Temperature and upsetting force control method of mirror friction stir welding based on digital twin

Abstract

The present invention discloses a method for controlling the temperature and upsetting force of mirror stir friction welding based on digital twins. The method comprises: constructing a mirror stir friction welding thermal coupling simulation model based on a modified single-sided stir friction welding thermal coupling simulation model; constructing the following neural network models, whose inputs include welding parameters and whose outputs are the initial rotation speed of the stirring head, the upsetting force, the temperature field of the workpiece, and the response time from issuing a command to the upsetting force and the temperature field of the workpiece reaching the next steady-state value: an initial rotation speed prediction model, an upsetting force prediction model, a temperature field prediction model, and a response time prediction model; constructing the following fuzzy controllers, whose inputs include the upsetting force error and the rate of change of the upsetting force error, and whose outputs are the change in the amount of downward pressure and the change in rotation speed: a downsetting amount change fuzzy controller and a speed change fuzzy controller, respectively. The downsetting amount or the stirring head rotation speed is controlled by the two fuzzy controllers. The present invention can achieve stable temperature-upsetting force control in mirror stir friction welding.

Description

Technical Field

[0001] The present invention relates to the technical field of friction stir welding, and in particular to a method for controlling temperature and upsetting force of mirror image friction stir welding based on digital twins. Background Art

[0002] Since its invention in 1991 by the British Welding Institute, friction stir welding (FSW) has garnered widespread attention due to its low welding temperature, low residual stress, and the absence of arc light, smoke, and other pollution during the welding process. Furthermore, it features low input power, which reduces production costs. FSW has been widely adopted as a new solid-state welding technology in various fields, including aerospace. As a thermomechanically coupled welding technique, FSW's temperature and forging force constantly influence the quality of the weld zone. Temperature softens the material and improves its fluidity. Excessively high temperatures cause the material to melt, affecting weld strength. Low temperatures can easily form defects such as grooves and holes. The forging force forges the softened material. Too little forging force can lead to loose weld structure and holes. Too much forging force can cause excessive flash and severe weld thinning. Therefore, maintaining stable temperature and forging force during welding is crucial for ensuring uniform and stable weld quality and improving overall weld strength.

[0003] With the continuous development of friction stir welding control technology, current single-sided friction stir welding temperature-forging force control requires temperature and force measuring instruments without damaging the tool or workpiece. These precision instruments can cost tens or even millions of dollars, are extremely expensive, and can only measure surface temperature. Mirror image friction stir welding, on the other hand, requires two sets of corresponding instruments, which doubles the cost. The huge cost and extremely low price-performance ratio for temperature measurement have seriously hindered researchers' progress. How to develop a low-cost mirror image friction stir welding temperature-forging force control method that can achieve full temperature monitoring of the weld zone, achieve stable temperature and forging force during the welding process, produce uniform and stable weld quality, and improve the overall strength of the weld has become a difficult problem that designers are constantly seeking to solve. Summary of the Invention

[0004] In view of the problems that the existing temperature-upsetting force control is costly and can only measure surface temperature, the present invention provides a mirror stir friction welding temperature and upsetting force control method based on digital twin to solve the technical problems existing in the known technology.

[0005] The present invention solves the technical problems existing in the known technology by adopting a technical solution: a method for controlling the temperature and upsetting force of mirror stir friction welding based on digital twin, comprising the following steps:

[0006] A mirror image friction stir welding thermal coupling simulation model is constructed based on the modified single-sided friction stir welding thermal coupling simulation model; the following neural network models are constructed, whose input includes welding parameters and whose outputs are the initial speed of the stirring head, the upsetting force, the temperature field of the workpiece, and the response time from the issuance of the instruction to the upsetting force and the temperature field of the workpiece reaching the next steady-state value: initial speed prediction model, upsetting force prediction model, temperature field prediction model, response time prediction model; the following fuzzy controllers are constructed, whose input includes the upsetting force error and the upsetting force error change rate and whose outputs are the downsetting amount change and the speed change, respectively: downsetting amount change fuzzy controller and speed change fuzzy controller;

[0007] Thermal experiments and variable parameter experiments were conducted using the mirror image friction stir welding thermal-mechanical coupling simulation model. The experimental data were used to compile corresponding training samples to train the initial speed prediction model, the upsetting force prediction model, the temperature field prediction model, and the response time prediction model.

[0008] The above-mentioned upset force prediction model, temperature field prediction model, and response time prediction model are combined to construct a digital twin prediction model;

[0009] The welding trajectory of the driven stirring head is obtained by using the mirror motion constraint and the welding trajectory of the active stirring head. The welding parameters except the rotation speed and upsetting force in the stable welding stage and the set upsetting force are input into the initial speed prediction model to obtain the initial welding speed of the stirring head.

[0010] The mirror-image friction stir welding robot performs welding operations based on the welding trajectory and initial rotational speed of the active and passive stir heads. It also feeds the current welding parameter values ​​into the digital twin prediction model, which then predicts and outputs the following data: upset force, workpiece temperature field, and the response time from issuing a command until the upset force and workpiece temperature field reach the next steady-state value.

[0011] Based on the above data output by the digital twin prediction model, the mirror friction stir welding robot chooses to use a fuzzy controller for the variation of the downward pressure or the variation of the speed to adjust and control the downward pressure of the stirring head or the speed of the stirring head accordingly, thereby controlling the upsetting force applied by the stirring head and the temperature field of the workpiece.

[0012] Furthermore, the upset force prediction model and the temperature field prediction model are divided into three welding stage sub-models according to the following three stages of the welding process: pressing stage, dwelling stage, and stable welding stage.

