Micro welding seam tracking method and system based on electromagnetic induction

Abstract

The invention relates to the technical field of automatic welding, and discloses a micro welding seam tracking method and system based on electromagnetic induction. According to the method, a high-frequency alternating magnetic field is applied to a welding seam area through an exciting coil in the micro electromagnetic sensing core; an induction coil in the electromagnetic sensing core is used for detecting magnetic field disturbance caused by geometric shape changes of the weld groove, and induction signals are obtained; processing the induction signal through a signal processing circuit integrated in the probe, namely converting the impedance change of the coil into a voltage signal, demodulating and filtering the voltage signal, and extracting a direct-current voltage signal corresponding to the deviation of the probe relative to the central position of the welding seam; the direct-current voltage signals serve as position feedback information to be output to a welding robot control system; and the movement track of the welding robot is adjusted in real time according to the position feedback information, and welding seam tracking and path deviation correction are achieved.

Description

Technical Field

[0001] This application relates to the field of automated welding technology, and in particular to a method and system for tracking micro weld seams based on electromagnetic induction. Background Technology

[0002] In the field of automated welding, real-time and accurate weld seam tracking is a core challenge in ensuring welding quality. Among the current mainstream technologies, laser vision sensors rely on optical imaging principles. However, in harsh welding environments with strong arc light, high levels of smoke and dust, and metal spatter, the optical signal is severely disturbed, leading to a sharp drop in the signal-to-noise ratio or even failure. Moreover, the high cost of these devices limits their widespread adoption in complex working conditions. While mechanical contact probes are less expensive, they rely on direct contact between a physical probe and the weld seam for guidance, inevitably resulting in mechanical wear and frequent component replacements. Furthermore, they are constrained by mechanical inertia, leading to lag in response during high-speed welding or complex path tracking, making it difficult to meet the requirements for high-precision real-time correction.

[0003] Therefore, a method is urgently needed to solve at least one of the above problems. Summary of the Invention

[0004] This application provides a method and system for tracking micro weld seams based on electromagnetic induction, aiming to solve the current industry's urgent need for a new weld seam tracking technology that can avoid optical interference and mechanical wear defects while taking into account the advantages of fast response and low cost.

[0005] In a first aspect, embodiments of this application provide a method for tracking micro weld seams based on electromagnetic induction, comprising: A high-frequency alternating magnetic field is applied to the weld area by an excitation coil in the micro electromagnetic sensing core; the magnetic field disturbance caused by the change in the geometric shape of the weld groove is detected by an induction coil in the electromagnetic sensing core, and the induction signal is acquired. The induced signal is processed by a signal processing circuit integrated inside the probe, including converting the coil impedance change into a voltage signal, demodulating and filtering the voltage signal, and extracting the DC voltage signal corresponding to the deviation of the probe position relative to the weld center. The DC voltage signal is output as position feedback information to the welding robot control system; the motion trajectory of the welding robot is adjusted in real time according to the position feedback information to achieve weld tracking and path correction.

[0006] In some embodiments, applying a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core includes: using a differential coil structure to pass a high-frequency alternating current with a frequency range of 100kHz to 2MHz to the excitation coil, wherein the differential coil structure consists of an excitation coil and two symmetrically arranged induction coils to generate a stable high-frequency alternating magnetic field.

[0007] In some embodiments, the method of using the induction coils in the electromagnetic sensing core to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal includes: when the probe is located at the center of the weld, the voltages induced by the two induction coils are equal in magnitude and have a zero difference; when the probe is offset, the voltage difference between the two induction coils is detected, and the voltage difference is proportional to the offset, while suppressing common-mode interference caused by minor changes in workpiece material and temperature drift.

[0008] In some embodiments, converting the coil impedance change into a voltage signal includes: using a precision bridge circuit or resonant circuit based on an operational amplifier to detect the Q value of the coil, and converting the minute changes in the equivalent inductance and equivalent resistance of the induction coil caused by magnetic field disturbance into a corresponding voltage signal.

[0009] In some embodiments, the demodulation and filtering of the voltage signal to extract the DC voltage signal corresponding to the deviation of the probe position relative to the weld center includes: demodulating the voltage signal using a phase-sensitive detector, extracting the real signal related to distance and position, suppressing quadrature components and noise, and then amplifying and low-pass filtering the signal to obtain a high-precision DC voltage signal.

[0010] In some embodiments, outputting the DC voltage signal as position feedback information to the welding robot control system includes: directly connecting the DC voltage signal to the I / O module of the robot control system through an analog interface or an industrial bus-type digital interface, wherein the analog interface outputs an analog voltage signal and the industrial bus-type digital interface outputs a digital signal.

[0011] In some embodiments, the step of adjusting the motion trajectory of the welding robot in real time according to the position feedback information to achieve weld seam tracking and path correction includes: after receiving a DC voltage signal, calculating the position deviation of the probe relative to the weld seam center through a preset control algorithm, adjusting the motion parameters of each axis of the welding robot in real time based on the deviation, and correcting the motion trajectory of the welding robot.

[0012] In some embodiments, the method further includes: training a machine learning model using historical welding data to establish a mapping relationship model between weld position deviation and welding process parameters and robot motion trajectory; during the welding process, predicting the changing trend of weld position through the mapping relationship model and adjusting the motion trajectory of the welding robot in advance.

[0013] In some embodiments, the method further includes: employing an adaptive filtering algorithm to monitor changing parameters of the welding environment in real time, including welding current and welding voltage, and dynamically adjusting demodulation filtering parameters in the signal processing circuit according to the monitored changing parameters, so as to improve the ability to suppress interference at the welding site.

[0014] Secondly, this application provides a micro weld seam tracking system based on electromagnetic induction, including a probe for micro weld seam tracking, a welding robot, and a controller; the probe is fixed to the front end or side of the welding torch corresponding to the welding robot by a mounting bracket, and is integrated with the welding torch; the controller is electrically connected to the probe and is used to implement the method provided in any embodiment of this application.

