Intelligent control method and system for thin-walled metal component welding process

Abstract

The present application relates to the technical field of welding, and discloses a kind of thin-walled metal component welding process intelligent control method and system, in which a kind of thin-walled metal component welding process intelligent control method includes: determining initial welding parameter set and carrying out time synchronization;Obtain the time-stamped multi-source original data in welding process;Time alignment is carried out;Respectively calculate current stability index, heat accumulation index and molten pool width measured value, obtain key deviation index;Based on key deviation index, respectively calculate to obtain current adjustment amount, pulse parameter adjustment amount and position compensation amount;Get adjustment amount set;Limit amplitude update is carried out to current welding parameter, and step transition method and high-priority fast instruction are used to obtain updated current welding parameter set;Record welding process data log, judge welding state;Comprehensive evaluation and data storage are carried out to welding process;The present application comprehensively responds to the multiple dynamic changes in welding process, and improves the stability and consistency of welding quality.

Description

Technical Field

[0001] This invention relates to the field of welding technology, and more specifically, to an intelligent control method and system for the welding process of thin-walled metal components. Background Technology

[0002] Thin-walled metal components are widely used in high-end equipment manufacturing fields such as aerospace and precision instrument manufacturing. Taking metal diaphragm boxes as an example, their diaphragm thickness is typically in the range of 0.05 to 0.1 mm, requiring circumferential welding of the inner and outer diameters using micro-beam argon arc welding technology. Due to the extremely thin material, narrow welding area, and low heat capacity, the welding process is extremely sensitive to parameter fluctuations; any slight parameter deviation can lead to serious defects such as burn-through, lack of fusion, and porosity.

[0003] Traditional welding methods for thin-walled metal components employ a fixed-parameter control strategy, where welding parameters are set before welding based on the workpiece material and thickness, and maintained constant throughout the welding process. However, in actual welding, the workpiece temperature gradually increases as welding progresses, resulting in a significant heat accumulation effect. For ultra-thin-walled components with a thickness of only 0.0635 mm, heat accumulation leads to an increase in actual heat input. If a fixed welding current is continued, burn-through is highly likely to occur in the later stages of welding. Simultaneously, due to the inherent poor rigidity and susceptibility to deformation of thin-walled components, slight gap fluctuations are unavoidable during clamping. These gap variations alter the welding conditions, making it difficult for fixed parameters to adapt to these changes. Furthermore, ultra-thin metal foils exhibit thickness variations during manufacturing, leading to different heat input requirements at different locations. During circumferential welding, the welding position and workpiece orientation are constantly changing, and the welding conditions vary at different locations due to the influence of gravity and airflow.

[0004] These problems lead to unstable quality in the welding of thin-walled metal components using traditional fixed-parameter welding methods, resulting in defects such as burn-through and lack of fusion, and a low product qualification rate, typically only 70% to 80%. While existing technologies have made some progress in welding process monitoring and defect identification, most methods only focus on post-defect detection and lack a real-time adaptive parameter adjustment mechanism based on monitoring results, failing to form a complete closed-loop control system. Some studies have proposed parameter adjustment methods, but these often employ single-parameter control or complex machine learning algorithms. The former struggles to cope with the coupled effects of multiple factors, while the latter has long computation times that fail to meet real-time requirements.

[0005] Therefore, there is an urgent need for a control method that can intelligently adjust welding parameters according to the real-time status of the welding process in order to solve technical problems such as heat accumulation effect, uneven gap, thickness deviation, and position change in the welding process of thin-walled metal components, and improve the stability of welding quality and product qualification rate. Summary of the Invention

[0006] This invention provides an intelligent control method and system for the welding process of thin-walled metal components, which solves the technical problems of unstable welding quality and low pass rate caused by the coupled influence of multiple factors such as heat accumulation effect, molten pool width deviation, and welding instability in the welding process of thin-walled metal components.

[0007] This invention provides an intelligent control method for the welding process of thin-walled metal components, comprising:

[0008] Determine the initial welding parameter set for the thin-walled metal component to be welded and synchronize it with time to obtain the time synchronization calibration parameters;

[0009] Based on the initial welding parameter set, the electric arc is ignited and circumferential welding begins, acquiring multi-source raw data with timestamps during the welding process; time alignment is performed by combining time synchronization calibration parameters to obtain time-aligned multi-source synchronized data.

[0010] Based on the time-aligned multi-source synchronous data, the current stability index, thermal accumulation index and measured value of molten pool width were calculated respectively, and the key deviation index was obtained by the threshold comparison method.

[0011] Based on the key deviation index, the current adjustment, pulse parameter adjustment, and position compensation are calculated respectively; a priority synthesis method is used to obtain the set of adjustment amounts.

[0012] The current welding parameters are obtained by combining the set of adjustment values ​​and the initial set of welding parameters; the current welding parameters are then updated with a limited amplitude, and the updated set of current welding parameters is obtained by using a step-by-step transition method and high-priority fast instructions;

[0013] Based on the updated current welding parameter set, the welding process data log is recorded to determine the welding status. If the welding has not been terminated, the welding process continues until a welding completion confirmation signal is received.

[0014] Based on the welding completion confirmation signal, the welding process is comprehensively evaluated and data is stored to obtain a welding summary report.

[0015] In a preferred embodiment, determining the initial set of welding parameters for the thin-walled metal component to be welded includes:

[0016] Establish a materials and processes database;

[0017] The material property dataset and welding requirement dataset of the thin-walled metal component to be welded are obtained. Material database matching and process knowledge base retrieval are used to determine the applicable welding method and shielding gas type according to the material type, the welding current reference value is determined according to the nominal thickness, and the heat dissipation rate and heat accumulation trend are estimated according to the thermophysical parameters to obtain the recommended process parameter range.

[0018] Based on the recommended process parameter range and welding requirement dataset, a multi-objective optimization method is adopted to comprehensively consider the penetration depth, weld width, and heat-affected zone size to obtain the initial welding parameter set.

[0019] In a preferred embodiment, the acquisition of timestamped multi-source raw data during the welding process includes welding process information from a sensor system, a rotary drive system, and a gas flow controller. The welding process information includes raw data of electrical parameter time series, raw data of temperature field, raw data of molten pool image sequence, raw data of welding position information, and raw data of gas protection status information. The sensor system includes a current sensor, a voltage sensor, an infrared temperature sensor, and a narrowband filter high-speed camera.

[0020] In a preferred embodiment, the calculation of the current stability index, thermal accumulation index, and measured values ​​of the molten pool width based on the time-aligned multi-source synchronization data includes:

[0021] Based on the original time series data of electrical parameters in the time-aligned multi-source synchronous data, the sliding window statistical method is used to calculate the current stability index. All current sampling data in the current window and the previous window are collected, and the sum of all current sampling data values ​​is divided by the number of sampling points to obtain the current mean. The difference between each current sampling value in the collection window and the current mean is calculated. The standard deviation of the current difference is calculated to obtain the current stability index.

[0022] The temperature field raw data in the time-aligned multi-source synchronous data is calculated using the temperature increment method to calculate the difference between the current temperature and the initial temperature, where the initial temperature is the temperature of the thin-walled metal component to be welded before welding begins; the temperature difference is normalized to obtain the heat accumulation index.

[0023] Based on the original data of the molten pool image sequence in the time-aligned multi-source synchronous data, the molten pool image is preprocessed and binarized to determine the center position of the molten pool and measure the pixel width in the vertical direction; the actual size is obtained by converting the pixel resolution coefficient.

[0024] In a preferred embodiment, the calculation of the current adjustment amount includes:

[0025] Based on the width deviation and thermal accumulation deviation in the key deviation indicators, set the thermal accumulation classification threshold; determine the thermal accumulation adjustment level based on the thermal accumulation status; set the width deviation classification threshold, including the upper limit threshold and the lower limit threshold of the width deviation;

[0026] When there is severe heat accumulation, the current adjustment amount is calculated by multiplying the current peak current by the severe heat accumulation adjustment factor.

[0027] When the temperature is at a moderate level of heat accumulation, the current adjustment amount is calculated by multiplying the current peak current by the moderate heat accumulation adjustment factor.

