting welding parameters and presetting specifications, and real-time controlling welding quality. Summary of the Invention

[0004] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title, and such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: An intelligent welding method for machine tool manufacturing, characterized by comprising a multi-sensor fusion module, a real-time welding quality evaluation module, a dynamic path planning module, an adaptive parameter adjustment module, and a human-machine interaction interface. The multi-sensor fusion module includes a vision sensor, a laser range finder sensor, and a temperature sensor. The real-time welding quality evaluation module is connected to the dynamic path planning module. Based on a defect detection algorithm of deep learning, the multi-sensor fusion module forms a unified data set. The dynamic path planning module combines 3D vision scanning to generate three-dimensional point cloud data of the production workpiece, and uses an ant colony optimization algorithm to enable the dynamic path planning module to generate a welding path. The adaptive parameter adjustment module combines the data collected by the multi-sensor fusion module to adjust the welding state, and combines an adaptive learning rate and regularization technology to optimize the training process of the anomaly detection model. The multi-sensor fusion module, the real-time welding quality evaluation module, the dynamic path planning module, the adaptive parameter adjustment module, and the human-machine interaction interface are connected. The robotic arm welding torch is adjusted and controlled through the human-machine interaction interface, and specifically includes the following steps: S1. Fix the workpiece to be welded on the welding platform, and start the robotic arm welding torch through the human-machine interaction interface; S2. Start the multi-sensor fusion module in combination with the dynamic path planning module, set the walking path of the robotic arm welding torch through the human-machine interaction interface, and then the robotic arm welding torch moves among multiple pre-welding points on the workpiece to be welded;

[0007] S3. Collect data during the welding process of the robotic arm welding torch through the multi-sensor fusion module, and at the same time evaluate the quality of the welded joints through the real-time welding quality evaluation module, and transmit the data recorded by the real-time welding quality evaluation module to the background human-machine interaction control unit; S4. Program the data of the dynamic path planning module and the adaptive parameter adjustment module in the human-machine interaction control unit through the background human-machine interaction control unit, and transmit the programming program to the robotic arm welding torch control unit. Then, the control unit controls the drive unit to drive the robotic arm welding torch to perform welding operations.

[0008] As a preferred solution of the intelligent welding method for machine tool manufacturing according to the present invention, the adaptive parameter adjustment module includes closed-loop feedback control of welding current, welding voltage, and welding speed.

[0009] As a preferred solution of the intelligent welding method for machine tool manufacturing according to the present invention, the dynamic path planning module adopts a two-layer optimization mechanism. The first layer generates a rough path based on the geometric features of the workpiece, and the second layer optimizes the welding sequence through thermodynamic simulation.

[0010] As a preferred solution of the intelligent welding method for machine tool manufacturing according to the present invention, the multi-sensor fusion module further includes a vision sensor, a weld tracking sensor, and a temperature sensor.

[0011] As a preferred embodiment of the intelligent welding method for machine tool manufacturing according to the present invention, the method further includes a multi-modal sensing unit, which includes a camera and a lidar. Through the fusion of the camera and the lidar, seam tracking at the 0.05 mm level is achieved. At the same time, combined with the combined algorithm of thermal imaging and arc spectrum, pores and lack of fusion defects are detected in real time.

[0012] As a preferred embodiment of the intelligent welding method for machine tool manufacturing according to the present invention, seam tracking is combined with mathematical modeling of seam formation to analyze the welding heat transfer process, which includes highly concentrated heat source input, mobility of the welding heat source, and complexity of welding heat transfer.

[0013] As a preferred embodiment of the intelligent welding method for machine tool manufacturing according to the present invention, this application adopts mathematical modeling and multiple regression analysis modeling methods, fuses and optimizes the mathematical analysis model based on the multiple regression model, and analyzes the accuracy of the fused model by comparing with actual welding data.

[0014] As a preferred embodiment of the intelligent welding method for machine tool manufacturing according to the present invention, the steps for establishing a multiple regression model are as follows:

[0015] (1) Collect data: including the values of the dependent variable corresponding to the independent variable at different factor levels;

[0016] (2) Establish a model: input the collected data into professional analysis software to establish a regression model;

[0017] (3) Model test: test the key parameters in the multiple regression model, including significance test of regression coefficients, standard error test, and goodness-of-fit test;

[0018] (4) Based on the test results, evaluate the fitting effect of the model to determine whether it has sufficient confidence.