[0013] Furthermore, the mirror friction stir welding robot is provided with: a rotating platform, a bracket fixed on the rotating platform, and a laser ranging sensor installed on the bracket; the mirror friction stir welding robot feeds back the following welding parameter values ​​corresponding to different welding stages to the digital twin prediction model: welding stage number k, workpiece thickness th , stirring head spindle speed n, inclination angle q, pressing speed v d , actual downward pressure h s , residence time t s , welding speed v w , the time t that the kth welding stage has been running rk ; where: t h Measured manually or with an instrument, t rk Obtained by the internal timer, v d 、v w The feed rate is determined by the G code, n is obtained by the spindle speed feedback, t s is the set value;

[0014] k: represents the numbers of the three different welding stages in the friction stir welding process. The values ​​of k are 1, 2, and 3, where 1 represents the pressing stage, 2 represents the dwell stage, and 3 represents the stable welding stage.

[0015] The motor drives the turntable, which rotates the laser distance sensor on the bracket. The data measured by the laser distance sensor is fitted into a cone whose top is cut by a plane. Its bottom diameter and generatrix inclination angle are known. The object is cut through by a plane passing through the axis. 50% of the sum of the heights of the two vertices of the two-dimensional figure of the cut surface is the axis length c. When the axis length c equals the distance H from the laser emission point to the top surface of the mixing head along the blade axis, k is set to 1, and the downward pressure stage begins.

[0016] When the axis length is Hv, the downward pressure reaches h, where: v = l + h / cos (q), l is the length of the stirring needle; k is 2, and the dwell stage begins;

[0017] After k is 2, the timing starts and the time reaches t s When k is 3, the welding stage begins;

[0018] q: Inclination angle. The laser ranging sensor rotates one circle to fit a plane in the sensor coordinate system. This is the smallest angle between the plane normal and the straight line on the z-axis of the sensor coordinate system.

[0019] h s : The actual downward pressure is obtained through the laser ranging sensor, h s =(Hcl)*cos(q).

[0020] Furthermore, the training input data of each welding stage sub-model of the upset force prediction model includes: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s, welding speed v w .

[0021] Furthermore, the training input data of each welding stage sub-model of the temperature field prediction model includes: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w , and the grid layer number i along the thickness direction in the mirror image friction stir welding thermal coupling simulation model; the training output data is the temperature data of the i-th grid along the thickness direction in the square area with a length and width of W centered on the stirring head in the corresponding mirror image friction stir welding thermal coupling simulation model.

[0022] Furthermore, the training input data of the response time prediction model includes: workpiece thickness t h , stirring head spindle speed n, inclination angle q, downward pressure h, welding speed v w , change in downward pressure Δh, change in speed Δn.

[0023] Furthermore, the training input data of the initial speed prediction model include: the average value of the upsetting force data F in the stable welding stage a , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w .

[0024] Furthermore, a visualization interface was built using the Unity development platform, combining the initial speed prediction model and the digital twin prediction model. The visualization interface is used to determine the initial speed and display the temperature field of the workpiece welding zone and the upsetting force exerted on the stirring head during the welding process in real time.

[0025] Furthermore, a thermomechanical coupling simulation model of single-sided friction stir welding was established using the fluid-solid coupling method based on ABAQUS; the heat generation and boundary conditions of the simulation model were adjusted according to the experimental data and existing literature data to correct the thermal data of the simulation model; based on the corrected single-sided friction stir welding simulation model, a mirror image friction stir welding thermomechanical coupling simulation model was constructed.

[0026] Furthermore, in the stable welding stage, the mirror friction stir welding robot selects to use the pressure variation fuzzy controller or the speed variation fuzzy controller according to the data output by the digital twin prediction model, which includes the following steps:

[0027] Step S1, set the initial response time Δt = 0; determine whether the time from the last adjustment of the welding parameters exceeds the response time Δt, if so, execute S3; if not, do not adjust the parameters;

[0028] Step S2: When the predicted upsetting force output by the digital twin prediction model is different from the set upsetting force, a downward pressure fuzzy controller or a speed fuzzy controller is selected based on a comparison result between the predicted upsetting force output by the digital twin prediction model and the set upsetting force, and a comparison result between the maximum temperature in the workpiece temperature field output by the digital twin prediction model and 80% of the workpiece melting temperature.

[0029] Step S3: When the ratio of the predicted upset force to the set upset force is different from the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, execute S4; when the ratio of the predicted upset force to the set upset force is the same as the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, execute S5;

[0030] Step S4, sending the upset force error and the upset force error change rate to the speed fuzzy controller to calculate the speed increment, and sending the adjusted speed to the controller to achieve speed regulation;

[0031] In step S5, the upset force error and the upset force error change rate are sent to the down-pressure fuzzy controller to calculate the adjusted down-pressure amount, and determine whether the adjusted down-pressure amount is between 0-0.5. If so, the adjusted down-pressure amount is sent to the mirror stir friction welding robot controller to implement the down-pressure amount adjustment; if not, return to step S4.

[0032] The advantages and positive effects of this invention include: digital twin technology accurately maps the physical entity during the machining process in virtual space. Using limited sensor data, it monitors data that would otherwise be impossible to measure directly, enabling full-scale monitoring of the physical entity during the machining process. Furthermore, based on feedback from the digital twin, it enables real-time predictive control of the physical entity's process parameters. This technology can achieve stable temperature-upsetting force during mirror-image friction stir welding, thereby improving the overall weld quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a structural schematic diagram of the mirror image friction stir welding robot used in the present invention.

[0034] Figure 2 The invention relates to a device for measuring downward pressure and inclination angle.

[0035] Figure 3 It is a mirror image friction stir welding digital twin visualization interface of the present invention.

[0036] Figure 4It is a flow chart of a method for controlling temperature and upsetting force of mirror image friction stir welding based on digital twins of the present invention.

[0037] Figure 5 It is a framework schematic diagram of a digital twin-based mirror stir friction welding temperature and upsetting force control method of the present invention.

[0038] In the figure: 1—workbench; 2—left friction stir welding robot; 3—workpiece; 4—fixture; 5—right friction stir welding robot; 6—base; 7—bracket; 8—rotating platform; 9—laser ranging sensor.

[0039] x1, the x-axis of the left sensor coordinate system; z1, the z-axis of the left sensor coordinate system; o1, the coordinate origin of the left sensor coordinate system; α1, the angle between the z-axis of the left sensor coordinate system and the plane normal fitted by the sensor; x2, the x-axis of the right sensor coordinate system; z2, the z-axis of the right sensor coordinate system; o2, the coordinate origin of the right sensor coordinate system; α2, the angle between the z-axis of the right sensor coordinate system and the plane normal fitted by the sensor.