[0028] When there is mild heat accumulation, the current is slightly reduced by using a mild heat accumulation adjustment coefficient, and at the same time, it is corrected according to the width deviation, and the current adjustment amount is calculated.

[0029] When there is no heat accumulation, the width adjustment level is determined by comparing the width deviation magnitude with the width deviation threshold.

[0030] In a preferred embodiment, the calculation of the pulse parameter adjustment amount includes:

[0031] Based on the stability deviation among the key deviation indicators, a stability threshold is set to determine the stability status of the welding process. If the stability deviation exceeds the stability threshold, it indicates that the current stability index is out of control and the welding process is unstable. The pulse parameters need to be adjusted to improve stability. If the stability deviation is within the safe range, it indicates that the welding process is stable and the pulse parameters do not need to be adjusted. The pulse frequency adjustment is set to zero.

[0032] The pulse parameters are adjusted by reading the currently set pulse frequency value from the welding equipment control system, using a pulse frequency adjustment coefficient, multiplying the current pulse frequency by the pulse frequency adjustment coefficient, and obtaining the pulse frequency adjustment amount.

[0033] In a preferred embodiment, the calculation of the position compensation amount includes:

[0034] Establish a preset position compensation table and determine the position compensation amount based on the welding position angle. The position compensation amount includes welding current compensation, welding speed compensation, and wire feed speed compensation. Find the corresponding compensation value in the preset compensation table based on the current welding position angle and use interpolation to obtain the accurate position compensation amount.

[0035] In a preferred embodiment, the step of combining the adjustment amount with the initial welding parameter set, and obtaining the current welding parameters through limited update and rapid execution includes:

[0036] The parameter adjustment process is divided into steps using a phased transition method, and the adjustment is completed gradually within a continuous control cycle to avoid the impact of sudden parameter changes on welding stability. High-priority fast commands are used to quickly transmit the updated welding parameters to the welding equipment through a high-speed communication interface to ensure the real-time performance and accuracy of parameter adjustment.

[0037] In a preferred embodiment, the comprehensive evaluation and data storage of the welding process includes:

[0038] Statistical analysis of welding process data yields statistical characteristics of the welding process;

[0039] Quality is graded based on statistical characteristics of the welding process, and the quality of width control, heat accumulation control, and stability are comprehensively evaluated. A weighted scoring method is used to calculate the overall quality evaluation, and the comprehensive evaluation result of welding quality is obtained.

[0040] Based on the comprehensive evaluation results of welding quality and the welding process data log, the welding records are stored in the database to obtain the welding record archive;

[0041] Based on the comprehensive evaluation results of welding quality, a welding task completion report is generated.

[0042] In a preferred embodiment, an intelligent control system for the welding process of thin-walled metal components is used to execute the above-described intelligent control method for the welding process of thin-walled metal components, including:

[0043] The initial parameter determination and time calibration module is used to determine the initial welding parameter set of the thin-walled metal component to be welded and to synchronize the time to obtain the time synchronization calibration parameters.

[0044] The multi-source information acquisition and alignment module, based on the initial welding parameter set, ignites the electric arc and begins circumferential welding, acquiring multi-source raw data with timestamps during the welding process; it then performs time alignment by combining time synchronization calibration parameters to obtain time-aligned multi-source synchronized data.

[0045] The rapid extraction module for welding status parameters calculates current stability index, heat accumulation index and measured value of weld pool width based on time-aligned multi-source synchronous data, and obtains key deviation index through threshold comparison method.

[0046] The intelligent parameter adjustment calculation module calculates the current adjustment, pulse parameter adjustment, and position compensation based on key deviation indicators; and uses a priority synthesis method to obtain the set of adjustment values.

[0047] The welding parameter update and execution module is used to combine the adjustment set and the initial welding parameter set to obtain the current welding parameters; it performs a limited update on the current welding parameters, using a step-by-step transition method and high-priority fast instructions to obtain the updated current welding parameter set.

[0048] The welding process loop control and termination judgment module records the welding process data log based on the updated current welding parameter set, judges the welding status, and continues the welding process until a welding completion confirmation signal is received if the welding has not been terminated.

[0049] The welding quality evaluation and data storage module, based on the welding completion confirmation signal, performs a comprehensive evaluation of the welding process and stores the data to generate a welding summary report.

[0050] The beneficial effects of this invention are as follows: By establishing a hardware timestamp synchronization mechanism and a delay compensation table, the problem of time synchronization of data acquisition from different sensors is solved. The sampling frequencies and inherent delays of current sensors, temperature sensors, and high-speed cameras are different. Traditional methods directly fuse data from different times, which leads to inaccurate state assessment. This invention adds a timestamp to each sensor data, measures and compensates for the inherent delay of each sensor, and uses time alignment algorithms and interpolation methods to align data with different sampling rates to a unified time reference, ensuring that the fused multi-source data corresponds to the same actual welding time. At the same time, narrowband filtering technology is used to effectively filter out interference from strong arc light, and a fast image processing method is used to shorten the processing time of a single frame of molten pool image compared to traditional methods, laying the foundation for accurate state assessment and real-time parameter adjustment, enabling the control system to accurately perceive the real-time state of the welding process.

[0051] To address the coupled effects of multiple factors such as heat accumulation, weld pool width deviation, and welding instability during the welding of thin-walled metal components, a graded adjustment strategy and a priority synthesis method are proposed. Based on the severity of the heat accumulation index, the welding state is divided into three levels: severe heat accumulation, moderate heat accumulation, and normal state. Different current adjustments are applied to different levels. In the case of severe heat accumulation, the current is preferentially reduced to avoid burn-through, while in the normal state, fine adjustments are made based on the weld pool width deviation. Simultaneously, a position preset compensation table and pulse parameter adjustment rules are introduced to achieve coordinated optimization control of multiple parameters such as current, pulse frequency, and position compensation. This avoids control lag caused by complex algorithms, ensuring both control effectiveness and real-time requirements. Compared to traditional fixed-parameter methods and single-parameter adjustment methods, this invention can comprehensively address various dynamic changes during the welding process, improving the stability and consistency of welding quality. Attached Figure Description

[0052] Figure 1 This is a flowchart of an intelligent control method for the welding process of thin-walled metal components according to the present invention;

[0053] Figure 2 This is a block diagram of an intelligent control system for the welding process of thin-walled metal components in this invention. Detailed Implementation

[0054] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0055] At least one embodiment of the present invention discloses an intelligent control method for the welding process of thin-walled metal components, such as... Figure 1 As shown, it includes:

[0056] Step 1: Determine the initial welding parameter set for the thin-walled metal component to be welded and synchronize it with time to obtain the time synchronization calibration parameters;

[0057] Specific implementation methods include:

[0058] Based on the material testing equipment and technical documents corresponding to the thin-walled metal components to be welded, material property information and welding requirement information are obtained, resulting in a material property dataset and a welding requirement dataset. The material property dataset includes material type (e.g., stainless steel SUS304, precision alloy Inconel 718, etc.), nominal thickness (e.g., 0.0635mm), and thermophysical parameters (including melting point temperature, thermal conductivity, specific heat capacity, density, etc.). The welding requirement dataset includes weld type (butt weld or lap weld) and quality standards (penetration depth range, weld width range, surface finish requirements, etc.).

[0059] Based on material property datasets and thermophysical parameters, a comprehensive approach combining material database matching and heat conduction calculation methods is used to determine the recommended range of process parameters.

[0060] Specifically, the recommended implementation steps for determining the range of process parameters are as follows:

[0061] Establish a materials and processes database. The database contains physical property parameters, chemical composition information, recommended welding methods, shielding gas types, current coefficient ranges, thermophysical parameters and other process data for different material types (such as stainless steel, precision alloys, titanium alloys, etc.). Each material is classified and stored according to its thickness range to form a structured process parameter data table.

[0062] Based on the input material type information and nominal thickness value of the thin-walled metal component to be welded, the corresponding material classification and thickness range are searched in the material process database to obtain a set of basic process parameters, including recommended welding method type, shielding gas type, current coefficient value range, etc.

[0063] From the range of matched current coefficient values, select an appropriate coefficient value based on the specific grade of the material, performance requirements, and welding quality standards. Multiply the selected current coefficient by the nominal thickness, and multiply the result by 1000 to convert the unit and obtain the reference current value.