[0019] As a preferred embodiment of the intelligent welding method for machine tool manufacturing according to the present invention, based on the assumption conditions in the welding heat transfer process, if the heat source is simplified as a concentrated heat source according to the characteristics of highly concentrated heat source, different analytical models can be established according to the thickness of the plate, including a moving point heat source analytical model and a moving line heat source model.

[0020] As a preferred embodiment of the intelligent welding device for machine tool manufacturing according to the present invention, it further includes improving the Gaussian heat source and using the Gaussian heat source as the heat input model in heat conduction analysis.

[0021] Advantages of the present invention: By adopting multi-modal perception technology, fusing RGB-D cameras with TOF lidar, the present invention achieves weld tracking accuracy. Combining thermal imaging with an arc spectrum joint algorithm, it can detect porosity and lack of fusion defects in real time. By constructing a twin model and combining it with a fusion optimization model after a multiple regression analysis model, it can more accurately predict arc welding weld formation data, thus realizing the real-time prediction function of the weld formation in the digital twin system. And by improving the temperature distribution function with a Gaussian distributed heat source, it solves the error caused by simplifying the heat source to a concentrated heat source. Based on the digital twin model for intelligent optimization of welding parameters using a BP neural network, it realizes the process parameter optimization function in the welding simulation of the robot workstation digital twin system. Through the analysis of weld size data, the optimization model can automatically adjust welding parameters, reduce human intervention, and realize the intelligence and automation of the welding process, thereby improving welding quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0023] Among them:

[0024] Figure 1 is a schematic flow chart of an intelligent welding method for machine tool manufacturing provided by an embodiment of the present invention;

[0025] Figure 2 is a flow chart of an intelligent optimization model of an intelligent welding method for machine tool manufacturing provided by an embodiment of the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments.

[0027] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0029] Example 1

[0030] Reference Figure 1 , according to an embodiment of the present invention, an intelligent welding method for machine tool manufacturing, characterized in that it includes a multi-sensor fusion module, a real-time welding quality evaluation module, a dynamic path planning module, an adaptive parameter adjustment module and a human-machine interaction interface. The multi-sensor fusion module includes a vision sensor, a laser range finder sensor and a temperature sensor; the real-time welding quality evaluation module is connected to the dynamic path planning module, and based on a defect detection algorithm of deep learning, the multi-sensor fusion module is used as a unified data set. The dynamic path planning module combines 3D vision scanning to generate three-dimensional point cloud data of the production workpiece, and through the ant colony optimization algorithm, the dynamic path planning module generates a welding path. The adaptive parameter adjustment module combines the data collected by the multi-sensor fusion module to adjust the welding state, combines the adaptive learning rate and regularization technology to optimize the training process of the anomaly detection model. The multi-sensor fusion module, the real-time welding quality evaluation module, the dynamic path planning module, the adaptive parameter adjustment module and the human-machine interaction interface are connected, and the manipulator welding torch is adjusted and controlled through the human-machine interaction interface, specifically including the following steps: S1. Fix the workpiece to be welded on the welding platform and start the manipulator welding torch through the human-machine interaction interface;

[0031] S2. Start the multi-sensor fusion module in combination with the dynamic path planning module, set the walking path of the manipulator welding torch through the human-machine interaction interface, and then the manipulator welding torch moves at multiple pre-welding points on the workpiece to be welded;

[0032] S3. Collect data during the welding process of the robotic arm welding torch through the multi-sensor fusion module. At the same time, evaluate the quality of the welded solder joints through the welding quality real-time evaluation module, and transmit the data recorded by the welding quality real-time evaluation module to the background human-machine interaction control unit; S4. Program the data of the dynamic path planning module and the adaptive parameter adjustment module in the human-machine interaction control unit through the background human-machine interaction control unit, and transmit the programmed program to the robotic arm welding torch control unit. Subsequently, control the driving unit through the control unit to drive the robotic arm welding torch to perform welding operations. Specifically, the adaptive parameter adjustment module includes closed-loop feedback control of welding current, welding voltage, and welding speed. It can dynamically optimize welding parameters (such as current, voltage, wire feeding speed, welding path, etc.) according to real-time working conditions, thereby improving welding quality and efficiency. During the welding process of the robotic arm welding torch, the adaptive parameter adjustment module automatically changes the welding state, records the data of current, voltage, and speed during the welding process and performs closed-loop feedback to the background control unit, and then evaluates the quality after welding through the welding quality real-time evaluation module. If it is unqualified, adjust the parameters of the robotic arm welding torch. The sensor and data acquisition module includes: current / voltage sensor, temperature sensor (including infrared or thermocouple), vision sensor (camera, laser scanner), weld tracking sensor (ultrasonic or laser ranging), gas flow / pressure sensor. Its function is to collect key data during the welding process in real time, such as molten pool shape, arc characteristics, weld position, ambient temperature, etc., providing an input basis for parameter adjustment.