[0015] This application, based on the principle of electromagnetic induction, completely avoids the effects of optical pollution from welding arc light, fumes, and spatter, enabling stable operation in extremely harsh environments. The non-contact measurement mode completely eliminates mechanical wear, significantly extending probe lifespan and reducing maintenance costs. With a microsecond-level electromagnetic induction response and no mechanical inertial delay, it precisely adapts to the high-speed dynamic correction requirements of robots. Compared to laser vision solutions, it features a simpler hardware structure and lower manufacturing costs, facilitating large-scale industrial applications. Its miniaturized design allows for easy integration with welding torches, enabling precise tracking of complex geometric weld seams even in confined spaces.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic flowchart illustrating the steps of a micro weld seam tracking method based on electromagnetic induction, provided in an embodiment of this application. Figure 2 This is a schematic diagram of a probe for tracking a miniature weld seam based on electromagnetic induction, provided in an embodiment of this application; Figure 3 This is a schematic diagram of a miniature weld seam tracking system based on electromagnetic induction provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a miniature weld seam tracking device based on electromagnetic induction provided in one embodiment of this application; Figure 5 This is a schematic block diagram of the controller provided in one embodiment of this application.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0022] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0023] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0024] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0025] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0026] In the field of automated welding, real-time and accurate weld seam tracking is a core challenge in ensuring welding quality. Among the current mainstream technologies, laser vision sensors rely on optical imaging principles. However, in harsh welding environments with strong arc light, high levels of smoke and dust, and metal spatter, the optical signal is severely disturbed, leading to a sharp drop in the signal-to-noise ratio or even failure. Moreover, the high cost of these devices limits their widespread adoption in complex working conditions. While mechanical contact probes are less expensive, they rely on direct contact between a physical probe and the weld seam for guidance, inevitably resulting in mechanical wear and frequent component replacements. Furthermore, they are constrained by mechanical inertia, leading to lag in response during high-speed welding or complex path tracking, making it difficult to meet the requirements for high-precision real-time correction.

[0027] Therefore, a method is urgently needed to solve at least one of the above problems.

[0028] Please refer to Figure 1 This application provides a method for tracking micro weld seams based on electromagnetic induction, combined with, for example... Figure 2-3 The diagram illustrates a probe and system implementation for electromagnetic induction-based miniature weld seam tracking. The provided system includes a miniature weld seam tracking probe, a welding robot, and a controller. The probe is fixed to the front or side of the welding torch corresponding to the welding robot via a mounting bracket, achieving integrated functionality with the welding torch. The controller is electrically connected to the probe and is used to implement the method provided in any embodiment of this application.

[0029] Specifically, such as Figure 1 As shown, the provided electromagnetic induction-based micro weld seam tracking method includes steps S101 to S103. Details are as follows: Step S101. Apply a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core; use the induction coil in the electromagnetic sensing core to detect the magnetic field disturbance caused by the change in the geometric shape of the weld bevel, and acquire the induction signal.

[0030] Specifically, this step is the perception stage of the method, the core of which is to use the principle of electromagnetic induction to non-contactly detect the geometric features of the weld.

[0031] The active excitation magnetic field converts high-frequency alternating current into a high-frequency alternating magnetic field through an excitation coil, serving as the "source" for detection.

[0032] Passive sensing disturbances induce eddy currents on the surface of a conductive weld workpiece when the alternating magnetic field is applied. The geometry of the weld bevel (such as edges, depth, and center position) directly alters the distribution intensity and path of the eddy currents.

[0033] The detection of magnetic field changes involves the generation of a secondary magnetic field by the changing eddy currents. This secondary magnetic field superimposes on the original excitation magnetic field, thereby altering the distribution of the original magnetic field. The induction coil detects this composite magnetic field resulting from the disturbance to obtain the original induced signal containing information about the weld location.

[0034] The excitation signal is generated and applied by a high-frequency oscillator integrated inside the probe, producing a frequency-stable sinusoidal signal. The preferred frequency range is 100 kHz to 2 MHz. Choosing a higher frequency helps improve sensitivity and response speed. This high-frequency alternating current is then applied to the excitation coil. The excitation coil is typically wound with Litz wire to reduce skin effect losses at high frequencies and ensure magnetic field strength.

[0035] To achieve high precision and interference resistance, a differential coil structure is preferred. This structure consists of an excitation coil and two spatially symmetrical induction coils. When the probe is precisely positioned on the weld centerline, the two induction coils are equidistant from the geometrical symmetry planes on both sides of the bevel, resulting in identical eddy current effects. Consequently, the voltage signals output by the two coils are equal in magnitude, with zero difference (ideal state). When the probe deviates from the center, one coil moves closer to the edge of the bevel, inducing a stronger magnetic field disturbance, while the other weakens, thus generating a differential voltage signal related to the magnitude and direction of the offset. This structure effectively suppresses common-mode interference caused by workpiece material inhomogeneity, temperature variations, etc.

[0036] Signal acquisition involves converting detected magnetic field changes into a weak AC voltage signal using an induction coil. The changes in the amplitude and phase of this signal relative to the excitation signal directly encode the relative position information between the probe and the weld.

[0037] Step S102. The induced signal is processed by the signal processing circuit integrated inside the probe, including converting the coil impedance change into a voltage signal, demodulating and filtering the voltage signal, and extracting the DC voltage signal corresponding to the deviation of the probe position relative to the weld center.

[0038] Specifically, this step is the signal processing and conversion stage, the purpose of which is to convert the weak, raw induction signal obtained in step S101 into a stable, accurate displacement signal that is easily recognized by the control system. Its technical essence is a sophisticated electromagnetic measurement signal chain.