[0064] The pulse parameter range is adjusted based on thermophysical parameters, and the heat accumulation trend is estimated using heat conduction calculation methods. The higher the thermal conductivity, the faster the heat dissipation and the weaker the heat accumulation effect; the lower the thermal conductivity, the slower the heat dissipation and the stronger the heat accumulation effect. The recommended range of pulse parameters is adjusted according to the thermal conductivity. When the thermal conductivity is high, a higher pulse frequency and a smaller duty cycle are used, and when the thermal conductivity is low, a lower pulse frequency and a larger duty cycle are used.

[0065] The recommended process parameter range is formed by combining the basic process parameter set with the thermophysical parameter adjustment results (i.e., based on the welding method, shielding gas, and reference current obtained by matching the database, combined with the pulse frequency adjustment coefficient and duty cycle adjustment coefficient calculated by the thermophysical parameters, the reference current value is finely adjusted to form the current range, and the pulse parameters are optimized to form the frequency and duty cycle range). This range includes welding method, shielding gas, reference current range, pulse frequency range, and duty cycle range.

[0066] Based on the recommended process parameter range and welding requirement dataset, a multi-objective optimization method is adopted to comprehensively consider multiple objectives such as penetration depth, weld width, and heat-affected zone size to determine the initial welding parameter set. Specifically, the arc-starting current is determined based on the recommended process parameter range, taking 50% to 70% of the reference current as the arc-starting current setting value (too small an arc-starting current will lead to difficulty in arc initiation, while too large an arc-starting current will cause the arc-starting point to burn through); the reference peak current is determined, using the reference current value as the peak current setting value; the reference base current is determined, taking 40% to 70% of the peak current as the base current setting value, the ratio of the base current to the peak current affects the amount of heat input; the pulse frequency is determined, selecting the middle value from the recommended pulse frequency range as the initial setting value, the pulse frequency affects the shaking and stirring effect of the molten pool; the pulse duty cycle is determined, selecting the middle value from the recommended duty cycle range as the initial setting value, the duty cycle is the proportion of the peak current duration to the pulse period; the rotation speed is determined, selecting an appropriate value within the range of 2 to 6 revolutions per minute based on the component size and welding requirements, the rotation speed determines the welding line speed; the shielding gas flow rate is determined, selecting an appropriate value within the range of 10 to 20 liters per minute based on the component size and protection requirements, too small a flow rate will result in insufficient protection, while too large a flow rate will blow away the molten pool, thus obtaining the initial welding parameter set.

[0067] Based on the initial welding parameter set, parameter issuance commands are used to transmit each parameter to the welding power source, rotary drive system, gas flow controller and other actuators through the communication interface of the control system. At the same time, sensor systems such as current sensor, voltage sensor, infrared temperature sensor, and narrowband filter high-speed camera are initialized to obtain the sensor system initialization completion status.

[0068] Based on the sensor system's initialization completion status, a unified time reference is established using a hardware timestamp synchronization mechanism. Specifically, the main controller's system clock is used as the time reference, with a clock frequency of 1 kHz and a time accuracy of 1 millisecond. A timestamp recording module is configured for each sensor, automatically adding a timestamp to the data packet at the moment of data acquisition to record the absolute time of data acquisition.

[0069] The inherent delay time of each sensor was measured using a standard test signal calibration method. A standard test signal at a known moment was simultaneously input to each sensor, and the timing of each sensor's output response was recorded. The difference between the two times is the inherent delay time of that sensor. Measurement results show that the inherent delay time of the current and voltage sensors is less than 1 millisecond, the inherent delay time of the infrared temperature sensor is 8 to 15 milliseconds, and the inherent delay time of the narrowband filter high-speed camera is 15 to 30 milliseconds. The inherent delay times of each sensor were recorded in a delay compensation table to obtain the time synchronization calibration parameters.

[0070] This step outputs the initial welding parameter set and time synchronization calibration parameters.

[0071] Furthermore, a parameter intelligent recommendation method based on historical data can replace the benchmark current value calculation formula and multi-objective optimization method. Specifically, historical welding records similar to the current workpiece in terms of material, thickness, and geometry are retrieved from the product database. Welding parameters rated as excellent in the historical records are extracted, and the mean and standard deviation of these parameters are calculated. The mean is used as the recommended parameter, and twice the standard deviation is used as the safe range for parameter adjustment. The purpose of this method is to fully utilize historical welding experience, improve the accuracy of initial parameters, and reduce parameter optimization time. For new products or workpieces being welded for the first time, the empirical formula method is still used; for workpieces produced repeatedly, the historical data recommendation method can improve the accuracy of initial parameters.

[0072] Step 2: Based on the initial welding parameter set, ignite the electric arc and start circumferential welding, and acquire multi-source raw data with timestamps during the welding process; combine the time synchronization calibration parameters to perform time alignment, and obtain time-aligned multi-source synchronized data;

[0073] Specific implementation methods include:

[0074] Based on the initial welding parameter set, a welding start command triggers the welding power supply and rotary drive system to begin operation. The welding power supply outputs an arc-starting current and generates a high-frequency, high-voltage arc-ignition signal, breaking down the air between the tungsten electrode of the welding torch and the workpiece to form a conductive path, establishing a welding arc and igniting it. After the arc ignites, the current quickly stabilizes to the arc-starting current value, forming an initial molten pool on the workpiece surface. As the current gradually increases to the reference peak current according to the preset rise time curve, the molten pool size gradually expands and stabilizes, ensuring the arc is in a stable welding state. Simultaneously, the rotary drive system drives the workpiece to rotate at a set rotation speed, and the welding torch moves relative to the workpiece along the circumferential seam direction, initiating the circumferential seam welding process.

[0075] When the electric arc is in a stable welding state, the sensor system, the rotary drive system and the gas flow controller are used to synchronously collect welding process information to obtain multi-source raw data with timestamps, including raw data of electrical parameter time series, raw data of temperature field, raw data of molten pool image sequence, raw data of welding position information and raw data of gas protection status information.

[0076] The sensor system includes a current sensor, a voltage sensor, an infrared temperature sensor, and a narrowband filter high-speed camera. Specifically, the current and voltage sensors acquire instantaneous values ​​of welding current and voltage in real time at a sampling frequency of 2 kHz. Each sampling point records the timestamp of the sampling moment and the corresponding current and voltage values ​​to obtain the raw data of the electrical parameter time series. The infrared temperature sensor measures the temperature distribution of the weld and heat-affected zone in real time at a sampling frequency of 20 Hz. The measurement point is located at an appropriate distance behind the welding point, and the temperature value can reflect the degree of heat accumulation. Each sampling point records the timestamp of the sampling moment and the measured temperature value to obtain the raw data of the temperature field. The narrowband filter high-speed camera captures the molten pool morphology in real time at a frame rate of 80 frames per second. It uses a narrowband filter to capture the thermal radiation signal of the molten pool and filter out the interference of strong arc light. Each frame of the image records the timestamp of the shooting moment and the complete pixel matrix data to obtain the raw data of the molten pool image sequence.

[0077] The encoder of the rotary drive system reads the actual values ​​of the current welding position angle and rotation speed in real time at a sampling frequency of 100 Hz. The encoder resolution is 0.1 degrees, which can accurately reflect the rotation position of the workpiece. Each sampling point records the timestamp, position angle and speed value to obtain the raw data of welding position information.

[0078] The feedback signal of the gas flow controller reads the actual flow value of the protective gas in real time at a sampling frequency of 10 Hz; the flow controller uses a mass flow meter with a measurement accuracy of 0.5 liters per minute, and records the timestamp and flow value at each sampling point to obtain the raw data of the gas protection status information.

[0079] Based on the timestamped multi-source raw data and time synchronization calibration parameters, a time alignment algorithm is used to align sensor data with different sampling rates and delays to a unified time reference, resulting in time-aligned multi-source synchronized data.