[0033] The data processing and analysis module uses signal filtering algorithms (to eliminate noise), feature extraction algorithms (such as weld pool width, weld deviation), and data fusion technologies (multi-sensor information integration). By preprocessing the original data, it extracts key features (such as weld pool dynamic behavior, weld offset) to provide structured information for subsequent decision-making. The parameter decision-making module uses control algorithms (PID, fuzzy logic, model predictive control), machine learning models (such as neural networks, reinforcement learning), and based on real-time data and preset goals (such as welding quality, efficiency), dynamically generates an optimal parameter adjustment plan. For example: Current / voltage adjustment: To cope with changes in plate thickness or thermal deformation; Welding speed optimization: To adapt to weld path deviation; Wire feeding speed matching: To ensure the stability of the deposited metal. The actuator module uses servo motors (to adjust the position of the welding torch), inverter power supplies (to adjust current / voltage), wire feeders (to control the wire feeding speed), and robotic arm motion controllers (to adjust the welding path) to convert the output instructions of the decision-making module into physical actions and achieve real-time parameter adjustment. The feedback control loop includes closed-loop control logic (to compare the target with the actual value in real time) and a dynamic compensation mechanism (such as heat input compensation). By continuously monitoring the difference between the output result and the expected goal, it corrects the parameter adjustment strategy to ensure system stability. For example: When the weld pool temperature is too high, automatically reduce the current or increase the welding speed; When the weld deviates, adjust the robotic arm path tracking accuracy. The user interaction and monitoring interface is an HMI (human-machine interface) with data visualization tools (such as dynamic display of the weld pool) and a parameter manual override function, allowing operators to monitor the system status, intervene in the parameter adjustment strategy, and set process constraint conditions (such as maximum current limit). Additionally, there is a safety protection module. By combining anomaly detection algorithms with the safety protection module, it can perform anomaly detection and emergency stop on the robotic arm, such as arc interruption, gas leakage, etc. The safety protection module triggers protection measures when detecting abnormal working conditions (such as parameter overrun, equipment failure) to avoid equipment damage or welding defects. At the same time, the control unit has a system calibration and self-learning module, which includes offline calibration tools, historical data storage and analysis, and model online update functions; By accumulating data over a long period, it optimizes the decision-making model to adapt to new materials, new processes, or environmental changes and improve the generalization ability of the system.

[0034] To improve the rationality of the dynamic path planning module, a two-layer optimization mechanism is adopted. The first layer generates a rough path based on the geometric features of the workpiece, and the second layer optimizes the welding sequence through thermodynamic simulation. The first step is geometric feature extraction for the 3D CAD model or scanned point cloud data of the workpiece. It can identify the weld types (butt joint, fillet weld, lap joint, etc.); extract geometric features such as weld position, length, and groove shape; and mark complex areas (such as corners, curved surfaces, and intersections of multiple welds). The second step is path topology planning to generate a continuous trajectory covering all welds. Through graph theory optimization: decompose the welds into nodes and connect them through the shortest path algorithm (such as Dijkstra); use the block strategy to plan complex geometric bodies in blocks (such as segmented welding of symmetric structures); combine the kinematic constraints of the manipulator with the torch pose (TCP pose) and joint movement range to generate a collision-free path. The third step is the output of the rough path, which can generate a geometrically feasible welding path sequence (such as the coordinate points of the torch movement, welding direction, and segmented marks); the second layer is the optimization of the welding sequence based on thermodynamic simulation. Predict the heat accumulation and deformation of different welding sequences through simulation, and select the optimal sequence to minimize the residual stress and deformation. The first step is thermo-mechanical coupling modeling for the model input, including the division of welding segments in the rough path, material properties (thermal conductivity, coefficient of thermal expansion), and process parameters (current, voltage, speed). Establish a three-dimensional transient heat conduction model through finite element analysis to simulate the evolution of the temperature field. The second step is welding sequence optimization. The optimization variable adjusts the execution order of each welding segment in the rough path. The optimization algorithm is heuristic search: the genetic algorithm (GA) generates candidate sequences and evaluates and iteratively optimizes the simulation results; the symmetry principle preferentially adopts an alternating welding sequence for symmetric structures to offset the thermal stress. Dynamic parameter adjustment is carried out through the human-machine interaction control unit.