[0040] k, welding stage number; t h , workpiece thickness; n, stirring head spindle speed; q, inclination angle; v d , pressing speed; h, pressing amount; t s , residence time; v w , welding speed; t rk , the time that the kth welding stage has been running; Δh, the change in the amount of downward pressure; Δn, the change in speed; F a , average value of upsetting force data in stable welding stage; F s , set the initial value of the upset force; i. The grid layer number along the thickness direction in the mirror image stir friction welding thermal coupling simulation model; x, y, z, a, b: represent the end position parameters of the robot stirring head in the workpiece coordinate system, where x, y, z are the end position coordinates, and a and b are the posture coordinates. DETAILED DESCRIPTION

[0041] To further understand the content, features and effects of the present invention, the following embodiments are listed and described in detail with reference to the accompanying drawings:

[0042] The Chinese meanings of the following English words, phrases and abbreviations in this application are as follows:

[0043] TCP / IP: Transmission Control Protocol / Internet Protocol.

[0044] LineGraph: curve graph.

[0045] Unity: Real-time 3D interactive content creation and operation platform.

[0046] Text: text.

[0047] Solidworks: A 3D mechanical design software.

[0048] ABAQUS: A finite element simulation software.

[0049] CEL: Fluid-structure coupling method (a finite element simulation method).

[0050] See Figures 1 to 5 A method for controlling temperature and upsetting force of mirror-image friction stir welding based on digital twins comprises the following steps:

[0051] A mirror image friction stir welding thermal coupling simulation model is constructed based on the modified single-sided friction stir welding thermal coupling simulation model; the following neural network models are constructed, whose input includes welding parameters and whose outputs are the initial speed of the stirring head, the upsetting force, the temperature field of the workpiece, and the response time from the issuance of the instruction to the upsetting force and the temperature field of the workpiece reaching the next steady-state value: initial speed prediction model, upsetting force prediction model, temperature field prediction model, response time prediction model; the following fuzzy controllers are constructed, whose input includes the upsetting force error and the upsetting force error change rate and whose outputs are the downsetting amount change and the speed change, respectively: downsetting amount change fuzzy controller and speed change fuzzy controller;

[0052] Thermal experiments and variable parameter experiments were conducted using the mirror image friction stir welding thermal-mechanical coupling simulation model. The experimental data were used to compile corresponding training samples to train the initial speed prediction model, the upsetting force prediction model, the temperature field prediction model, and the response time prediction model.

[0053] The above-mentioned upset force prediction model, temperature field prediction model, and response time prediction model are combined to construct a digital twin prediction model;

[0054] The welding trajectory of the driven stirring head is obtained by using the mirror motion constraint and the welding trajectory of the active stirring head. The welding parameters except the rotation speed and upsetting force in the stable welding stage and the set upsetting force are input into the initial speed prediction model to obtain the initial welding speed of the stirring head.

[0055] The mirror-image friction stir welding robot performs welding operations based on the welding trajectory and initial rotational speed of the active and passive stir heads. It also feeds the current welding parameter values ​​into the digital twin prediction model, which then predicts and outputs the following data: upset force, workpiece temperature field, and the response time from issuing a command until the upset force and workpiece temperature field reach the next steady-state value.

[0056] Based on the above data output by the digital twin prediction model, the mirror friction stir welding robot chooses to use a fuzzy controller for the variation of the downward pressure or the variation of the speed to adjust and control the downward pressure of the stirring head or the speed of the stirring head accordingly, thereby controlling the upsetting force applied by the stirring head and the temperature field of the workpiece.

[0057] The collection of temperatures at various points within a workpiece is called the temperature field. It is a function of time and space coordinates, reflecting the distribution of temperature in space and time. The variable temperature T is typically a function of the space coordinates (x, y, z) and the time variable t, i.e., T = T(x, y, z, t). This formula describes a three-dimensional, unsteady (transient) temperature field, and the heat conduction occurring in this temperature field is three-dimensional unsteady (transient) heat conduction. A temperature field that does not change with time is called a steady-state temperature field, i.e., T = T(x, y, z), which is three-dimensional steady-state heat conduction. For one-dimensional and two-dimensional temperature fields, the steady-state can be expressed as T = f(x) and T = f(x, y), respectively, and the unsteady-state can be expressed as T = f(x, t) and T = f(x, y, t), respectively.

[0058] Preferably, the upset force prediction model and the temperature field prediction model can be divided into three welding stage sub-models according to the following three stages of the welding process: the pressing stage, the dwell stage, and the stable welding stage.

[0059] Preferably, the mirror friction stir welding robot may include: a workbench 1 , a left friction stir welding robot 2 , and a right friction stir welding robot 5 . A clamp 4 may be provided on the workbench 1 , and the workpiece 3 is fixed on the workbench 1 through the clamp 4 .

[0060] The left or right friction stir welding robot in the mirror friction stir welding robot can be equipped with: a rotating platform 8, a bracket 7 fixed to the rotating platform 8, and a laser ranging sensor 9 installed on the bracket 7; the mirror friction stir welding robot can feed back the following welding parameter values ​​corresponding to different welding stages to the digital twin prediction model: welding stage number k, workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , actual downward pressure h s , residence time t s , welding speed v w , the time t that the kth welding stage has been running rk ; where: t h It can be measured manually or by instruments, rk It can be obtained through the internal timer of the robot human-computer interaction interface, v d 、v w It can be determined by the feed rate set in the G code, n can be obtained by the spindle speed feedback, t s Can be a set value;

[0061] k: represents the numbers of the three different welding stages in the friction stir welding process. The values ​​of k can be 1, 2, or 3, where 1 represents the pressing stage, 2 represents the dwell stage, and 3 represents the stable welding stage.