[0039] Impedance-to-voltage conversion converts minute changes in coil impedance (including equivalent inductance L and resistance R) caused by magnetic field disturbances into voltage changes. Demodulation recovers the low-frequency (or DC) effective signal containing position information from the high-frequency carrier signal modulated by weld information. Filtering and amplification remove unavoidable electromagnetic noise in the welding environment and high-frequency residual components in the demodulated signal, and amplify the weak effective signal to a standard voltage level.

[0040] Impedance-to-voltage conversion circuits can employ precision bridge circuits (such as Wheatstone bridges), with the sensing coil serving as one arm of the bridge. When the coil impedance changes, it disrupts the bridge's balance, outputting a voltage signal proportional to the impedance change. Another preferred approach is to use a resonant circuit with Q-factor detection. The sensing coil and capacitor form an LC parallel resonant circuit. When the coil impedance changes, the quality factor (Q) and resonant frequency of the resonant circuit change. The impedance change is indirectly measured by detecting the amplitude or phase change of the resonant voltage. This method has higher sensitivity at specific frequencies.

[0041] The demodulation circuit employs a phase-sensitive detector or a synchronous demodulator. It requires a reference signal that is in phase and frequency with the original excitation signal. The phase-sensitive detector simultaneously demodulates the amplitude and phase information of the input signal. It not only outputs the signal amplitude related to the probe-workpiece distance but also determines the direction of probe offset (e.g., left or right) through phase information. This is crucial for achieving bidirectional correction and effectively suppresses noise components orthogonal to the reference signal.

[0042] The filtering and amplification circuit uses a low-pass filter after demodulation to remove high-frequency harmonics and residual carrier waves generated by demodulation, while retaining low-frequency or DC signals that reflect position changes.

[0043] Subsequently, the DC signal is amplified to a standard range (e.g., ±10V or 0-5V) using an operational amplifier circuit, forming the final high-precision DC voltage signal. This signal has a precise functional relationship with the probe's offset.

[0044] Step S103. Output the DC voltage signal as position feedback information to the welding robot control system; adjust the motion trajectory of the welding robot in real time according to the position feedback information to achieve weld tracking and path correction.

[0045] Specifically, this step is the control and execution phase of the method, completing the closed loop from perceived information to robot action. Its technical essence is the construction of a real-time feedback control system.

[0046] Information transmission reliably transmits the standard analog signals processed by the probe to the robot's "brain"—the control system.

[0047] Closed-loop control compares the received actual position signal with the preset ideal weld path (expected value) through the robot controller, and calculates the position deviation.

[0048] The path correction system adjusts the welding torch's trajectory in real time based on the magnitude and direction of the deviation, ensuring that the welding torch is always precisely aligned with the center of the weld.

[0049] The probe outputs a DC voltage signal through its output interface (such as a robust aviation connector). The interface type can be an analog interface (such as a ±10V voltage signal) or a digital interface with an integrated fieldbus protocol (such as PROFIBUS, EtherCAT, etc.). The latter has stronger anti-interference capabilities and is suitable for complex industrial environments.

[0050] The probe is integrated with the welding torch via a compact mounting bracket with fine-tuning capabilities. Before welding begins, initial position calibration is required. By fine-tuning the bracket, the DC voltage signal output by the probe is made zero (or a preset reference value) when the welding torch is aligned with the weld seam.

[0051] The robot controller receives the probe's voltage signal in real time through its I / O module. The controller has a pre-installed control algorithm (usually a PID algorithm or a variant thereof). This algorithm compares the received voltage signal (representing the actual offset) with zero (representing the desired weld center) in real time to calculate the correction command.

[0052] The controller sends this command to the robot's servo driver, which drives the motors of the corresponding joints to perform minute, rapid movements, thereby moving the welding torch in the direction of eliminating deviation and achieving real-time, dynamic path tracking and correction. Because the electromagnetic induction response speed is in the microsecond range, the entire perception-decision-execution closed loop can meet the real-time requirements of high-speed robot welding.

[0053] In some embodiments, applying a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core includes: using a differential coil structure to pass a high-frequency alternating current with a frequency range of 100kHz to 2MHz to the excitation coil, wherein the differential coil structure consists of an excitation coil and two symmetrically arranged induction coils to generate a stable high-frequency alternating magnetic field.

[0054] This embodiment details a method for generating a detection magnetic field, the core of which lies in defining the preferred structure of the sensing core and the key parameters of the excitation signal. Employing a differential coil structure is crucial for improving measurement accuracy, while limiting the excitation frequency to a high-frequency range of 100 kHz to 2 MHz represents an engineering trade-off between the depth of the eddy current effect and signal sensitivity.

[0055] The coil winding and arrangement involves first precisely winding an excitation coil on the frame at the probe's front end. Then, behind or to both sides of this excitation coil, two induction coils are precisely wound symmetrically about the probe's central axis. The coil frame is preferably made of high-temperature resistant ceramic material to ensure the stability of its physical and electromagnetic properties under the high-temperature welding environment.

[0056] Signal generation is achieved through a high-frequency oscillator circuit within the probe, configured to produce a stable, pure sine wave signal. The frequency of this signal is precisely controlled within the aforementioned preferred range through circuit design (e.g., using a high-precision crystal oscillator or phase-locked loop technology), for example, a typical implementation frequency of 400 kHz.

[0057] The current drive generates an AC signal that passes through a power drive amplifier circuit, enabling it to provide sufficient current to the excitation coil, thereby generating a sufficiently strong and stable high-frequency alternating magnetic field in front of the probe.

[0058] In some embodiments, the method of using the induction coils in the electromagnetic sensing core to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal includes: when the probe is located at the center of the weld, the voltages induced by the two induction coils are equal in magnitude and have a zero difference; when the probe is offset, the voltage difference between the two induction coils is detected, and the voltage difference is proportional to the offset, while suppressing common-mode interference caused by minor changes in workpiece material and temperature drift.