[0035] Collaborative optimization: While adjusting the sequence, the welding parameters may be slightly adjusted (for example, the current in section B is reduced by 10% to balance the heat input); the final welding sequence and supporting parameters (such as section sequence, speed, energy) are adjusted according to the output results. The dynamic path planning module uses a two-layer cooperation mechanism for data transfer. The first layer outputs a rough path (geometric segmentation) → the second layer uses it as the input for simulation optimization; if the second layer finds that a certain path will inevitably cause serious deformation (such as single-sided welding of a long straight weld), it may trigger the first layer to re-segment (such as splitting it into multiple sections for reverse welding). Iterative optimization is carried out according to different materials. The iterative optimization process is geometric planning → thermodynamics optimization → output the final path; and a feedback mechanism is used for supervision. If the simulation shows that the deformation exceeds the limit, the second layer can require the first layer to adjust the path topology (such as increasing the heat dissipation gap). For fixed workpieces, thermodynamic optimization is completed in advance through offline pre-computation; online adjustment is carried out according to the welding conditions on the welding site: for dynamic working conditions (such as temporary weld changes), a lightweight model is used for rapid response. The dynamic path planning module can reduce the computational complexity through geometric and thermodynamics decoupling and hierarchical processing; it can meet both path feasibility and welding quality by using multi-objective balance; in addition, it has strong adaptability and can be compatible with complex workpieces and multiple processes.

[0036] In addition, the multi-sensor fusion module further includes a vision sensor, a weld tracking sensor, and a temperature sensor. It also includes a multi-modal perception unit. The multi-modal perception unit includes a camera and a lidar. Through the fusion of the camera and the lidar, weld tracking at the 0.05 mm level can be achieved. At the same time, combined with the combined algorithm of thermal imaging and arc spectrum, pores and lack of fusion defects can be detected in real time. The weld tracking combines with the mathematical modeling of weld formation to analyze the welding heat transfer process. The welding heat transfer process includes highly concentrated heat source input, the mobility of the welding heat source, and the complexity of welding heat transfer. For the heat conduction problem in general welding, when the thermal conductivity, density ρ, and specific heat capacity c of the object are known p The differential equation is shown in Equation (1.1):

[0037]

[0038] where λ, ρ, and c in the equation p are all functions that vary with temperature. If it is assumed to be a constant, Equation (1.1) can be simplified to:

[0039]

[0040] To solve Equation (1.2), boundary conditions and initial conditions of the material, such as the boundary conditions and initial conditions of the material, must be added. The welding heat transfer process is a complex three-dimensional heat conduction process. It is mainly manifested in:

[0041] (1) Highly concentrated heat source input

[0042] The heat source input during welding comes from the end of the welding torch, which causes only local areas of the welded part to be affected by the heat source during the heating process, and the overall heat absorption and dissipation of the workpiece are extremely uneven. Generally, the temperature change is relatively large in the area around the welding torch, and as the distance from the welding torch increases, the temperature change becomes smaller and smaller. Moreover, since the heating and temperature rise process of the welded part is relatively fast, the gradient of the temperature change curve during the heating process is relatively large.

[0043] (2) Mobility of the welding heat source

[0044] Since the welding torch always moves during the welding process, the welding heat source also remains in a moving state. This causes the welding heat-affected zone to constantly change. The points in the heat-affected zone are heated and their temperature rises, while the points that leave the heat-affected zone cool down rapidly, resulting in the solution process finally only reaching a "quasi-steady state".