[0062] The turntable can be driven by a motor to rotate the laser distance sensor 9 on the bracket 7. The data measured by the laser distance sensor 9 can be used to fit a cone whose top is cut by a plane. The bottom diameter and the generatrix inclination angle of the cone are known. The object can be cut by a plane passing through the axis. 50% of the sum of the heights of the two vertices of the two-dimensional figure of the section can be the axis length c; when the axis length c is equal to the distance H from the laser emission point to the top surface of the stirring head along the knife axis direction, k can be taken as 1, and the downward pressure stage begins.

[0063] When the axis length is Hv, the downward pressure h is reached, where: v = l + h / cos (q), l is the length of the stirring needle; k can be 2, and the dwell stage begins.

[0064] After k is 2, the timing starts and the time reaches t s When k is 3, the welding stage begins.

[0065] q: tilt angle. The laser ranging sensor 9 rotates one circle to fit a plane in the sensor coordinate system, which is the smallest angle between the plane normal and the straight line where the z-axis of the sensor coordinate system is located.

[0066] h s : The actual downward pressure can be obtained by the laser distance sensor 9, h s =(Hcl)*cos(q).

[0067] Preferably, the training input data of each welding stage sub-model of the upset force prediction model may include: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w .

[0068] For the forging force prediction model:

[0069] The input variables of the sub-model in the downward pressure stage can be: t h , t r1 (k=1),n,q,v d , h, residence time t s and welding speed v w The value of is 0.

[0070] The input variables of the dwell phase sub-model can be: h , tr2 (k=2),n,q,v d ,h,t s , welding speed v w The value of is 0.

[0071] The input variables of the stable welding stage sub-model can be: t h , t r (r=3),n,q,v d ,h,t s 、v w .

[0072] The output variables of the three sub-models of the upsetting force prediction model are all upsetting force.

[0073] The three sub-models of the upset force prediction model can be trained using the corresponding input / output variables according to the different stages mentioned above.

[0074] Preferably, the training input data of each welding stage sub-model of the temperature field prediction model may include: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w , and the grid layer number i along the thickness direction in the mirror image friction stir welding thermal coupling simulation model; the training output data is the temperature data of the i-th grid along the thickness direction in the square area with a length and width of W centered on the stirring head in the corresponding mirror image friction stir welding thermal coupling simulation model.

[0075] On the welded workpiece, a square area with a length and width of W, centered on the stirring head, can be selected. The temperature field of the rectangular portion of the workpiece taken along the thickness direction can be used as the parameter sampling and prediction object. W can be selected to be 3-5mm larger than the shoulder diameter of the stirring head corresponding to the thickest workpiece.

[0076] That is, for the temperature field prediction model:

[0077] The input variables of the sub-model in the downward pressure stage can be: grid layer number i, t h , t r1 (k=1),n,q,v d , h. Residence time t s and welding speed v w The value of is 0.

[0078] The input variables of the dwell phase sub-model can be: grid layer number i, t h , t r2 (k=2),n,q,v d ,h,ts . Welding speed v w The value of is 0.

[0079] The input variables of the stable welding stage sub-model can be: mesh layer number i, t h , t r3 (k=3),n,q,v d ,h,t s 、v w .

[0080] The output variables of the three sub-models of the temperature field prediction model are: the temperature data of the i-th layer of grid from top to bottom along the thickness direction in the square area with length and width W centered on the stirring head on the welding workpiece.

[0081] The three sub-models of the temperature field prediction model can be trained using the corresponding input / output variables according to the different stages mentioned above.

[0082] Preferably, the training input data of the response time prediction model may include: workpiece thickness t h , stirring head spindle speed n, inclination angle q, downward pressure h, welding speed v w , change in downward pressure Δh, change in speed Δn.

[0083] Preferably, the training input data of the initial speed prediction model may include: the average value of the upset force data F during the stable welding stage a , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w .

[0084] Preferably, the Unity development platform can be used to build a visualization interface in combination with the initial speed prediction model and the digital twin prediction model. The visualization interface can be used to determine the initial speed and display in real time the temperature field of the workpiece welding zone and the upsetting force exerted on the stirring head during the welding process.

[0085] Preferably, a thermomechanical coupling simulation model of unilateral friction stir welding can be established based on ABAQUS using the fluid-solid coupling method; the heat generation and boundary conditions of the simulation model can be adjusted according to experimental data and existing literature data to correct the thermal data of the simulation model; and a mirror image friction stir welding thermomechanical coupling simulation model can be constructed based on the corrected unilateral friction stir welding simulation model.

[0086] Preferably, in the stable welding stage, the method for the mirror friction stir welding robot to select the use of the pressure variation fuzzy controller or the speed variation fuzzy controller according to the data output by the digital twin prediction model may include the following steps:

[0087] In step S1, the initial response time Δt may be set to 0; it is determined whether the time from the last adjustment of the welding parameters exceeds the response time Δt. If so, step S3 is executed; if not, no parameter adjustment is performed.

[0088] Step S2: When the predicted upsetting force output by the digital twin prediction model is different from the set upsetting force, the downforce fuzzy controller or the speed fuzzy controller may be selected based on the comparison result between the predicted upsetting force output by the digital twin prediction model and the set upsetting force, and the comparison result between the highest temperature in the workpiece temperature field output by the digital twin prediction model and 80% of the workpiece melting temperature.

[0089] Step S3: When the ratio of the predicted upset force to the set upset force is different from the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, step S4 may be executed; when the ratio of the predicted upset force to the set upset force is the same as the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, step S5 may be executed;

[0090] Step S4, sending the upset force error and the upset force error change rate to the speed fuzzy controller to calculate the speed increment, and sending the adjusted speed to the controller to achieve speed regulation;

[0091] In step S5, the upset force error and the upset force error change rate are sent to the down-pressure fuzzy controller to calculate the adjusted down-pressure amount, and determine whether the adjusted down-pressure amount is between 0 and 0.5. If so, the adjusted down-pressure amount can be sent to the mirror stir friction welding robot controller to realize the down-pressure amount adjustment; if not, return to step S4.