[0080] Specifically, the data processing cycle of the control system is set to 60 milliseconds, meaning that data fusion and control decisions are performed every 60 milliseconds. At the beginning of each processing cycle, data is extracted from the data buffers of the sensor system, rotary drive system, and gas flow controller: For the raw data of electrical parameter time series, all sampling points within the corresponding time period of the current processing cycle are extracted (120 sampling points at a sampling frequency of 2 kHz within a 60-millisecond cycle). Based on the timestamp and delay compensation table (the delay of the current sensor is less than 1 millisecond and can be ignored), the actual welding time corresponding to these sampling points is determined. For the raw data of temperature field, the temperature measurement value closest to the current processing time is extracted. Based on the timestamp, the inherent delay time in the delay compensation table (assumed to be 10 milliseconds) is subtracted to obtain the actual welding time corresponding to the raw data of temperature field. Since the temperature change is relatively slow, the temperature value after delay compensation can better reflect the current heat accumulation state. For the raw data of molten pool image sequence, the image frame closest to the current processing time is extracted. Based on the image timestamp, the inherent delay time (assumed to be 20 milliseconds) is subtracted to obtain the actual welding time corresponding to the image. For the raw data of welding position information and gas protection status information, time alignment is also performed based on the timestamp and delay compensation. A linear interpolation method is used to interpolate data with different sampling rates to a unified processing time.

[0081] After time alignment and interpolation, this step outputs time-aligned multi-source synchronous data, including time-aligned raw data of electrical parameter time series, raw data of temperature field, raw data of molten pool image sequence, raw data of welding position information, and raw data of gas protection status information.

[0082] Furthermore, acoustic emission sensors can be used as a supplementary sensing method to enhance the comprehensiveness of welding condition monitoring. Acoustic emission sensors can detect high-frequency acoustic signals generated during welding, which are closely related to physical processes such as molten pool flow, gas escape, and crack formation. Specifically, acoustic emission sensors are installed on the workpiece fixture. The sensor's frequency response range is 50 to 500 kHz, and the sampling frequency is 1 MHz. Acoustic emission signals are acquired in real time during welding, and characteristic parameters such as amplitude, energy, and duration are recorded. A sudden increase in the energy of the acoustic emission signal may indicate the presence of defects such as porosity or cracks; a change in the signal frequency characteristics may indicate a change in the molten pool flow state. Integrating acoustic emission characteristic parameters with other sensing information can improve the early identification capability of welding defects. The purpose of this alternative embodiment is to achieve a more comprehensive state perception of the welding process and improve the sensitivity and accuracy of defect identification by adding acoustic emission monitoring methods.

[0083] Step 3: Calculate the current stability index, thermal accumulation index and measured value of molten pool width based on the time-aligned multi-source synchronous data, and obtain the key deviation index through the threshold comparison method.

[0084] Specific implementation methods include:

[0085] Based on the time-aligned time series data of electrical parameters, a sliding window statistical method is used to calculate the current stability index. Specifically, the sliding window length is set to 80 milliseconds, which contains 160 current sampling points (sampling frequency 2 kHz). At the current processing time, all current sampling values ​​within the past 80 milliseconds are extracted, and all current sampling data within the window are collected. Then, 80 milliseconds are traced back from the current processing time, and all current sampling values ​​within that time period are extracted, forming a current array containing 160 data points. All current sampling values ​​within the window are summed to obtain the total current value. The total current value is divided by the number of sampling points (i.e., 160) to obtain the average current within the time window, reflecting the average current level of the current period. The current stability index is calculated based on the average current value by calculating the difference between each current sampling value within the collection window and the average current value. The standard deviation of the difference is then calculated to obtain the current stability index, reflecting the severity of current fluctuations.

[0086] When the welding process is stable, the current fluctuation is small, and the current stability index is low. When unstable factors occur in the welding process (such as sudden gap changes, molten pool sloshing, etc.), the current fluctuation increases, and the current stability index increases. A stability threshold is set at 10% of the average current. When the current stability index exceeds this threshold, the welding process is determined to be unstable, and the pulse parameters need to be adjusted to improve stability.

[0087] Based on the time-aligned raw temperature field data, a temperature increment calculation method is used to obtain the heat accumulation index. Specifically, before welding begins, the initial ambient temperature of the workpiece and fixture (typically room temperature, ranging from 15 to 30 degrees Celsius) is recorded. During welding, the temperature at a point 8 mm behind the weld is measured in real time, and the temperature rise is calculated using a temperature difference calculation method. The specific calculation method for the temperature difference is as follows: the real-time temperature value at a point 8 mm behind the weld is read from an infrared temperature sensor; the initial ambient temperature data of the workpiece and fixture, pre-recorded before welding begins, is retrieved; the current measured temperature is subtracted from the initial ambient temperature to obtain the temperature rise, which reflects the degree of heat accumulation during the welding process.

[0088] The heat accumulation index is obtained by normalization to eliminate the influence of different ambient temperature conditions. The specific calculation method is as follows: use the temperature rise as the basic data; call the pre-recorded initial ambient temperature as the normalization benchmark; divide the temperature rise by the initial ambient temperature to obtain the heat accumulation index. The heat accumulation index can provide a consistent heat accumulation evaluation standard under different ambient temperature conditions.

[0089] Based on the original data of the time-aligned molten pool image sequence, a fast image processing method was used to obtain the measured value of the molten pool width. Specifically, the image processing procedure is as follows: the molten pool image is preprocessed by filtering to remove image noise; histogram equalization is used to enhance image contrast; a fixed threshold binarization method is used to segment the image into the molten pool region and the background region to obtain a binary image of the molten pool; the center position of the molten pool is determined in the binary image of the molten pool, and the centroid coordinates of the molten pool region are calculated using the image moment method; the continuous width of white pixels is counted along the transverse section perpendicular to the welding direction to obtain the pixel value of the molten pool width.

[0090] The method for calculating the measured value of the molten pool width is as follows: obtain the pixel value of the molten pool width from the image processing process; obtain the pixel resolution coefficient and call the actual size parameter corresponding to each pixel determined during the camera calibration process; perform unit conversion calculation, multiply the pixel value of the molten pool width by the pixel resolution coefficient to obtain the measured value of the molten pool width.

[0091] Based on the current stability index, thermal accumulation index, and measured molten pool width, a threshold comparison method was used to obtain three key deviation indices: width deviation, thermal accumulation deviation, and stability deviation. The specific calculation process is as follows:

[0092] Calculate the width deviation by obtaining the measured value of the molten pool width, which is the actual measurement value obtained during image processing; set the target molten pool width according to the nominal thickness and welding requirements (generally, the target molten pool width is 2 to 4 times the nominal thickness); subtract the target molten pool width from the measured molten pool width to obtain the difference; divide the width difference by the target width to obtain the width deviation; when the width deviation is positive, it indicates that the molten pool is too wide and the heat input is too large, and when the width deviation is negative, it indicates that the molten pool is too narrow and the heat input is insufficient.

[0093] Calculate the thermal accumulation deviation by comparing the thermal accumulation index with a safety threshold (usually set to 0.3 to 0.5 for thin-walled components, and 0.4 in this embodiment). When the thermal accumulation index is less than the safety threshold, the thermal accumulation deviation is zero, indicating that no thermal accumulation compensation is required. When the thermal accumulation index is greater than or equal to the safety threshold, the thermal accumulation deviation is equal to the thermal accumulation index minus the safety threshold.

[0094] The stability deviation is calculated by comparing the current stability index with the stability threshold. When the current stability index is less than the stability threshold, the stability deviation is zero, indicating that the welding process is stable. When the current stability index is greater than or equal to the stability threshold, the stability deviation is equal to the current stability index minus the stability threshold.

[0095] This step outputs key deviation metrics, including width deviation, thermal accumulation deviation, and stability deviation.

[0096] Furthermore, the molten pool area assessment method can replace the molten pool width assessment method to more comprehensively reflect the penetration state. Molten pool width only reflects the size of the molten pool in the direction perpendicular to the weld, while molten pool area can comprehensively reflect the length and width of the molten pool, more accurately characterizing the heat input. Specifically, connected component analysis is performed on the binary image of the molten pool, and the total number of white pixels is counted, which is the pixel value of the molten pool area. The pixel area is converted into the actual area according to the camera calibration parameters to obtain the measured value of the molten pool area. The target molten pool area is determined according to the nominal thickness and welding requirements, and the measured area is compared with the target area to obtain the area deviation. The purpose of this alternative embodiment is to improve the accuracy of welding state assessment through the more comprehensive feature parameter of molten pool area. The computational load of the area assessment method is slightly increased, but it can still be completed in a short time, meeting real-time requirements.