[0045] (3) Complexity of welding heat transfer

[0046] Since the welding torch moves, the molten pool during welding also remains in a moving state. Inside the molten pool, due to the phase change of the welded part, heat transfer mainly occurs by convective heat transfer at this time; in the non-molten area, it is mainly solid heat conduction. On the boundary elements of the welded part, there are also heat conduction forms such as convective heat dissipation and radiative heat dissipation. At the same time, the change of the stress field in the welded part will also make the temperature calculation more complex. Therefore, in order to achieve real-time prediction of the weld formation, during the welding heat conduction calculation process, assumptions are generally needed to simplify the calculation:

[0047] (1) During the robotic arc welding process, the heat input of the welded part is only affected by the heat source parameters;

[0048] (2) The thermophysical properties of the material do not change with temperature and do not undergo phase change after reaching the melting point; (3) The heat conduction properties of the material are the same in different directions, that is, isotropic.

[0049] This application improves the Gaussian heat source and combines the improved analytical formula of the Gaussian heat source. During the process of the welding torch heating the welding plate, the heat flux is mainly concentrated in a relatively small area, and the radius of this area is the effective radius r of the heating spot h . Since the heat flux on the heating spot can be approximately represented by a Gaussian function. Therefore, the Gaussian heat source is used as the heat input model in the heat conduction analysis. The mathematical expression of the Gaussian heat source is:

[0050]

[0051] In the formula, q(r) is the heat flux density at a distance Q from the center of the circle; Q is the effective power of the welding arc; K is the heat energy concentration coefficient. For the Gaussian heat source, there is also the following expression:

[0052]

[0053] σ q is the Gaussian heat source distribution parameter, and its relationship with the heat concentration coefficient K and the effective radius r of the Gaussian heat source h is as follows:

[0054]

[0055] On an infinitely large plate, assume there is a Gaussian distributed heat source with an effective power of Q on the surface of the welded part. Establish a coordinate system with the center of the Gaussian distributed heat source as the origin. At any point (x', y'0) within the heat source action area, the heat input dQ = q(r')dx'dy'dt, and the temperature change it causes at any point (x, y, z) on the welded part is:

[0056]

[0057] Among them,

[0058]

[0059] Substitute Equation (1.4) into Equation (1.6) to obtain:

[0060]

[0061] Regarding the Gaussian heat source as the sum of the actions of countless point heat sources, the analytical formula for the temperature change caused by the Gaussian distributed heat source instantaneously acting on a thick and large welded part at point (x, y, z) at time t can be obtained:

[0062]

[0063] When the Gaussian distributed heat source moves along the surface of the welded part at a speed v 0 then within its continuous movement time [0, t], by dividing the time period into countless infinitesimals, then integrating them, and transforming to the moving coordinate system, the mathematical expression for the temperature change caused by the Gaussian distributed heat source continuously acting on a thick and large welded part at point (x, y, z) at time t can be obtained:

[0064]

[0065] When σ q = 0, Equation (1.9) becomes the mathematical expression for the heating process of the welding material under the action of a moving point heat source:

[0066] T| t＝0 = δ(x, y, z)

[0067] By combining the Gaussian heat source with the analytical solution model, an improved analytical solution model is obtained. Based on this, an analytical algorithm for calculating the weld size during the welding process can be designed. The real-time prediction model of weld formation has great significance in the welding robot workstation. By predicting the size and shape of the weld in real time, the process parameters can be adjusted in a timely manner during the welding process to ensure the weld quality and consistency. This prediction ability can identify potential welding defects in advance in complex welding tasks, thereby reducing the scrap rate and improving production efficiency and product quality. Digital twin technology provides strong support for this, enabling the close integration of virtual simulation and actual operation to achieve efficient and precise control of the welding process.

[0068] This application uses mathematical modeling and multiple regression analysis modeling methods to fuse and optimize the mathematical analytical model based on the multiple regression model, and analyzes the accuracy of the fused model by comparing it with the actual welding data.