[0092] The following is a preferred embodiment of the present invention to further illustrate the workflow and working principle of the present invention:

[0093] A method for controlling temperature and upsetting force of mirror-image friction stir welding based on digital twins includes the following specific steps:

[0094] Step 1: A thermomechanical coupling simulation model for single-sided FSW was established using the CEL (fluid-structure interaction) method in ABAQUS. The heat generation and boundary conditions of the simulation model were adjusted based on experimental data and existing literature to calibrate the model's thermal data. Based on the corrected single-sided FSW simulation model, a mirror image FSW thermomechanical coupling simulation model was constructed.

[0095] Step 2: Use the mirror image stir friction welding thermal coupling simulation model to conduct thermal experiments and variable parameter experiments to collect the required data. Thermal experiment: For welding workpieces of different thicknesses, the orthogonal experiment method is used to test the combination of process parameters in the welding process (workpiece thickness t h, stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w ) is designed, and the mirror image stir friction welding thermal coupling simulation model of the corresponding workpiece thickness is used to experiment with the corresponding process parameter combination to generate thermal data of different process parameters and different workpiece thicknesses; variable parameter experiment: for welding workpieces with different workpiece thicknesses, the orthogonal experiment method is used to analyze the process parameters that affect the thermal data of the stable welding stage: stirring head spindle speed n, inclination angle q, downward pressure h, welding speed v w On this basis, the time required for the forging force to reach a steady state after the speed / downward pressure amount changes is studied, and the time data required to reach a steady state after the speed / downward pressure amount changes under different workpiece thicknesses and different process parameters are generated.

[0096] Step 3: Take the average value F of the upset force data in the stable welding stage of each thermal test a , with workpiece thickness t h ,q,h,v w 、F a As input variables, n as output variables, use the above variables to train the initial speed prediction model and build the initial speed prediction model. When using, input t h ,q,h,v w , set the initial value of the forging force F s =F a , the initial speed can be obtained.

[0097] Step 4: For the three welding stages in the friction stir welding process, establish the sub-models of the forging force prediction model, the sub-model of the temperature field prediction model, and the sub-model of the response time prediction model for the pressing stage, the dwell stage, and the stable welding stage. For the above thermal data, extract the temperature data of the i-th layer of grid along the thickness direction of the square area with a length and width of W (W is selected according to the shoulder diameter of the stirring head corresponding to the thickest workpiece, which is slightly larger than 3-5mm) centered on the stirring head on the welding workpiece, and the forging force data of the stirring head as output variables, combined with the input variables: the duration t from the start of the k-th welding stage to the present. rk ; k represents the three different welding stage numbers in the friction stir welding process, k takes values ​​1, 2, and 3, 1 represents the pressing stage, 2 represents the dwell stage, and 3 represents the stable welding stage; workpiece thickness t h , grid layer number i, stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w ; Used to train the forging force prediction model, temperature field prediction model, and response time prediction model.

[0098] The input variables of the down-pressing stage sub-model of the upset force prediction model are: workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , the pressing amount h, the duration from the start of the pressing stage to the present t r1 .

[0099] The input variables of the dwell phase sub-model of the upset force prediction model are: workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , the duration t from the start of the dwell phase to the present r2 .

[0100] The input variables of the welding stage sub-model of the upset force prediction model are: workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w , Duration t from the start of stable welding process to now r3 .

[0101] The output variables of the three sub-models of the upsetting force prediction model are the upsetting force of the stirring head.

[0102] The prediction model of the temperature field in the welding area is based on the input variables of the upset force prediction model. The prediction models at different stages add the input variable i of the number of grid layers along the thickness direction. The output of the sub-models at different stages is the temperature of the i-th layer of grid.

[0103] Using the corresponding input / output variables for each stage, the sub-models of the upsetting force prediction model and the sub-models of the temperature field prediction model are trained to ultimately construct a digital twin prediction model. When the digital twin prediction model predicts, the upsetting force prediction model inputs the input variables for the corresponding stage to obtain the corresponding upsetting force. The temperature field prediction model inputs the corresponding input variables. The input variable i is iterated from 1 to the total number of grid cells m, while other variables remain unchanged. m predictions are performed to obtain the corresponding weld zone temperature field.

[0104] Step 5: Based on the time data obtained from the variable parameter experiment, the workpiece thickness t h , stirring head spindle speed n, inclination angle q, downward pressure h, welding speed v wThe steady-state welding response time neural network is trained to construct a response time prediction model, using the change in downforce Δh, and the change in speed Δn as input variables and the time required for the upset force to reach a steady state after the change as the output variable. Δh and Δn cannot be non-zero at the same time; only one of h and n is changed at a time. When using these input variables, the time required to reach a steady state after the process parameters of the steady-state welding phase are output.

[0105] Step 6: Use the above-mentioned upset force prediction model, temperature field prediction model, and response time prediction model to build a digital twin prediction model. The input of the digital twin prediction model is: welding stage number k, workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w , the time t that the kth welding stage has been running rk , the change in the downward pressure Δh, the change in the rotational speed Δn, and the output is the response time after the parameter change, the upsetting force data and the temperature field at the corresponding moment.

[0106] The selected temperature field is divided equally into a cubic grid with a length, width, and height of 1 mm. The temperature field is then layered along the thickness of the grid, resulting in m layers, where m is the integer obtained by dividing the workpiece thickness by 1 mm. The number of layers is represented by i. Each cube in each layer is numbered j = 1 to 900.