[0097] Step 4: Based on the key deviation index, calculate the current adjustment, pulse parameter adjustment, and position compensation amount respectively; use the priority synthesis method to obtain the set of adjustment amounts;

[0098] The specific implementation process is as follows:

[0099] Based on thermal accumulation deviation and width deviation, a graded adjustment strategy is adopted. The direction and magnitude of current adjustment are determined according to the current welding state to obtain the current adjustment amount. Specifically, thermal accumulation classification thresholds are set, including mild thermal accumulation threshold, moderate thermal accumulation threshold, and severe thermal accumulation threshold, which correspond to different levels of thermal accumulation state judgment standards. Width deviation classification thresholds are set, including upper limit threshold and lower limit threshold, which are used to determine whether the molten pool width is within a reasonable range.

[0100] When there is severe heat accumulation, that is, when the heat accumulation index is greater than or equal to the severe heat accumulation threshold, the heat accumulation effect is the main problem and the welding current needs to be reduced. The current peak current value is read from the welding equipment control system, and the current peak current is multiplied by the severe heat accumulation adjustment coefficient to obtain the current adjustment amount.

[0101] When the heat accumulation is moderate, that is, when the heat accumulation index is between the moderate heat accumulation threshold and the severe heat accumulation threshold, the welding current needs to be appropriately reduced. The current adjustment amount is obtained by multiplying the current peak current by the moderate heat accumulation adjustment coefficient.

[0102] When the heat accumulation is mild, i.e., the heat accumulation index is between the mild and moderate heat accumulation thresholds, a preventive fine-tuning strategy is adopted, which is combined with the width deviation for comprehensive adjustment. The current is slightly reduced by the mild heat accumulation adjustment coefficient, and at the same time, it is corrected according to the width deviation to calculate the current adjustment amount.

[0103] When there is no heat accumulation, i.e., the heat accumulation index is less than the mild heat accumulation threshold, the width adjustment level is determined by comparing the width deviation magnitude with the width deviation classification threshold. When the width deviation is greater than the upper limit threshold, it indicates that the molten pool is too wide, and the current adjustment amount is calculated using the excessive width adjustment coefficient. When the width deviation is less than the lower limit threshold, it indicates that the molten pool is too narrow, and the current adjustment amount is calculated using the insufficient width adjustment coefficient. When the width deviation is within the safe range, the current adjustment amount is zero.

[0104] Based on the stability deviation, pulse parameter adjustment rules are adopted to obtain the pulse parameter adjustment amount. Specifically, a stability classification threshold parameter is set to determine the stability status of the welding process; it is determined whether the stability deviation is greater than the stability safety threshold. If the stability deviation exceeds the threshold, it indicates that the current stability index is out of control and the welding process is unstable, requiring adjustment of the pulse parameters to improve stability; if the stability deviation is within the safe range, it indicates that the welding process is stable, the pulse parameters do not need to be adjusted, and the pulse frequency adjustment amount is set to zero.

[0105] The pulse parameters are adjusted by reducing the pulse frequency. The current pulse frequency value is read from the welding equipment control system, and the pulse frequency adjustment coefficient is used to multiply the current pulse frequency by the pulse frequency adjustment coefficient to obtain the pulse frequency adjustment amount.

[0106] Based on the welding position angle, a position preset compensation table is used to obtain the position compensation amount. The position preset compensation table is an empirical data table established based on multiple experiments, recording the welding parameter compensation amounts corresponding to different position angles. In the experiment, the same workpiece was welded at different position angles, welding quality indicators were measured, quality differences at different positions were analyzed, and the required parameter compensation amount for each position was determined. The compensation table sets sampling points at preset angle intervals, including multiple equally spaced angle positions, and each sampling point records the current compensation coefficient for the corresponding position.

[0107] The degree of influence of gravity varies depending on the location: at the top, the molten pool is prone to sagging due to gravity, requiring an appropriate increase in current; at the bottom, the molten pool is supported by gravity, requiring an appropriate decrease in current; and at the side, the influence of gravity is relatively small.

[0108] During actual welding, the compensation table is consulted based on the current welding position angle. If the current angle is exactly equal to the angle of a certain sampling point, the compensation coefficient of that sampling point is directly used. If the current angle is between two sampling points, the compensation coefficient is calculated using a linear interpolation method. The compensation coefficient is multiplied by the reference peak current to obtain the position current compensation amount.

[0109] Based on current adjustment, pulse parameter adjustment, and position compensation, a priority synthesis method is used to obtain a set of adjustment values. Specifically, if the thermal accumulation index is greater than or equal to the severe thermal accumulation threshold, it belongs to a severe thermal accumulation state, and thermal accumulation adjustment has the highest priority. In this case, the final current adjustment is equal to the current adjustment value, without considering width deviation and position compensation, to ensure that the thermal accumulation problem is resolved first and burn-through is avoided. If the thermal accumulation index is less than the severe thermal accumulation threshold, the current adjustment and position compensation are considered comprehensively. The current adjustment value calculated using the aforementioned hierarchical adjustment strategy is used as the basic data. The position compensation value data obtained by looking up the position preset compensation table is added to the current adjustment value to obtain the final current adjustment value.

[0110] This step outputs a set of adjustment values, including the final current adjustment value and the pulse frequency adjustment value. At the same time, the adjustment value of the base current value and the peak current value maintain the same proportional relationship, that is, the adjustment value of the base current value is equal to the ratio of the base current value to the peak current value multiplied by the peak current adjustment value.

[0111] Furthermore, fuzzy control strategies can be used to replace hierarchical adjustment strategies and priority synthesis methods. Fuzzy control does not require precise mathematical models, can handle nonlinear and uncertain problems, and is suitable for complex welding process control. Specifically, a fuzzy controller is established, with input variables including thermal accumulation index, width deviation, and stability deviation, and output variables including current adjustment and pulse frequency adjustment. Fuzzy sets (e.g., low, medium, high) and membership functions (e.g., triangular or trapezoidal functions) are defined for each input and output variable. A fuzzy rule base is established, containing multiple "if-then" rules, such as "if the thermal accumulation index is high and the width deviation is positive, then the current adjustment is reduced." In actual control, the membership degree of the current input variable to each fuzzy set is calculated based on its current value, fuzzy inference is performed according to the fuzzy rules to obtain the fuzzy result of the output variable, and finally, the precise adjustment value is obtained through defuzzification (e.g., the centroid method). The purpose of this alternative embodiment is to achieve more flexible multi-factor comprehensive decision-making through fuzzy control strategies and improve the level of control intelligence. The computational cost of fuzzy control is slightly higher than that of hierarchical strategies, but through optimization of the rule base and algorithm, the computation time is shorter, still meeting real-time requirements.

[0112] Step 5: Combine the adjustment set and the initial welding parameter set to obtain the current welding parameters; perform a limit update on the current welding parameters, and use a step-by-step transition method and high-priority fast instructions to obtain the updated current welding parameter set;

[0113] The specific implementation process is as follows:

[0114] Based on the set of adjustment values, the system adds the corresponding parameters in the initial welding parameter set to the adjustment values ​​to obtain new parameter values, and dynamically registers these new parameter values ​​as the current welding parameters. The current welding parameters are the parameter states maintained by the system in real time, reflecting the latest set values ​​of the parameters during the welding process. Based on the set of adjustment values ​​and the current welding parameters, a limiting update algorithm is used to obtain the limited target parameter set. Specifically, the currently set peak current value is read from the welding equipment control system; the final current adjustment value is used as the adjustment reference; the current peak current is added to the final current adjustment value to obtain the new target peak current.

[0115] Perform a limiting constraint check on the target peak current:

[0116] Check the single adjustment range constraint. If the absolute value of the final current adjustment is greater than the single adjustment range limit ratio of the current peak current, then limit the adjustment amount to within that ratio range.