[0069] Multiple regression analysis model

[0070] As a type of regression analysis model, it is different from general regression analysis in the number of independent variables. Generally speaking, a regression model with two or more dependent variables is called a multiple regression model. The mathematical expression for analyzing the correlation between the effective radius and the independent variables using the multiple linear regression model is:

[0071] y = β 0 + β 1 x 1 + β 2 x 2 + … + β k x k + ε

[0072] In the formula, y is the effective radius r of the heat source h ; x i (i = 1, 2 ···, k) are welding process parameters;

[0073] β i (i = 1, 2 ···, k) are regression coefficients; ε is a random error variable, referring to the difference between the actual observed value and the predicted value of the regression model. According to the least squares principle, the estimation of the regression coefficient is β = (X′X) -1 X′y

[0074] The general steps for establishing a multiple regression model are as follows:

[0075] (1) Collect data: including the values of the dependent variable corresponding to the independent variable at different factor levels;

[0076] (2) Establish a model: Input the collected data into professional analysis software such as SPSS, Stata, EXCEL to establish a regression model;

[0077] (3) Model verification: Verify the key parameters in the multiple regression model, including the significance test of regression coefficients, standard error test, goodness-of-fit test, etc.

[0078] (4) Evaluate the fitting effect of the model based on the test results to determine whether it has sufficient confidence. If the effect is poor, problems need to be identified and the model rebuilt. A real-time prediction digital twin model for weld formation based on the multiple regression analysis model is used to solve the real-time generation and size prediction functions of welds in the digital twin system.

[0079] Digital twins have high requirements for real-time performance in order to achieve services such as real-time mapping of physical entities and online monitoring of physical data. For example: The sampling frequency of robot operation data is 20Hz, and the resulting basic time delay between the digital twin and the physical entity is 50ms. This time delay is controllable and can be adjusted according to actual needs. If higher real-time performance is required, data with a sampling frequency of 10kHz or higher can be used to drive the digital twin within the allowable range of Redis read and write speeds, reducing the delay to 0.1ms or less.

[0080] For the acquisition of data on the twin evolution of robotic arms, there are various evaluation methods for welding quality. Commonly used ones include qualitative evaluation of macroscopic morphology, such as whether the weld surface is flat and whether there are cracks and pores; quantitative evaluation of macroscopic morphology, such as the width, height, and penetration depth of the weld; mechanical property evaluation, such as the tensile strength and hardness of the weld; qualitative evaluation of microscopic morphology, such as the morphology of grains in the weld; and microscopic quantitative evaluation, such as the composition ratio of elements in the weld. In this paper, the commonly used evaluation indicators for weld quality, namely weld width, weld height, and weld penetration depth, are selected to evaluate the quality of the welds obtained by welding. Welding quality is affected by various factors. Theoretically, all factors involved in the welding process will affect welding quality. This application also involves the influence of controllable welding parameters on welding quality, including six parameters such as welding current and voltage, welding torch movement speed and angle, wire length, and shielding gas flow rate. At the same time, the influence of three groups of interaction factors on welding quality, such as welding current and welding speed, welding current and wire length, and welding current and welding torch angle, is considered. In the design of welding experiments, it is necessary to determine the combination of welding parameters so that the determined parameter combination is representative among all factor level combinations, enabling the data samples obtained from the experiments to reflect the overall relationship between the parameters and the indicators, and achieving a comprehensive description of the relationship between the parameters and the indicators with fewer experiment times, shorter experiment time, and lower experiment cost. Commonly used methods include the orthogonal design method in the field of engineering experiments and the Pairwise algorithm in the field of software testing. The orthogonal design method is a classic experimental design method. The combinations of experimental parameters designed have the characteristics of "even dispersion and neat comparability", that is, the number of occurrences of different levels of each parameter is equal, and the level combinations of any two parameters are comprehensive and balanced. The Pairwise algorithm is an improvement of the orthogonal experimental method based on statistics. It is believed that research objectives such as quality and faults usually involve at most combinations between two factors. Compared with orthogonal experiments, it can further reduce the number of experiments. The generated parameter combinations still have a certain degree of representativeness and are particularly suitable for experimental designs with mixed levels. To study the influence of three groups of interaction factors, namely welding current and welding speed, wire length, and welding angle, on welding quality, it is necessary to determine the corresponding interaction table while determining the orthogonal table, so as to determine the position of the interaction column in the orthogonal table. The allowable values of the corresponding values of welding test parameters are shown in Table 1:

[0081] Table 1

[0082] Item Welding Current Welding Voltage Torch Speed Torch Angle Wire Length Gas Flow Level 1 120 14 9 60 10 14 Level 2 170 20 11 90 15 20 Level 3 250 24 13 120 20 23