[0059] This embodiment reveals the core working mechanism and anti-interference principle of the differential structure. Its core technology is to use symmetry and differential comparison to extract small useful signals (displacement signals) while suppressing common environmental interference (common-mode interference).

[0060] Signal detection generates induced voltage signals V1 and V2 through two induction coils. Subsequent signal processing circuitry (typically a differential amplifier) ​​first reads these two voltage values. Differential processing calculates the difference between the two signals, ΔV = V1 - V2. Zero-point calibration is performed during initial calibration by fine-tuning the probe's mounting position to precisely align it with the weld center. At this point, due to symmetry, V1 ≈ V2, therefore ΔV ≈ 0V. This point is the output "zero point".

[0061] Offset detection works by detecting that if the probe shifts to the left during welding, the left coil is closer to the metal, resulting in a stronger eddy current effect and greater magnetic field disturbance. This causes a greater change in V1 than V2, and ΔV outputs a positive voltage (conversely, a negative voltage). The amplitude of this voltage represents the magnitude of the offset, and its polarity (positive or negative) represents the direction of the offset. Slight changes in the workpiece material or an increase in ambient temperature will almost equally affect the two symmetrical induction coils, causing V1 and V2 to increase or decrease simultaneously (common-mode change). However, when calculating the difference ΔV, this common change is canceled out, greatly improving the system's anti-interference capability and stability.

[0062] In some embodiments, converting the coil impedance change into a voltage signal includes: using a precision bridge circuit or resonant circuit based on an operational amplifier to detect the Q value of the coil, and converting the minute changes in the equivalent inductance and equivalent resistance of the induction coil caused by magnetic field disturbance into a corresponding voltage signal.

[0063] This embodiment clarifies two classic circuit implementation paths for converting the abstract "impedance change" into a measurable "voltage change". The bridge method is a direct measurement, while the Q-value detection method uses the resonance effect for indirect measurement. Both are mature technologies for achieving high-precision measurement of micro-variables.

[0064] The precision bridge circuit design uses two induction coils (or one of them and a reference coil) in a differential coil structure as two adjacent arms of a Wheatstone bridge, with the other two arms consisting of high-precision, low-temperature-coefficient resistors. The bridge is powered by an AC signal at the same frequency as the excitation source. Initially, the bridge is balanced through tuning, and the output voltage is zero. When the coil impedance changes due to weld disturbances, the bridge becomes unbalanced, and the output voltage signal is proportional to the impedance change.

[0065] The resonant circuit Q-factor detection scheme forms an LC parallel resonant circuit by connecting each induction coil in parallel with a suitable capacitor. This circuit has maximum impedance at its resonant frequency. Changes in coil impedance alter the quality factor (Q-factor) (i.e., frequency selectivity or sharpness) and the amplitude at the resonant point. By detecting the change in the amplitude of the voltage across the resonant circuit, the change in coil impedance can be deduced. This method exhibits extremely high sensitivity near the resonant point.

[0066] In some embodiments, the demodulation and filtering of the voltage signal to extract the DC voltage signal corresponding to the deviation of the probe position relative to the weld center includes: demodulating the voltage signal using a phase-sensitive detector, extracting the real signal related to distance and position, suppressing quadrature components and noise, and then amplifying and low-pass filtering the signal to obtain a high-precision DC voltage signal.

[0067] This embodiment specifies the key steps for extracting the DC position signal from the modulated signal. Phase-sensitive detection is the core process, which demodulates the phase-related amplitude information (i.e., retains the direction information), while low-pass filtering is used to filter out high-frequency components, resulting in a smooth DC signal.

[0068] Phase-sensitive detection involves simultaneously inputting the AC voltage signal output from the embodiment and a reference signal that is in phase and frequency with the original excitation signal into an analog multiplier or a dedicated phase-sensitive detection chip.

[0069] The multiplier's output is a signal containing a DC component, the voltage of which is proportional to the amplitude of the input signal and the cosine of its phase difference with the reference signal. Thus, not only is the magnitude of the offset converted into a voltage value, but the direction of the offset (whether the phase difference is 0° or 180°) is also indicated by the polarity of the output voltage.

[0070] The signal output from the phase-sensitive detector is complex, containing the desired DC component and many high-frequency harmonics. This signal is first amplified by an operational amplifier to adjust the level to a range suitable for subsequent processing. It then passes through an active low-pass filter with a cutoff frequency set much lower than the excitation frequency (e.g., excitation at 400 kHz, cutoff frequency at 1 kHz). This filter removes all high-frequency noise and residual carrier waves, retaining only a smooth, high-precision DC voltage signal proportional to the position deviation.

[0071] In some embodiments, outputting the DC voltage signal as position feedback information to the welding robot control system includes: directly connecting the DC voltage signal to the I / O module of the robot control system through an analog interface or an industrial bus-type digital interface, wherein the analog interface outputs an analog voltage signal and the industrial bus-type digital interface outputs a digital signal.

[0072] This embodiment relates to the interface standard between the probe and the upper-level control system, covering both analog and digital, two mainstream industrial interface forms, to adapt to robot control systems of different brands and models.

[0073] The analog interface solution outputs a DC voltage signal (e.g., ±10V or 0-5V) obtained through internal probe processing directly via a shielded cable. The other end of the cable connects to the analog input module standard on the welding robot control system. This is the most common and direct integration method; the system can directly read the voltage value and convert it into position deviation.

[0074] Industrial bus-based digital interface solutions integrate a microcontroller and a fieldbus protocol chip (such as PROFINET, EtherCAT, EtherNet / IP, etc.) within the probe. The microcontroller converts the analog voltage values ​​acquired by the ADC into digital signals, which are then packaged into standard data frames by the bus protocol chip. The digital signals are transmitted to the robot controller's bus I / O module via a network cable or dedicated bus cable. This solution offers stronger immunity to electromagnetic interference, longer transmission distances, and facilitates the integration of more complex diagnostic and parameter configuration functions.