[0117] Check the cumulative adjustment range constraint, record the reference peak current in the initial welding parameters as a reference, calculate the difference between the target peak current and the reference peak current, and if the absolute value of the difference is greater than the cumulative adjustment range limit ratio of the reference peak current, then limit the adjustment to ensure that the cumulative adjustment amount does not exceed the ratio range.

[0118] Check the physical feasibility constraints of the parameters. The target peak current cannot be lower than the arc ignition current, otherwise a stable arc cannot be maintained. The target peak current cannot exceed the rated output capacity of the welding power source. If the target peak current violates these physical constraints, it should be corrected to the constraint boundary value.

[0119] After amplitude limiting protection, the target peak current that meets all constraints is obtained. Similarly, parameters such as the base current, pulse frequency, and rotation speed are subjected to corresponding amplitude limiting processing to obtain the set of target parameters after amplitude limiting. The amplitude limiting protection mechanism ensures the safety of parameter adjustment and avoids sudden parameter changes or exceeding reasonable ranges that could lead to welding process instability or equipment damage.

[0120] Based on the target parameter set after amplitude limiting, a step-by-step transition method is adopted to obtain a step-by-step parameter update sequence. Specifically, to avoid abrupt changes in the welding process caused by a one-time parameter adjustment, the parameter adjustment process is divided into a preset number of small steps, which are gradually completed over multiple consecutive control cycles. The adjustment amount of each small step is equal to the average distribution ratio of the total adjustment amount.

[0121] In the current control cycle, only the first adjustment is performed, updating the parameters to the values ​​after the first adjustment. In the subsequent three control cycles, the second, third, and fourth adjustments are performed sequentially, eventually allowing the parameters to smoothly transition to the target value. During the parameter transition process, the system continuously monitors the welding status. If the welding status changes (such as further aggravation of heat accumulation), the adjustment amount will be recalculated according to steps 3 and 4, and the target parameters will be updated. The original step-by-step transition sequence is replaced by the new sequence, realizing dynamic continuous adjustment, effectively avoiding welding instability caused by parameter mutations, and improving the robustness of the control system.

[0122] Based on the initial parameter value of the step-by-step parameter update sequence, a high-priority fast command is used to obtain the parameter execution confirmation result. The control system sends a parameter update command to the welding power source, and the welding power source immediately updates its internal control parameters. If the execution is successful, the initial parameter value is recorded as the new current welding parameter; if the execution fails, the original parameters remain unchanged.

[0123] Upon receiving the command, the welding power source immediately updates its internal control parameters and returns an execution confirmation message to the control system via the same communication interface. This confirmation message includes the command sequence number and execution status (success or failure). Upon receiving the confirmation message, the control system checks if the command sequence number matches and if the execution status is successful. If the confirmation is successful, the initial parameter values ​​are recorded as the new current welding parameters for status evaluation and parameter adjustment calculations in the next control cycle. If the confirmation fails or no confirmation message is received within the timeout period, a communication anomaly alarm is triggered, the original parameters are maintained, and the command is resent.

[0124] This step outputs the updated set of current welding parameters.

[0125] Furthermore, predictive feedforward control can be employed to improve the foresight and accuracy of the control. Specifically, when calculating the parameter adjustment, not only the current deviation index but also its changing trend is considered. By calculating the rate of change of the deviation index (such as the growth rate of the heat accumulation index or the rate of change of the width deviation) over multiple consecutive control cycles, the deviation value after the next one or several control cycles is predicted. Based on the predicted deviation value, the parameter adjustment is calculated in advance, and the parameter adjustment begins before the deviation reaches the threshold, thus achieving feedforward control. For example, if the current heat accumulation index is 0.35, which has not yet reached the threshold of 0.4, but the growth rate of the heat accumulation index is 0.03 per cycle, it is predicted that it will reach 0.41 after two cycles, exceeding the threshold. At this time, the current is reduced in advance, and control begins before the heat accumulation reaches the threshold, avoiding lag. The purpose of this alternative embodiment is to reduce control lag and improve the timeliness and effectiveness of the control through feedforward control. Predictive feedforward control requires maintaining historical data for multiple cycles, which slightly increases the computational load, but the increased computation time is within 2 to 3 milliseconds, still meeting the real-time requirements.

[0126] Step 6: Record the welding process data log based on the updated current welding parameter set, determine the welding status, and if the welding has not been terminated, continue the welding process until a welding completion confirmation signal is received.

[0127] The specific implementation process is as follows:

[0128] Based on the updated set of current welding parameters, a data logging module is used to obtain the welding process data log for the current moment. Specifically, at the end of each control cycle, key information for the current moment is packaged and stored in the process data buffer. The process data buffer adopts a circular queue structure with a capacity of 1000 records, capable of storing approximately 1 minute of welding process data (control cycle 60 milliseconds, 1 minute corresponds to 1000 cycles). When the buffer is full, the oldest log record is overwritten. The welding process data log records the current timestamp, welding position angle, current welding parameters (peak current, base current, pulse frequency, pulse duty cycle, rotation speed, gas flow rate), key deviation indicators (width deviation, heat accumulation deviation, stability deviation), measured molten pool width, measured temperature, average current, average voltage, etc.

[0129] After the welding process is completed, all records in the process data buffer will be dumped to the product database for permanent storage, for quality traceability, statistical analysis and process optimization.

[0130] Based on the welding position information, a termination condition judgment algorithm is used to obtain a welding termination signal or a welding continuation signal. Specifically, the current welding position angle is read from the encoder of the rotary drive system. At the start of welding, the initial position angle is recorded, which is usually set to 0 degrees. During the welding process, as the workpiece rotates, the current position angle continuously increases.

[0131] The calculation steps are as follows: read the current angle of the welding head from the rotary positioning system; call the starting angle data recorded at the start of welding; subtract the starting angle from the current welding angle to obtain the completed welding angle difference, which represents the angle through which the welding head has rotated from the starting position to the current position.

[0132] Based on the calculated welding angle difference, it is determined whether the completed angle has reached the termination condition. Theoretically, circumferential welding closes after 360 degrees, but considering the need for overlap at the start and end points to ensure weld continuity and strength, the actual termination angle is set to 365 degrees, meaning welding continues for 5 degrees from the starting point to form an overlap zone. Based on the comparison between the angle difference and the termination angle, corresponding control signals are output. If the completed welding angle difference is greater than or equal to 365 degrees, it indicates that the circumferential welding is complete, and a welding termination signal is output; if the completed welding angle difference is less than 365 degrees, it indicates that the welding is not yet complete, and a welding continue signal is output.

[0133] Based on the welding continue signal or welding terminate signal, a loop control logic or arc termination control strategy is used to obtain the welding continue signal or welding completion confirmation signal, specifically including:

[0134] If a welding continue signal is received, a loop control logic is adopted, and the control system returns to step 2 to continue real-time acquisition and time alignment of key welding process information. In the next control cycle, the updated current welding parameter set is used as the current parameter, and the entire process from step 2 to step 6 is repeated to achieve continuous adaptive adjustment of welding parameters, forming a closed-loop control. The control cycle is approximately 60 milliseconds. During the total welding time of the circumferential seam is approximately 15 seconds (rotation speed 4 revolutions per minute, 365 degrees takes approximately 15 seconds), a total of approximately 250 control cycles are executed, achieving 250 adaptive parameter adjustments to fully cope with dynamic changes during the welding process.

[0135] If a welding termination signal is received, a welding arc-extinguishing control strategy is adopted. The control system sends an arc-extinguishing command to the welding power source, which executes current decay according to a preset arc-extinguishing program. The peak current gradually decays from the current value to the arc-extinguishing current (usually 50% to 80% of the arc-starting current) according to a linear or exponential curve, with a decay time of 1 to 2 seconds. This decay process allows the molten pool to solidify slowly, avoiding craters or cracks caused by sudden current interruption. After the current decay is complete, the welding power source shuts off its output, and the welding arc extinguishes. Simultaneously, the gas flow controller continues to output shielding gas for a delay of 5 to 10 seconds. During this period, the high-temperature weld and heat-affected zone are continuously protected by the shielding gas, preventing oxidation and contamination. After the delay, the gas flow controller shuts off its output, the rotary drive system stops rotating, and the entire welding process ends.

[0136] This step outputs a welding process data log and a welding completion confirmation signal.