[0083] As can be seen from the above table, all six factors in the experiment are at three levels, and the degrees of freedom of these six factors are all 2, while the degrees of freedom of the three pairs of interaction factors are 4. The mathematical expressions are as follows:

[0084] f A×B =f A·f B = 2×2 = 4

[0085] where f A and f B are the degrees of freedom of two factors A and B respectively, and f A×B is the degree of freedom of their interaction factor. Therefore, there are two schemes when designing an orthogonal experiment to select an orthogonal array: (1) Select an orthogonal array with a single level, where the levels of each column in the table are all 3 and the degrees of freedom are 2. Since the degrees of freedom of each interaction factor are all 4, each requires 2 columns in the orthogonal array. The mathematical expression for the number of columns of the orthogonal array is:

[0086] k min = 6 + 2×3 = 12

[0087] The lowest - specification orthogonal array is L N1 (3 12 );

[0088] (2) Select an orthogonal array with mixed levels, where the levels of the columns where 6 single factors are located are all 3, and the levels of the columns where three pairs of interaction factors are located are all 5. The mathematical expression for the number of columns of the orthogonal array is:

[0089] k min = 6 + 3 = 9

[0090] The lowest - specification orthogonal array is L N2 (3 6 , 5 3 ).

[0091] In the above two schemes, k min is the minimum number of columns. This is because the interaction columns cannot be arranged randomly in the orthogonal array to prevent confounding. Therefore, the positions of the interaction columns need to be determined according to the positions of their two single factors. When there is a conflict with the positions of other factors, an orthogonal array with more columns is required.

[0092] Among them, the acquisition of BP neural network learning samples mainly includes operations such as data acquisition and cleaning, data transformation, and data normalization. Data cleaning refers to discovering and removing some recognizable errors in the data, so as to provide more accurate and reliable data for subsequent data analysis, mainly including invalid values, duplicate values, etc. After removing duplicate and abnormal data through data cleaning, the remaining data is saved as the neural network training and validation data set. Data transformation mainly includes the transformation of non-numerical data. Here, the one-hot encoding technology is mainly used. This encoding technology is mainly applicable to non-numerical data where there is no size relationship between categories, such as plate thickness, joint form, welding pose, etc. Its basic principle is to represent each category as a binary vector, where only one element is 1 and the other elements are 0. Combining with the actual situation of the BP neural network parameter optimization model, when using the one-hot encoding technology, certain adjustments are made to it, that is, the data is transformed into 1 and -1 respectively. Such transformation can improve the feature discrimination of the data, because the difference between -1 and 1 is larger than the difference between 0 and 1, which improves the anti-noise ability of the model.

[0093] Data normalization is a common data preprocessing technology. Through specific mathematical transformation methods, the obtained original data is scaled to a small specific interval according to a certain ratio, so as to adjust the data to a unified scale, make the features in the data comparable, and facilitate better data analysis and processing. The commonly used normalization intervals are [0,1] or [-1,1]. Since in data transformation, the one-hot encoding is used to change non-numerical data into -1 and 1, in order to reduce the amount of normalization operations, the normalization interval is set to [-1,1].

[0094] The formula used for data normalization is:

[0095]

[0096] Normalization processing can effectively eliminate the influence brought by different magnitudes and dimensions between data. At the same time, the solution speed of gradient descent is also improved, so that the convergence speed of the model is correspondingly improved.

[0097] In summary, the weld formation real-time prediction model is based on the Gaussian heat source model and multiple regression analysis, and can predict the size and shape of the weld in real time during the welding process. The welding process parameter optimization model uses the BP neural network to analyze the input weld size data, such as weld width and reinforcement height, and then designs and optimizes more reasonable welding current, welding voltage, and welding speed to ensure welding quality and efficiency.

[0098] It should be understood that, during the development of any actual implementation, such as in any engineering or design project, a large number of specific implementation decisions can be made. Such development efforts may be complex and time-consuming, but for those of ordinary skill in the art who benefit from this disclosure, without undue experimentation, such development efforts will be routine work of design, manufacturing, and production.