[0107] Therefore, the temperature field can be expressed as a matrix T = [T1, T2, ..., T i ] T ,i=1:m,where T i Represents the temperature array of the i-th layer of the grid in the rectangular area on the workpiece of different thicknesses. The array number corresponds to the grid number, and the temperature of the grid with the corresponding number is stored in different positions in the array. i =[T i,1 ,T i,2 ,…,T i,j ],j=1:900,

[0108]

[0109] The temperature of each grid T i,j It is also related to the welding process parameters of the corresponding stage, such as:

[0110] Pressing stage: the time from the start of the current welding process to the present time t r1 (k=1),n,q,v d ,h,

[0111] T i,j=f(t r1 ,n,q,v d ,h)

[0112] Dwell stage model: t r2 (k=2),n,q,v d ,h,t s ,

[0113] T i,j =f(t r2 ,n,q,v d ,h,t s )

[0114] Stable welding stage model: t r3 (k=3),n,q,v d ,h,t s 、v w ,

[0115] T i,j =f(t r3 ,n,q,v d ,h,t s ,v w )

[0116] When training the temperature field prediction model:

[0117] The input variable of the model in the pressing stage is the thickness of the welding plate t h , the number of mesh layers i, the time from the start of the current welding stage to the present t rk (k=1),n,q,v d , h, the input variable of the dwell stage model is t h ,i,t rk (k=2),n,q,v d ,h,t s , the input variable of the stable welding stage model is t h ,i,t rk (k=3),n,q,v d ,h,t s 、v w The output variable is the temperature array T of the i-th grid on the rectangular area of ​​the workpiece with different thicknesses. i .

[0118] When the temperature field of the rectangular area is finally constructed, i is taken as 1-m, and other parameters are combined to input the temperature field prediction model, and predictions are made m times to finally obtain the temperature field T.

[0119] like Figure 5As shown in the figure, the mirror stir friction welding robot controller first judges the different welding stages (pressing stage, dwelling stage, welding stage) through the value of k, and then selects the corresponding parameters to input into the forging force prediction model and temperature field prediction model of the corresponding stage, and obtains t rk The forging force and temperature field data at the moment. For the stable welding stage, the response time model must also be considered. When the pressing amount h or the stirring head spindle speed n remains unchanged, the response time Δt is 0, and the forging force and temperature field prediction is performed as described above. When the pressing amount h or the stirring head spindle speed n changes, the response time prediction model is used to predict the time Δt required to reach the steady state after the parameter changes. Then, according to the time t before the parameter change b The time t when the parameter reaches steady state after the change can be obtained a =t b +Δt, predict the steady-state upset force and temperature field data according to the above prediction method. The upset force before reaching the steady state and the temperature of the j-th grid can be approximately expressed as F = F1 + (F2-F1) / Δt, T j =T j1 +(T j2 -T j1 ) / Δt, (j=1, 2, 3...m), where F1 / F2 corresponds to the upsetting force before / after the steady state, T j1 / T j2 It corresponds to the temperature of the jth grid before / after steady state, and m is the number of temperature field grids.

[0120] Step 7: Use Unity to combine the initial speed prediction model and the digital twin prediction model to build a visualization interface. The visualization interface consists of four parts: welding animation area, welding temperature area, process parameter area, and thermal curve area. Figure 3 The stirring head is built in SolidWorks and imported into Unity. The workpiece is created by Unity. The position of the workpiece is fixed. The position and rotation of the stirring head are determined by the robot end point coordinates and the robot spindle speed respectively. The corresponding data is converted to the corresponding object through C# code. The specific effect is displayed on the screen through the Unity main camera. The welding temperature zone is used to display the temperature field of the left and right welding zones and the temperature field distribution of the middle section parallel to the welding direction (such as Figure 3 As shown in the lower left, first construct a cube with a length and width of W and a thickness of the workpiece t based on 1mm in Unity. hThe welding zone temperature field visualization model and the cross-section temperature field visualization model with a length of W and a width of 1 are rendered using the temperature data of the grids with different serial numbers transmitted by the thermal prediction module. The temperature field of the welding zone and the cross-section temperature field visualization model are rendered according to the principle that the color changes from blue to red as the temperature increases from low to high. The temperature field distribution of the left / right welding zone temperature field visualization model and the cross-section temperature field visualization model are recorded by three cameras respectively, and the three camera images are placed in the main camera image through the unity depth function. Figure 3 The process parameter area (such as Figure 3 The function shown in the upper right corner is as follows: 1. Predict initial speed: Click the Prediction button to go to the initial speed prediction interface, enter F s , t h ,q,h,v w , sent to the initial speed prediction model, and the predicted speed is displayed on the interface; 2. Real-time display of the process parameters used for the current processing: Click the display button to go to the parameter real-time process parameter display interface. The process parameter data passed by the robot to the digital twin model is displayed in real time through the text function of the canvas in unity. The thermal curve area is used to display the real-time upsetting force curve obtained by the digital twin model during the welding process, as well as the maximum temperature curve of the welding area on both sides and the temperature curve of the tool tip point. By constructing a LineGraph component in the canvas, a coordinate system is established with time as the horizontal coordinate and temperature / force as the vertical coordinate, and the two discrete points obtained in chronological order are connected to achieve a dynamic curve effect, and by clicking the unity button (such as Figure 3 The three buttons ("Force," "Maximum Temperature," and "Temperature Difference") in the lower right corner switch between force and heat curves. The digital twin's real-time data is visualized through a visual interface, allowing users to monitor physical quantities such as weld zone temperature and stir head forging force during the welding process. The initial speed prediction model, digital twin prediction model, and visual interface together constitute the mirror friction stir welding digital twin.

[0121] Step 8: Based on the upsetting force error and the rate of change of the upsetting force error in the simulation data, two dual-input and single-output fuzzy controllers are established. The inputs of the two fuzzy controllers are the upsetting force error and the rate of change of the upsetting force error, and the outputs are the downward pressure increment and the speed increment respectively.

[0122] Step 9: Before welding begins, generate the active side robot offline based on the process parameters except the rotation speed (such as Figure 1 The theoretical welding trajectory is shown in Figure 2, and the theoretical welding trajectory of the slave robot is generated by the mirror motion constraint. Figure 5 As shown in the preparation stage, the initial value of the forging force F s , workpiece thickness t h , inclination angle q, pressure h, welding speed vw Enter the initial speed prediction interface in the visualization interface and determine n using the initial speed prediction model. Complete the process parameters.