[0137] Step 7: Based on the welding completion confirmation signal, perform a comprehensive evaluation of the welding process and store the data to obtain a welding summary report;

[0138] The specific implementation process is as follows:

[0139] Based on welding process data logs, statistical analysis methods are used to obtain statistical characteristics of the welding process, including width deviation, heat accumulation, current, parameter adjustment, and abnormal event records. Specifically, all records are extracted from the process data buffer, and statistical parameters of key indicators are calculated. For width deviation, the average, standard deviation, maximum, and minimum values ​​are calculated to obtain width deviation statistical characteristics; for heat accumulation index, the maximum and average values ​​are calculated to obtain heat accumulation statistical characteristics; for current average, the average value and range of variation are calculated to obtain current statistical characteristics; for the frequency and magnitude of statistical parameter adjustments, the number of effective adjustments (number of adjustments not equal to zero), the maximum adjustment magnitude per adjustment, and the cumulative adjustment amount are calculated to obtain parameter adjustment statistical characteristics; abnormal events are checked for, including stability deviations continuously exceeding thresholds for more than a preset number of cycles, heat accumulation index exceeding abnormal heat accumulation thresholds, width deviation exceeding abnormal width deviation thresholds, communication anomalies, or execution failures. The type, occurrence time, and duration of abnormal events are statistically analyzed to obtain abnormal event records.

[0140] Based on the statistical characteristics of the welding process, a quality grading evaluation method is used to obtain a comprehensive evaluation result of the welding quality. Specifically, this includes:

[0141] The effectiveness of molten pool width control is judged based on the average width deviation. If the absolute value of the average width deviation is less than the excellent width control threshold, it indicates excellent width control. If it is between the excellent width control threshold and the good width control threshold, it indicates good width control. If it is between the good width control threshold and the qualified width control threshold, it indicates qualified width control. If it is greater than the qualified width control threshold, it indicates unqualified width control.

[0142] The effectiveness of heat accumulation control is judged based on the maximum value of the heat accumulation index. If the maximum value of the heat accumulation index is less than the excellent threshold for heat accumulation control, it indicates that the heat accumulation control is excellent; if it is between the excellent threshold and the good threshold for heat accumulation control, it indicates that the heat accumulation control is good; if it is between the good threshold and the qualified threshold for heat accumulation control, it indicates that the heat accumulation control is qualified; if it is greater than the qualified threshold for heat accumulation control, it indicates that the heat accumulation control is unqualified.

[0143] The stability of the welding process is judged based on the abnormal event records. If there are no abnormal events, the stability is excellent. If the number of abnormal events is less than the threshold for the number of events with good stability and the duration is less than the threshold for the duration of events with good stability, the stability is good. If the number of abnormal events is between the threshold for the number of events with good stability and the threshold for the number of events with acceptable stability, or the duration is between the threshold for the duration of events with good stability and the threshold for the duration of events with acceptable stability, the stability is acceptable. If the number of abnormal events exceeds the threshold for the number of events with acceptable stability or the duration exceeds the threshold for the duration of events with acceptable stability, the stability is unacceptable.

[0144] Considering the effectiveness of width control, heat accumulation control, and stability, weighting coefficients for width control, heat accumulation control, and stability are determined. A weighted scoring method is used to calculate the overall quality score. A total weighted score above the excellent quality threshold is considered excellent; between the good and excellent quality thresholds is considered good; between the acceptable and good quality thresholds is considered acceptable; and below the acceptable quality threshold is considered unacceptable. This yields the comprehensive evaluation result of the welding quality.

[0145] Based on the comprehensive evaluation results of welding quality and welding process data logs, database storage operations are used to obtain archived welding records. The welding process data logs are stored in a compressed format to save storage space, with a compression rate of approximately 3 to 5 times. Welding records support multiple query methods, including querying by workpiece number, by date range, and by quality grade, facilitating quality traceability and statistical analysis. Welding records can also be used for process optimization and knowledge base updates. Through data mining of a large number of welding records, optimal parameter patterns under different materials, thicknesses, and welding conditions can be discovered, updating the material database and process knowledge base, and improving the accuracy of initial parameter recommendations.

[0146] Based on the comprehensive evaluation results of welding quality, a welding task completion report is generated using a report generation module. Specifically, the welding task completion report generates a text report containing the following information: workpiece number, welding date and time, quality level (Excellent / Good / Pass / Fail), key parameters (initial current, final current, average current, number of current adjustments), quality indicators (average width deviation, maximum heat accumulation index, number of abnormal events), abnormality alerts (if abnormal events exist, a detailed description of the abnormality type and time of occurrence), and recommended measures (improvement suggestions are given for non-conformities or abnormal situations).

[0147] Welding task completion reports are presented in both text and graphical interface formats: the text format facilitates archiving and printing, while the graphical interface allows operators to quickly view the data. The graphical interface displays color-coded indicators for quality levels (green for excellent, blue for good, yellow for acceptable, and red for unacceptable), as well as trend curves for key parameters and quality indicators. Once generated, the report is automatically sent to the operator's workstation display terminal and can also be uploaded to the upper-level management system via a network interface, enabling information-based management of the production process.

[0148] This step outputs a welding summary report, including a comprehensive evaluation of welding quality, archived welding records, and a welding task completion report.

[0149] An intelligent control system for the welding process of thin-walled metal components, such as Figure 2 As shown, an intelligent control method for performing the above-described welding process of a thin-walled metal component includes:

[0150] The initial parameter determination and time calibration module is used to determine the initial welding parameter set of the thin-walled metal component to be welded and to synchronize the time to obtain the time synchronization calibration parameters.

[0151] The multi-source information acquisition and alignment module, based on the initial welding parameter set, ignites the electric arc and begins circumferential welding, acquiring multi-source raw data with timestamps during the welding process; it then performs time alignment by combining time synchronization calibration parameters to obtain time-aligned multi-source synchronized data.

[0152] The rapid extraction module for welding status parameters calculates current stability index, heat accumulation index and measured value of weld pool width based on time-aligned multi-source synchronous data, and obtains key deviation index through threshold comparison method.

[0153] The intelligent parameter adjustment calculation module calculates the current adjustment, pulse parameter adjustment, and position compensation based on key deviation indicators; and uses a priority synthesis method to obtain the set of adjustment values.

[0154] The welding parameter update and execution module is used to combine the adjustment set and the initial welding parameter set to obtain the current welding parameters; it performs a limited update on the current welding parameters, using a step-by-step transition method and high-priority fast instructions to obtain the updated current welding parameter set.

[0155] The welding process loop control and termination judgment module records the welding process data log based on the updated current welding parameter set, judges the welding status, and continues the welding process until a welding completion confirmation signal is received if the welding has not been terminated.

[0156] The welding quality evaluation and data storage module, based on the welding completion confirmation signal, performs a comprehensive evaluation of the welding process and stores the data to generate a welding summary report.

[0157] In one embodiment of the present invention, a specific example is provided:

[0158] To verify the effectiveness of the method of the present invention, a practical application was carried out using the welding of a certain type of metal diaphragm box as an example. The workpiece is a corrugated diaphragm box made of stainless steel SUS304, with an inner diameter of 5mm, an outer diameter of 10mm, and a diaphragm thickness of 0.0635mm. Circumferential welding of the outer diameter is required. The welding equipment is a micro-beam argon arc welding system, including a T-100i digital micro-precision argon arc welding power supply and a WH300 outer diameter welding mechanism.

[0159] The initial welding parameters are shown in Table 1:

[0160] Table 1: Initial welding parameter diagram;

[0161]

[0162] During the welding process, the control system performed cyclical control according to steps 2 to 6, executing a total of approximately 333 control cycles (total welding time approximately 20 seconds, control cycle 60 milliseconds); typical process data is shown in Table 2 (data from 5 key moments are selected):

[0163] Table 2: Typical data for the welding process;

[0164]

[0165] As shown in Table 2, as the welding process progressed, the workpiece temperature gradually increased from the initial 25 degrees Celsius to 118 degrees Celsius, and the heat accumulation index increased from 0.80 to 3.72. Based on the increase in the heat accumulation index, the control system gradually reduced the peak current from the initial 3.5 amps to 2.9 amps, a reduction of 17%. Through adaptive current adjustment, the weld pool width remained consistently within the range of 0.21 to 0.24 mm, with width deviation controlled within ±10%, resulting in stable welding quality.