[0075] In some embodiments, the step of adjusting the motion trajectory of the welding robot in real time according to the position feedback information to achieve weld seam tracking and path correction includes: after receiving a DC voltage signal, calculating the position deviation of the probe relative to the weld seam center through a preset control algorithm, adjusting the motion parameters of each axis of the welding robot in real time based on the deviation, and correcting the motion trajectory of the welding robot.

[0076] This embodiment describes how a robot control system utilizes position feedback information to form a closed-loop control. Its core is a real-time control loop that compares sensor feedback with preset target values ​​and generates correction commands through a control algorithm.

[0077] Deviation calculation is performed by periodically (e.g., every millisecond) reading the DC voltage signal V_offset from the probe by the robot controller. Based on a preset calibration coefficient K (unit: mm / V), the actual position deviation value Error = K * V_offset is immediately calculated.

[0078] The control algorithm operates by inputting the error value into a digital PID controller. The PID controller calculates a control output based on the current error (P), the historical cumulative error (I), and the error trend (D), which is the correction amount and direction that the welding torch needs to move.

[0079] The trajectory correction is converted into a displacement command for the robot tool center point (TCP) through this control quantity. The robot's inverse kinematics algorithm calculates in real time the additional or reduced motion required for each joint axis (such as J1-J6 axes), and adds this correction to the originally planned welding path, thereby driving the servo motor to complete real-time, smooth, and precise path correction.

[0080] In some embodiments, the method further includes: training a machine learning model using historical welding data to establish a mapping relationship model between weld position deviation and welding process parameters and robot motion trajectory; during the welding process, predicting the changing trend of weld position through the mapping relationship model and adjusting the motion trajectory of the welding robot in advance.

[0081] This embodiment is an advanced, intelligent extended implementation. By introducing machine learning, it no longer merely reacts to current deviations but attempts to predict future deviation trends, achieving proactive correction and thus further improving tracking accuracy and process quality.

[0082] Model training involves collecting a large amount of historical welding data before system deployment, including: weld seam tracking probe signals (input), welding current / voltage / speed (process parameters), and the final weld seam quality data (labels). This data is then used to train a machine learning model (such as an LSTM recurrent neural network or a gradient boosting tree) that learns the complex mapping relationship of "how the weld seam position will change under specific process parameters".

[0083] Online predictive and feedforward control works by inputting current probe signals and process parameters into a pre-trained model in real time during the welding process. The model outputs a predicted value for the position change trend in the near future. The robot controller combines traditional PID feedback control with feedforward predictive control from the machine learning model.

[0084] In this way, the system can start adjusting the robot trajectory in advance before the deviation actually occurs but has not yet been detected by the probe, achieving smoother and more accurate tracking, which is especially suitable for handling complex welds with regular changes.

[0085] In some embodiments, the method further includes: employing an adaptive filtering algorithm to monitor changing parameters of the welding environment in real time, including welding current and welding voltage, and dynamically adjusting demodulation filtering parameters in the signal processing circuit according to the monitored changing parameters, so as to improve the ability to suppress interference at the welding site.

[0086] This embodiment is another advanced enhanced implementation, focusing on addressing the strong electromagnetic interference generated by the welding process itself. Its core is adaptive filtering technology, which dynamically identifies interference characteristics and automatically adjusts filter parameters to achieve optimal filtering results.

[0087] Interference monitoring involves the system monitoring key parameters of the welding power source, such as welding current and arc voltage, in real time. These parameters are the strongest sources of interference during the welding process, and changes in them directly introduce specific noise.

[0088] Reference signal generation involves appropriately conditioning these monitored current / voltage signals and then inputting them as reference signals characterizing the noise source into an adaptive filter (such as a Wiener filter or an LMS algorithm filter).

[0089] Dynamic filtering uses an adaptive filter whose main input is the raw signal from the probe (containing true position information and noise). The filter continuously adjusts its internal parameters through an algorithm, enabling it to "learn" from the main input signal and subtract noise components associated with the reference signal.

[0090] The parameter adjustment, through the filter output, yields a purified position signal. Simultaneously, this process dynamically optimizes the demodulation and filtering parameters of the signal processing link (such as the center frequency and bandwidth of the equivalent filter), ensuring it consistently suppresses the most prevalent interference frequency. This significantly improves the system's signal-to-noise ratio and robustness in highly volatile welding environments.

[0091] In some embodiments, during the sensing phase S101, not only are electromagnetic induction signals acquired, but visual information of the weld area is also simultaneously obtained through a low-cost miniature CMOS image sensor. A lightweight convolutional neural network (CNN) is used to analyze the image in real time, automatically identifying the weld type (such as V-groove, lap fillet weld, I-butt weld, etc.), and intelligently switching or fusing the signal processing model parameters in S102 accordingly. Through multimodal information fusion, the system can maintain stable tracking even when strong arc light or spatter interference causes transient distortion of the electromagnetic signal.

[0092] A miniature camera resistant to arc light interference and a ring-shaped LED illumination source are integrated next to the electromagnetic probe. Data is acquired synchronously by the camera and the probe.

[0093] Offline training involves collecting a large number of images of different types of welds and their corresponding electromagnetic signal data to train a multi-task deep learning model. One branch of the model is a CNN for image classification, identifying weld types; the other branch is a network (such as 1D-CNN or LSTM) that processes the temporal features of electromagnetic signals to predict offsets.

[0094] Online inference and fusion include: simultaneously acquiring image I(t) and raw electromagnetic signal S(t). I(t) is input into a CNN classifier, which outputs weld type Type and confidence level. The system automatically selects a preset optimal set of signal processing parameters (such as filter cutoff frequency and PID control parameters) based on Type. Simultaneously, S(t) is input into the corresponding signal processing network branch, which outputs offset prediction ΔX_dnn. The final offset ΔX_final is obtained by weighted fusion of the traditional signal processing output ΔX_traditional and ΔX_dnn based on confidence level. When the image is briefly occluded, the weights automatically favor the traditional electromagnetic signal; when the electromagnetic signal is strongly interfered with, the weights favor the visually assisted DNN prediction. The robot corrects the offset based on ΔX_final.