[0123] Step 10: Figure 5 The real-time mapping part of the digital twin. The theoretical welding trajectory is input to the master / slave robot to drive the robot movement; the speed is input to the master / slave robot spindle to drive the spindle rotation. The welding process begins. During the entire welding operation, the robot feedback parameters and transmits the following parameters to the digital twin model via the TCP / IP protocol: welding stage number k, workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , actual downward pressure h s , residence time t s , welding speed v w , the time t that the kth welding stage has been running rk , pressure change Δh, speed change Δn, x, y, z, a, b and other parameters.

[0124] Where: t h Obtained by manual measurement, t rk Obtained by the internal timer, v d 、v w The feed rate is determined by the G code, n is obtained by the spindle speed feedback, t s is the set value.

[0125] k: represents the numbers of the three different welding stages in the friction stir welding process. The values ​​of k are 1, 2, and 3, where 1 represents the pressing stage, 2 represents the dwell stage, and 3 represents the stable welding stage.

[0126] like Figure 2 As shown, the motor drives the gear installed in the base 6, which in turn drives the laser distance sensor 9 installed on the bracket 7 on the rotating platform 8 to rotate. The data measured by the laser distance sensor 9 fits a cone whose top is cut by a plane. Its bottom diameter and generatrix inclination are known. The object is cut through by a plane passing through the axis, and the sum of the heights of the two vertices of the two-dimensional figure of the section is halved to obtain the axis length c. When the axis length c is equal to the distance H from the laser emission point to the top surface of the stirring head along the knife axis, k is set to 1, and the downward pressure stage begins. Figure 5 As shown on the left, when the axis length c is the distance H from the laser emission point to the top surface of the stirring head along the knife axis, k is 1; the downward pressure stage begins.

[0127] When the axis length is Hv, the downward pressure reaches h, where: v = l + h / cos (q), l is the length of the stirring needle; k is 2, and the dwell stage begins;

[0128] After k is 2, the timing starts and the time reaches t s When k is 3, the welding stage begins;

[0129] q: Figure 2 As shown, the laser ranging sensor 9 rotates one circle to fit a Figure 2 The smaller angle between the plane normal and the straight line of the sensor coordinate system z-axis in the o1 / o2 sensor coordinate system is the process inclination angle q;

[0130] h s : The actual pressing amount is obtained by the laser distance sensor 9: h s =(Hcl)*cos(q),h s ≈h.

[0131] h: Reference pressure, the actual pressure h is obtained by the circular scanning laser ranging sensor 9 s :h s =(Hcl)*cos(q), compare with the reference pressure h, adjust the robot motion trajectory so that h s Keep it as h.

[0132] The values ​​of Δh and Δn are output by the fuzzy controller.

[0133] The digital twin prediction model predicts the real-time upsetting force and temperature field based on the parameters, and transmits the data to the visualization interface for real-time display, so that the digital twin can map the welding process in real time.

[0134] Step 11: In the stable welding stage, the mirror stir friction welding temperature-upsetting force control is performed based on the fuzzy controller and the digital twin prediction model, such as Figure 5In the temperature / forging force control section, the first step is to determine whether the time since the last adjustment of the process parameters exceeds the response time Δt (Δt = 0 at the beginning or / no adjustment). If so, the parameters can be adjusted and the next step can be carried out; if not, the parameters are not adjusted and there are no subsequent steps. Then, based on the difference between the forging force and maximum temperature fed back by the digital twin and the set forging force and 80% melting temperature of the welded workpiece, the choice of using a downforce fuzzy controller or a speed fuzzy controller is made: when the forging force and maximum temperature are one small or one large (one large and one small), the forging force error and the rate of change of the forging force error are sent to the speed fuzzy controller to calculate the speed increment, and the adjusted speed is sent to the controller to achieve speed regulation; when the forging force and maximum temperature are both small / large, the forging force error and the rate of change of the forging force error are sent to the downforce fuzzy controller to calculate the downforce. After the downforce fuzzy controller calculates the downforce adjustment value, it determines whether the adjusted downforce is between 0 and 0.5. If so, the adjusted downforce is sent to the controller to achieve downforce adjustment; if not, the speed fuzzy controller is used to adjust the speed. By adjusting the downforce and speed, the temperature-forging force control of mirror stir friction welding based on digital twin is finally achieved.

[0135] The embodiments described above are only used to illustrate the technical ideas and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of the patent of the present invention cannot be limited by these embodiments alone. That is, any equivalent changes or modifications made to the spirit disclosed by the present invention still fall within the scope of the patent of the present invention.

Claims

1. A method for controlling temperature and upsetting force of mirror stir friction welding based on digital twin, characterized in that: The steps include: A mirror image friction stir welding thermal coupling simulation model is constructed based on the modified single-sided friction stir welding thermal coupling simulation model; the following neural network models are constructed, whose input includes welding parameters and whose outputs are the initial speed of the stirring head, the upsetting force, the temperature field of the workpiece, and the response time from the issuance of the instruction to the upsetting force and the temperature field of the workpiece reaching the next steady-state value: initial speed prediction model, upsetting force prediction model, temperature field prediction model, response time prediction model; the following fuzzy controllers are constructed, whose input includes the upsetting force error and the upsetting force error change rate and whose outputs are the downsetting amount change and the speed change, respectively: downsetting amount change fuzzy controller and speed change fuzzy controller; Thermal experiments and variable parameter experiments were conducted using the mirror image friction stir welding thermal-mechanical coupling simulation model. The experimental data were used to compile corresponding training samples to train the initial speed prediction model, the upsetting force prediction model, the temperature field prediction model, and the response time prediction model. The above-mentioned upset force prediction model, temperature field prediction model, and response time prediction model are combined to construct a digital twin prediction model; The welding trajectory of the driven stirring head is obtained by using the mirror motion constraint and the welding trajectory of the active stirring head. The welding parameters except the rotation speed and upsetting force in the stable welding stage and the set upsetting force are input into the initial speed prediction model to obtain the initial welding speed of the stirring head. The mirror-image friction stir welding robot performs welding operations based on the welding trajectory and initial rotational speed of the active and passive stir heads. It also feeds the current welding parameter values ​​into the digital twin prediction model, which then predicts and outputs the following data: upset force, workpiece temperature field, and the response time from issuing a command until the upset force and workpiece temperature field reach the next steady-state value. Based on the above data output by the digital twin prediction model, the mirror friction stir welding robot chooses to use a fuzzy controller for the variation of the downward pressure or the variation of the speed to adjust and control the downward pressure of the stirring head or the speed of the stirring head accordingly, thereby controlling the upsetting force applied by the stirring head and the temperature field of the workpiece.

2. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twinning according to claim 1, characterized in that: The upset force prediction model and the temperature field prediction model are divided into three welding stage sub-models according to the following three stages of the welding process: pressing stage, dwelling stage, and stable welding stage.

3. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twin according to claim 2, characterized in that: The mirror friction stir welding robot is equipped with: a rotating platform, a bracket fixed to the rotating platform, and a laser ranging sensor installed on the bracket; the mirror friction stir welding robot feeds back the following welding parameter values ​​corresponding to different welding stages to the digital twin prediction model: welding stage number k, workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , actual downward pressure h s , residence time t s , welding speed v w , the time t that the kth welding stage has been running rk ; where: t h Measured manually or with an instrument, t rk Obtained by the internal timer, v d 、v w The feed rate is determined by the G code, n is obtained by the spindle speed feedback, t s is the set value; k: represents the numbers of the three different welding stages in the friction stir welding process. The values ​​of k are 1, 2, and 3, where 1 represents the pressing stage, 2 represents the dwell stage, and 3 represents the stable welding stage. The motor drives the turntable, which rotates the laser distance sensor on the bracket. The data measured by the laser distance sensor is fitted into a cone whose top is cut by a plane. Its bottom diameter and generatrix inclination angle are known. The object is cut through by a plane passing through the axis. 50% of the sum of the heights of the two vertices of the two-dimensional figure of the cut surface is the axis length c. When the axis length c equals the distance H from the laser emission point to the top surface of the mixing head along the blade axis, k is set to 1, and the downward pressure stage begins. When the axis length is Hv, the downward pressure reaches h, where: v = l + h / cos (q), l is the length of the stirring needle; k is 2, and the dwell stage begins; After k is 2, the timing starts and the time reaches t s When k is 3, the welding stage begins; q: Inclination angle. The laser ranging sensor rotates one circle to fit a plane in the sensor coordinate system. This is the smallest angle between the plane normal and the straight line on the z-axis of the sensor coordinate system. h s : The actual downward pressure is obtained through the laser ranging sensor, h s =(Hcl)*cos(q).

4. The method for controlling temperature and upsetting force of mirror-image friction stir welding based on digital twin according to claim 2, characterized in that: The training input data of each welding stage sub-model of the upset force prediction model includes: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w .

5. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twin according to claim 2, characterized in that: The training input data of each welding stage sub-model of the temperature field prediction model includes: the time t that the kth welding stage has been running rk , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w , and the grid layer number i along the thickness direction in the mirror image friction stir welding thermal coupling simulation model; the training output data is the temperature data of the i-th grid along the thickness direction in the square area with a length and width of W centered on the stirring head in the corresponding mirror image friction stir welding thermal coupling simulation model.

6. The method for controlling temperature and upsetting force of mirror-image friction stir welding based on digital twin according to claim 1, characterized in that: The training input data of the response time prediction model include: workpiece thickness t h , stirring head spindle speed n, inclination angle q, downward pressure h, welding speed v w , change in downward pressure Δh, change in speed Δn.

7. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twin according to claim 1, characterized in that: The training input data of the initial speed prediction model include: the average value of the upsetting force data F in the stable welding stage a , workpiece thickness t h , stirring head spindle speed n, inclination angle q, pressing speed v d , pressure h, residence time t s , welding speed v w .

8. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twinning according to claim 1, characterized in that: Using the Unity development platform, a visualization interface was built by combining the initial speed prediction model and the digital twin prediction model. The visualization interface is used to determine the initial speed and display the temperature field of the workpiece welding zone and the upsetting force exerted on the stirring head in real time during the welding process.

9. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twinning according to claim 1, characterized in that: A thermomechanical coupling simulation model of single-sided friction stir welding was established using the fluid-solid coupling method based on ABAQUS. The heat generation and boundary conditions of the simulation model were adjusted according to experimental data and existing literature data to correct the thermal data of the simulation model. Based on the corrected single-sided friction stir welding simulation model, a mirror image friction stir welding thermomechanical coupling simulation model was constructed.

10. The method for controlling temperature and upset force of mirror-image friction stir welding based on digital twin according to claim 1, characterized in that: In the stable welding stage, the mirror friction stir welding robot selects to use the fuzzy controller of the pressure variation or the fuzzy controller of the speed variation according to the data output by the digital twin prediction model, which includes the following steps: Step S1, set the initial response time Δt = 0; determine whether the time from the last adjustment of the welding parameters exceeds the response time Δt, if so, execute S3; if not, do not adjust the parameters; Step S2: When the predicted upsetting force output by the digital twin prediction model is different from the set upsetting force, a downward pressure fuzzy controller or a speed fuzzy controller is selected based on a comparison result between the predicted upsetting force output by the digital twin prediction model and the set upsetting force, and a comparison result between the maximum temperature in the workpiece temperature field output by the digital twin prediction model and 80% of the workpiece melting temperature. Step S3: When the ratio of the predicted upset force to the set upset force is different from the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, execute S4; when the ratio of the predicted upset force to the set upset force is the same as the ratio of the highest temperature in the workpiece temperature field to 80% of the workpiece melting temperature, execute S5; Step S4, sending the upset force error and the upset force error change rate to the speed fuzzy controller to calculate the speed increment, and sending the adjusted speed to the controller to achieve speed regulation; In step S5, the upset force error and the upset force error change rate are sent to the down-pressure fuzzy controller to calculate the adjusted down-pressure amount, and determine whether the adjusted down-pressure amount is between 0-0.

5. If so, the adjusted down-pressure amount is sent to the mirror stir friction welding robot controller to implement the down-pressure amount adjustment; if not, return to step S4.

Images