[0099] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An intelligent welding method for machine tool manufacturing, characterized in that: It includes a multi-sensor fusion module, a welding quality real-time evaluation module, a dynamic path planning module, an adaptive parameter adjustment module and a human-computer interaction interface. The multi-sensor fusion module includes a visual sensor, a laser ranging sensor and a temperature sensor. The real-time welding quality evaluation module is connected to the dynamic path planning module. The multi-sensor fusion module is unified as a data set based on a deep learning defect detection algorithm. The dynamic path planning module combines 3D visual scanning to produce three-dimensional point cloud data of workpieces. The dynamic path planning module generates a welding path through an ant colony optimization algorithm. The adaptive parameter adjustment module adjusts the welding state in combination with the data collected by the multi-sensor fusion module. The training process of the abnormality detection model is optimized by combining adaptive learning rate and regularization technology. The multi-sensor fusion module, the real-time welding quality evaluation module, the dynamic path planning module, the adaptive parameter adjustment module and the human-computer interaction interface are connected. The robot arm welding gun is adjusted and controlled through the human-computer interaction interface. Specifically, it includes the following steps: S1. Fix the workpiece to be welded on the welding platform and start the robot arm welding gun through the human-computer interaction interface. S2, start the multi-sensor fusion module combined with the dynamic path planning module, set the walking path of the robotic arm welding gun through the human-computer interaction interface, and then move the robotic arm welding gun to multiple pre-welding points on the workpiece to be welded; S3, collecting data during the welding process of the robot welding gun through the multi-sensor fusion module, and at the same time evaluating the quality of the welded welds after welding through the welding quality real-time evaluation module, and transmitting the data recorded by the welding quality real-time evaluation module to the background human-computer interaction control unit; S4. The data of the dynamic path planning module and the adaptive parameter adjustment module are programmed in the human-machine interaction control unit through the human-machine interaction control unit, and the programming program is transmitted to the robot arm welding gun control unit, and then the control unit controls the driving unit to drive the robot arm welding gun to perform welding operations.

2. The intelligent welding method for machine tool manufacturing according to claim 1, characterized in that: The adaptive parameter adjustment module includes closed-loop feedback control of welding current, welding voltage and welding speed.

3. The intelligent welding method for machine tool manufacturing according to claim 1, characterized in that: The dynamic path planning module adopts a two-layer optimization mechanism. The first layer generates a rough path based on the geometric features of the workpiece, and the second layer optimizes the welding sequence through thermodynamic simulation.

4. The intelligent welding method for machine tool manufacturing according to claim 1, characterized in that: The multi-sensor fusion module also includes a visual sensor, a weld tracking sensor and a temperature sensor.

5. The intelligent welding method for machine tool manufacturing according to claim 1, characterized in that: The method also includes a multimodal sensing unit, which includes a camera and a laser radar. Through the cooperation of the camera and the laser radar, 0.05mm level weld tracking is achieved. At the same time, thermal imaging and arc spectrum combined algorithm are combined to detect pores and unfused defects in real time.

6. The intelligent welding method for machine tool manufacturing according to claim 5, characterized in that: Weld tracking is combined with mathematical modeling of weld formation to analyze the welding heat transfer process, which includes the highly concentrated heat source input, the mobility of the welding heat source, and the complexity of the welding heat transfer.

7. The intelligent welding method for machine tool manufacturing according to claim 5, characterized in that: This application adopts mathematical modeling and multivariate regression analysis modeling methods, fuses and optimizes the mathematical analytical model based on the multivariate regression model, and analyzes the accuracy of the fused model by comparing it with actual welding data.

8. The intelligent welding method for machine tool manufacturing according to claim 7, characterized in that: The steps to build a multiple regression model are: (1) Collect data: including the values ​​of the dependent variable corresponding to the independent variable at different factor levels; (2) Model building: Input the collected data into professional analysis software to build a regression model; (3) Model testing: testing the key parameters in the multivariate regression model, including significance test of regression coefficients, standard error test, and goodness of fit test; (4) By testing the results, the model fitting effect is evaluated to determine whether it has sufficient confidence.

9. The intelligent welding method for machine tool manufacturing according to claim 5, characterized in that: Based on the assumptions in the welding heat transfer process, if the heat source is simplified to a concentrated heat source according to its highly concentrated characteristics, different analytical models can be established according to the thickness of the plate. Including moving point heat source analytical model and moving line heat source model.

10. The intelligent welding method for machine tool manufacturing according to claim 1, characterized in that: Also included is an improvement to the Gaussian heat source as the heat input model in heat transfer analysis.

Images