[0095] In some embodiments, by introducing a reinforcement learning (RL) agent, it takes the signal characteristics and welding process parameters processed by S102 as the state, fine-tuning the PID parameters as the action, and tracking the accumulation of errors as the reward. Through online learning, it autonomously finds and dynamically adjusts the optimal control strategy, achieving unprecedented adaptive capabilities.

[0096] State space definition: State s_t includes: current offset e(t), differential of offset de(t) / dt, welding current I_weld, and welding voltage V_weld. Action space definition: Action a_t is defined as the small adjustment to the three PID parameters: ΔKp, ΔKi, ΔKd. Reward function design: Reward r_t = -(α * |e(t)| + β * |ΔAction|), which punishes large tracking errors and large parameter jitter, and encourages the system to achieve precise stability with minimal control actions.

[0097] Training and deployment include: Simulation training: First, in a welding simulation environment, the RL agent (e.g., using the PPO algorithm) interacts with the simulation model millions of times to learn and optimize strategies. Online fine-tuning: The trained policy network is deployed to the actual robot controller. In stage S103, the RL agent observes the state s_t in real time, outputs action a_t, fine-tunes the PID parameters, then executes the adjusted PID calculation and outputs control commands to the servo driver. The system continuously fine-tunes online, constantly adapting to new challenges brought about by workpiece clamping, thermal deformation, etc.

[0098] In some embodiments, the same intelligent tracking system is deployed on multiple welding robots (or workstations). Each robot collects its own tracking performance data locally (during processes S101-S103) and trains an improved model locally. Then, using a federated learning framework, only the model parameter updates of each local model (rather than the original data) are encrypted and uploaded to a cloud server for aggregation, generating a more powerful "global model." Finally, the optimized global model is distributed to all robots. In this way, the entire robot ensemble achieves knowledge and experience sharing and collaborative evolution while protecting the data privacy of each production point.

[0099] The local model comprises: each welding robot node has an initial built-in intelligent signal processing or control model (such as the DNN model in Implementation 1 or the RL policy network in Implementation 2). Local training involves incrementally training the local model after each welding task using the (signal, offset) data collected during that task. Federated aggregation involves each node periodically (e.g., daily) sending encrypted weight gradients or weight updates of its local model to a central server. The server uses a federated averaging algorithm (FedAvg) to calculate a weighted average of all received updates and update the global model.

[0100] Model distribution and updates are handled by the server, which distributes the aggregated new global model to all nodes. In the next welding task, each robot uses this refined and improved global model to perform signal analysis in S102 and decision control in S103. The entire system functions like a constantly learning bee colony, becoming smarter with use.

[0101] In some embodiments, a spiking neural network (SNN) is designed to process the raw electromagnetic signal in S101. The SNN uses discrete pulse sequences to transmit information, and its event-driven nature means that computation is only performed when the input signal changes, making it well-suited for processing such continuously sensed temporal signals. This approach aims to achieve ultra-low latency processing at the microsecond level with extremely low power consumption, providing a technological foundation for future miniaturized, battery-powered welding robots or drones.

[0102] The encoder design converts the analog voltage signal S(t) acquired by the induction coil into two pulse sequences (Spike_train_left and Spike_train_right) through a differential pulse encoder. The frequency and timing of the pulses encode the amplitude and direction of change of the signal.

[0103] SNN network processing involves inputting a pulse sequence into a pre-trained SNN. This network consists of an input layer, hidden layers, and an output layer (composed of spiking neurons, such as in the LIF model). The network learns to decode positional offsets from the pulse sequence.

[0104] Output and execution are achieved through the generation of corresponding pulses by neurons in the output layer. The decoder converts these output pulses into analog voltage signals (i.e., the output of S102) or directly into control commands (directly to S103). The entire process is event-driven; the neural network consumes almost no power when the signal remains unchanged, triggering sparse computation only when the signal changes, thus achieving ultra-low power consumption and extremely fast response. This network can be deployed on neuromorphic chips to achieve extreme optimization of energy efficiency.

[0105] This application, based on the principle of electromagnetic induction, completely avoids the effects of optical pollution from welding arc light, fumes, and spatter, enabling stable operation in extremely harsh environments. The non-contact measurement mode completely eliminates mechanical wear, significantly extending probe lifespan and reducing maintenance costs. With a microsecond-level electromagnetic induction response and no mechanical inertial delay, it precisely adapts to the high-speed dynamic correction requirements of robots. Compared to laser vision solutions, it features a simpler hardware structure and lower manufacturing costs, facilitating large-scale industrial applications. Its miniaturized design allows for easy integration with welding torches, enabling precise tracking of complex geometric weld seams even in confined spaces.

[0106] Please see Figure 4 As shown, Figure 4This is a schematic diagram of the structure of the electromagnetic induction-based miniature weld seam tracking device 200 provided in the embodiments of this application. The electromagnetic induction-based miniature weld seam tracking device 200 is used to execute the steps of the electromagnetic induction-based miniature weld seam tracking method shown in the above embodiments. The electromagnetic induction-based miniature weld seam tracking device 200 can be a single server or a server cluster, or it can be a terminal, such as a handheld terminal, a laptop computer, a wearable device, or a robot.