[0166] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for intelligent control of the welding process of thin-walled metal components, characterized in that, include: Determine the initial welding parameter set for the thin-walled metal component to be welded and synchronize it with time to obtain the time synchronization calibration parameters; Based on the initial welding parameter set, the electric arc is ignited and circumferential welding begins, acquiring multi-source raw data with timestamps during the welding process; time alignment is performed by combining time synchronization calibration parameters to obtain time-aligned multi-source synchronized data. Based on the time-aligned multi-source synchronous data, the current stability index, thermal accumulation index and measured value of molten pool width were calculated respectively, and the key deviation index was obtained by the threshold comparison method. Based on the key deviation index, the current adjustment amount, pulse parameter adjustment amount, and position compensation amount are calculated respectively. The set of adjustment values ​​is obtained by using a priority synthesis method; The calculation of the current adjustment amount includes: Based on the width deviation and thermal accumulation deviation in the key deviation indicators, set the thermal accumulation classification threshold; determine the thermal accumulation adjustment level based on the thermal accumulation status; set the width deviation classification threshold, including the upper limit threshold and the lower limit threshold of the width deviation; When there is severe heat accumulation, the current adjustment amount is calculated by multiplying the current peak current by the severe heat accumulation adjustment factor. When the temperature is at a moderate level of heat accumulation, the current adjustment amount is calculated by multiplying the current peak current by the moderate heat accumulation adjustment factor. When there is mild heat accumulation, the current is slightly reduced by using a mild heat accumulation adjustment coefficient, and at the same time, it is corrected according to the width deviation, and the current adjustment amount is calculated. When there is no heat accumulation, the width adjustment level is determined by comparing the width deviation magnitude with the width deviation threshold. The calculation of the pulse parameter adjustment includes: Based on the stability deviation among the key deviation indicators, a stability threshold is set to determine the stability status of the welding process. If the stability deviation exceeds the stability threshold, it indicates that the current stability index is out of control and the welding process is unstable. The pulse parameters need to be adjusted to improve stability. If the stability deviation is within the safe range, it indicates that the welding process is stable and the pulse parameters do not need to be adjusted. The pulse frequency adjustment is set to zero. The pulse parameters are adjusted by reading the currently set pulse frequency value from the welding equipment control system, using a pulse frequency adjustment coefficient, multiplying the current pulse frequency by the pulse frequency adjustment coefficient, and obtaining the pulse frequency adjustment amount. The current welding parameters are obtained by combining the set of adjustment values ​​and the initial set of welding parameters; the current welding parameters are then updated with a limited amplitude, and the updated set of current welding parameters is obtained by using a step-by-step transition method and high-priority fast instructions; Based on the updated current welding parameter set, the welding process data log is recorded to determine the welding status. If the welding has not been terminated, the welding process continues until a welding completion confirmation signal is received. Based on the welding completion confirmation signal, the welding process is comprehensively evaluated and data is stored to obtain a welding summary report.

2. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The initial welding parameter set for determining the thin-walled metal component to be welded includes: Establish a materials and processes database; The material property dataset and welding requirement dataset of the thin-walled metal component to be welded are obtained. Material database matching and process knowledge base retrieval are used to determine the applicable welding method and shielding gas type according to the material type, the welding current reference value is determined according to the nominal thickness, and the heat dissipation rate and heat accumulation trend are estimated according to the thermophysical parameters to obtain the recommended process parameter range. Based on the recommended process parameter range and welding requirement dataset, a multi-objective optimization method is adopted to comprehensively consider the penetration depth, weld width, and heat-affected zone size to obtain the initial welding parameter set.

3. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The acquisition of multi-source raw data with timestamps during the welding process includes welding process information from the sensor system, the rotary drive system, and the gas flow controller. The welding process information includes raw data of electrical parameter time series, raw data of temperature field, raw data of molten pool image sequence, raw data of welding position information, and raw data of gas protection status information. The sensor system includes a current sensor, a voltage sensor, an infrared temperature sensor, and a narrowband filter high-speed camera.

4. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The calculation of current stability index, thermal accumulation index, and measured values ​​of molten pool width based on time-aligned multi-source synchronization data includes: Based on the original time series data of electrical parameters in the time-aligned multi-source synchronous data, the sliding window statistical method is used to calculate the current stability index. All current sampling data in the current window and the previous window are collected, and the sum of all current sampling data values ​​is divided by the number of sampling points to obtain the current mean. The difference between each current sampling value in the collection window and the current mean is calculated. The standard deviation of the current difference is calculated to obtain the current stability index. The temperature field raw data in the time-aligned multi-source synchronous data is calculated using the temperature increment method to calculate the difference between the current temperature and the initial temperature, where the initial temperature is the temperature of the thin-walled metal component to be welded before welding begins; the temperature difference is normalized to obtain the heat accumulation index. Based on the original data of the molten pool image sequence in the time-aligned multi-source synchronous data, the molten pool image is preprocessed and binarized to determine the center position of the molten pool and measure the pixel width in the vertical direction; the actual size is obtained by converting the pixel resolution coefficient.

5. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The calculation of the position compensation amount includes: Establish a preset position compensation table and determine the position compensation amount based on the welding position angle. The position compensation amount includes welding current compensation, welding speed compensation, and wire feed speed compensation. Find the corresponding compensation value in the preset compensation table based on the current welding position angle and use interpolation to obtain the accurate position compensation amount.

6. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The adjustment amount is combined with the initial welding parameter set, and the current welding parameters are obtained through limited update and fast execution, including: The parameter adjustment process is divided into steps using a phased transition method, and the adjustment is completed gradually within a continuous control cycle to avoid the impact of sudden parameter changes on welding stability. High-priority fast commands are used to quickly transmit the updated welding parameters to the welding equipment through a high-speed communication interface to ensure the real-time performance and accuracy of parameter adjustment.

7. The intelligent control method for welding thin-walled metal components according to claim 1, characterized in that, The comprehensive evaluation and data storage of the welding process includes: Statistical analysis of welding process data yields statistical characteristics of the welding process; Quality is graded based on statistical characteristics of the welding process, and the quality of width control, heat accumulation control, and stability are comprehensively evaluated. A weighted scoring method is used to calculate the overall quality evaluation, and the comprehensive evaluation result of welding quality is obtained. Based on the comprehensive evaluation results of welding quality and the welding process data log, the welding records are stored in the database to obtain the welding record archive; Based on the comprehensive evaluation results of welding quality, a welding task completion report is generated.

8. An intelligent control system for the welding process of thin-walled metal components, characterized in that, A method for intelligent control of the welding process of a thin-walled metal component as described in any one of claims 1-7 includes: The initial parameter determination and time calibration module is used to determine the initial welding parameter set of the thin-walled metal component to be welded and to synchronize the time to obtain the time synchronization calibration parameters. The multi-source information acquisition and alignment module, based on the initial welding parameter set, ignites the electric arc and begins circumferential welding, acquiring multi-source raw data with timestamps during the welding process; it then performs time alignment by combining time synchronization calibration parameters to obtain time-aligned multi-source synchronized data. The rapid extraction module for welding status parameters calculates current stability index, heat accumulation index and measured value of weld pool width based on time-aligned multi-source synchronous data, and obtains key deviation index through threshold comparison method. The intelligent parameter adjustment calculation module calculates the current adjustment, pulse parameter adjustment, and position compensation based on key deviation indicators; and uses a priority synthesis method to obtain the set of adjustment values. The welding parameter update and execution module is used to combine the adjustment set and the initial welding parameter set to obtain the current welding parameters; it performs a limited update on the current welding parameters, using a step-by-step transition method and high-priority fast instructions to obtain the updated current welding parameter set. The welding process loop control and termination judgment module records the welding process data log based on the updated current welding parameter set, judges the welding status, and continues the welding process until a welding completion confirmation signal is received if the welding has not been terminated. The welding quality evaluation and data storage module, based on the welding completion confirmation signal, performs a comprehensive evaluation of the welding process and stores the data to generate a welding summary report.

Images