[0107] like Figure 4 As shown, the electromagnetic induction-based miniature weld seam tracking device 200 includes: The magnetic field application unit 201 is used to apply a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core; and to use the induction coil in the electromagnetic sensing core to detect the magnetic field disturbance caused by the change in the geometric shape of the weld groove and to acquire the induction signal. The signal processing unit 202 is used to process the induced signal through a signal processing circuit integrated inside the probe, including converting the coil impedance change into a voltage signal, demodulating and filtering the voltage signal, and extracting the DC voltage signal corresponding to the deviation of the probe from the center of the weld. The correction implementation unit 203 is used to output the DC voltage signal as position feedback information to the welding robot control system; and adjust the motion trajectory of the welding robot in real time according to the position feedback information to realize weld seam tracking and path correction.

[0108] In some embodiments, applying a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core includes: using a differential coil structure to pass a high-frequency alternating current with a frequency range of 100kHz to 2MHz to the excitation coil, wherein the differential coil structure consists of an excitation coil and two symmetrically arranged induction coils to generate a stable high-frequency alternating magnetic field.

[0109] In some embodiments, the method of using the induction coils in the electromagnetic sensing core to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal includes: when the probe is located at the center of the weld, the voltages induced by the two induction coils are equal in magnitude and have a zero difference; when the probe is offset, the voltage difference between the two induction coils is detected, and the voltage difference is proportional to the offset, while suppressing common-mode interference caused by minor changes in workpiece material and temperature drift.

[0110] In some embodiments, converting the coil impedance change into a voltage signal includes: using a precision bridge circuit or resonant circuit based on an operational amplifier to detect the Q value of the coil, and converting the minute changes in the equivalent inductance and equivalent resistance of the induction coil caused by magnetic field disturbance into a corresponding voltage signal.

[0111] In some embodiments, the demodulation and filtering of the voltage signal to extract the DC voltage signal corresponding to the deviation of the probe position relative to the weld center includes: demodulating the voltage signal using a phase-sensitive detector, extracting the real signal related to distance and position, suppressing quadrature components and noise, and then amplifying and low-pass filtering the signal to obtain a high-precision DC voltage signal.

[0112] In some embodiments, outputting the DC voltage signal as position feedback information to the welding robot control system includes: directly connecting the DC voltage signal to the I / O module of the robot control system through an analog interface or an industrial bus-type digital interface, wherein the analog interface outputs an analog voltage signal and the industrial bus-type digital interface outputs a digital signal.

[0113] In some embodiments, the step of adjusting the motion trajectory of the welding robot in real time according to the position feedback information to achieve weld seam tracking and path correction includes: after receiving a DC voltage signal, calculating the position deviation of the probe relative to the weld seam center through a preset control algorithm, adjusting the motion parameters of each axis of the welding robot in real time based on the deviation, and correcting the motion trajectory of the welding robot.

[0114] In some embodiments, the method further includes: training a machine learning model using historical welding data to establish a mapping relationship model between weld position deviation and welding process parameters and robot motion trajectory; during the welding process, predicting the changing trend of weld position through the mapping relationship model and adjusting the motion trajectory of the welding robot in advance.

[0115] In some embodiments, the method further includes: employing an adaptive filtering algorithm to monitor changing parameters of the welding environment in real time, including welding current and welding voltage, and dynamically adjusting demodulation filtering parameters in the signal processing circuit according to the monitored changing parameters, so as to improve the ability to suppress interference at the welding site.

[0116] It should be noted that those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the electromagnetic induction-based micro weld seam tracking device and its modules described above can be referred to the corresponding processes in the embodiments of the electromagnetic induction-based micro weld seam tracking method described above, and will not be repeated here.

[0117] The aforementioned electromagnetic induction-based micro weld seam tracking method can be implemented as a computer program, which can be used in, for example... Figure 4 It runs on the device shown.

[0118] Please see Figure 5 , Figure 5This is a schematic block diagram of the controller provided in an embodiment of this application. The controller includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0119] The storage medium can store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform any electromagnetic induction-based micro-weld seam tracking method.

[0120] The processor provides computing and control capabilities to support the operation of the entire controller.

[0121] The internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any electromagnetic induction-based micro weld seam tracking method.

[0122] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the terminal to which the present application is applied. The specific controller may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0123] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0124] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: A high-frequency alternating magnetic field is applied to the weld area by an excitation coil in the micro electromagnetic sensing core; the magnetic field disturbance caused by the change in the geometric shape of the weld groove is detected by an induction coil in the electromagnetic sensing core, and the induction signal is acquired. The induced signal is processed by a signal processing circuit integrated inside the probe, including converting the coil impedance change into a voltage signal, demodulating and filtering the voltage signal, and extracting the DC voltage signal corresponding to the deviation of the probe position relative to the weld center. The DC voltage signal is output as position feedback information to the welding robot control system; the motion trajectory of the welding robot is adjusted in real time according to the position feedback information to achieve weld tracking and path correction.

[0125] In some embodiments, applying a high-frequency alternating magnetic field to the weld area through the excitation coil in the micro electromagnetic sensing core includes: using a differential coil structure to pass a high-frequency alternating current with a frequency range of 100kHz to 2MHz to the excitation coil, wherein the differential coil structure consists of an excitation coil and two symmetrically arranged induction coils to generate a stable high-frequency alternating magnetic field.

[0126] In some embodiments, the method of using the induction coils in the electromagnetic sensing core to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal includes: when the probe is located at the center of the weld, the voltages induced by the two induction coils are equal in magnitude and have a zero difference; when the probe is offset, the voltage difference between the two induction coils is detected, and the voltage difference is proportional to the offset, while suppressing common-mode interference caused by minor changes in workpiece material and temperature drift.

[0127] In some embodiments, converting the coil impedance change into a voltage signal includes: using a precision bridge circuit or resonant circuit based on an operational amplifier to detect the Q value of the coil, and converting the minute changes in the equivalent inductance and equivalent resistance of the induction coil caused by magnetic field disturbance into a corresponding voltage signal.

[0128] In some embodiments, the demodulation and filtering of the voltage signal to extract the DC voltage signal corresponding to the deviation of the probe position relative to the weld center includes: demodulating the voltage signal using a phase-sensitive detector, extracting the real signal related to distance and position, suppressing quadrature components and noise, and then amplifying and low-pass filtering the signal to obtain a high-precision DC voltage signal.

[0129] In some embodiments, outputting the DC voltage signal as position feedback information to the welding robot control system includes: directly connecting the DC voltage signal to the I / O module of the robot control system through an analog interface or an industrial bus-type digital interface, wherein the analog interface outputs an analog voltage signal and the industrial bus-type digital interface outputs a digital signal.

[0130] In some embodiments, the step of adjusting the motion trajectory of the welding robot in real time according to the position feedback information to achieve weld seam tracking and path correction includes: after receiving a DC voltage signal, calculating the position deviation of the probe relative to the weld seam center through a preset control algorithm, adjusting the motion parameters of each axis of the welding robot in real time based on the deviation, and correcting the motion trajectory of the welding robot.

[0131] In some embodiments, the method further includes: training a machine learning model using historical welding data to establish a mapping relationship model between weld position deviation and welding process parameters and robot motion trajectory; during the welding process, predicting the changing trend of weld position through the mapping relationship model and adjusting the motion trajectory of the welding robot in advance.

[0132] In some embodiments, the method further includes: employing an adaptive filtering algorithm to monitor changing parameters of the welding environment in real time, including welding current and welding voltage, and dynamically adjusting demodulation filtering parameters in the signal processing circuit according to the monitored changing parameters, so as to improve the ability to suppress interference at the welding site.

[0133] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the steps of the electromagnetic induction-based micro weld seam tracking method provided in any embodiment of this application.

[0134] The computer-readable storage medium may be an internal storage unit of the controller as described in the foregoing embodiments, such as the hard disk or memory of the controller. Alternatively, the computer-readable storage medium may be an external storage device of the controller, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for tracking micro weld seams based on electromagnetic induction, characterized in that, include: A high-frequency alternating magnetic field is applied to the weld area by the excitation coil in the core of the miniature electromagnetic sensor. The induction coil in the electromagnetic sensing core is used to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal. The induced signal is processed by a signal processing circuit integrated inside the probe, including converting the coil impedance change into a voltage signal, demodulating and filtering the voltage signal, and extracting the DC voltage signal corresponding to the deviation of the probe position relative to the weld center. The DC voltage signal is output as position feedback information to the welding robot control system; the motion trajectory of the welding robot is adjusted in real time according to the position feedback information to achieve weld tracking and path correction.

2. The method according to claim 1, characterized in that, The application of a high-frequency alternating magnetic field to the weld area via the excitation coil in the micro electromagnetic sensing core includes: A differential coil structure is used to supply a high-frequency alternating current with a frequency range of 100kHz to 2MHz to the excitation coil. The differential coil structure consists of an excitation coil and two symmetrically arranged induction coils to generate a stable high-frequency alternating magnetic field.

3. The method according to claim 1, characterized in that, The method of using the induction coil in the electromagnetic sensing core to detect magnetic field disturbances caused by changes in the geometric shape of the weld groove and to acquire the induced signal includes: When the probe is located at the center of the weld, the voltages induced by the two induction coils are equal in magnitude and the difference is zero. When the probe deviates, the voltage difference between the two induction coils is detected. The voltage difference is proportional to the amount of deviation, while suppressing common-mode interference caused by minor changes in workpiece material and temperature drift.

4. The method according to claim 1, characterized in that, The process of converting coil impedance changes into voltage signals includes: The Q-value detection method based on operational amplifier precision bridge circuit or resonant circuit is used to convert the minute changes in equivalent inductance and equivalent resistance of the induction coil caused by magnetic field disturbance into corresponding voltage signals.

5. The method according to claim 1, characterized in that, The demodulation and filtering of the voltage signal to extract the DC voltage signal corresponding to the deviation of the probe position relative to the weld center includes: The voltage signal is demodulated using a phase-sensitive detector to extract the real signal related to distance and position, suppress orthogonal components and noise, and then the signal is amplified and low-pass filtered to obtain a high-precision DC voltage signal.

6. The method according to claim 1, characterized in that, The step of outputting the DC voltage signal as position feedback information to the welding robot control system includes: The DC voltage signal can be directly connected to the I / O module of the robot control system through an analog interface or an industrial bus-type digital interface. The analog interface outputs an analog voltage signal, while the industrial bus-type digital interface outputs a digital signal.

7. The method according to claim 1, characterized in that, The step of adjusting the motion trajectory of the welding robot in real time based on the position feedback information to achieve weld seam tracking and path correction includes: After receiving the DC voltage signal, the positional deviation of the probe relative to the weld center is calculated through a preset control algorithm. Based on the deviation, the motion parameters of each axis of the welding robot are adjusted in real time to correct the motion trajectory of the welding robot.

8. The method according to claim 1, characterized in that, The method further includes: A machine learning model was trained using historical welding data to establish a mapping relationship between weld position deviation and welding process parameters and robot motion trajectory. During the welding process, the changing trend of the weld position is predicted by the mapping relationship model, and the motion trajectory of the welding robot is adjusted in advance.

9. The method according to claim 1, characterized in that, The method further includes: An adaptive filtering algorithm is adopted to monitor the changing parameters of the welding environment in real time, including welding current and welding voltage. Based on the monitored changing parameters, the demodulation and filtering parameters in the signal processing circuit are dynamically adjusted to improve the ability to suppress interference at the welding site.

10. A miniature weld seam tracking system based on electromagnetic induction, characterized in that, The invention includes a probe for tracking miniature weld seams, a welding robot, and a controller; the probe is fixed to the front end or side of the welding torch of the welding robot via a mounting bracket, and is integrated with the welding torch; the controller is electrically connected to the probe and is used to implement the method as described in any one of claims 1-9.

Images