Laser welding method based on self-adaptive pulse waveform control and intelligent welding system

Abstract

The invention relates to the technical field of intelligent laser welding, and discloses a laser welding method and an intelligent welding system based on self-adaptive pulse waveform control, which are characterized in that a detection return light ratio is quantized as a difficulty index, and unit length energy consistency is constrained to a power generation link by utilizing unit length energy and an energy error; risk suppression is triggered when coupling fluctuation occurs through the coupling stability and the risk sign, so that the power waveform not only changes along with the speed, but also can make prospective and constraining response to the coupling state change, and by executing slope constraint and acceleration consistency correction on the set waveform queue, the accuracy of the power waveform is improved. Control data can be stably executed by the laser and is kept consistent with the motion state of the robot, so that the occurrence probability of corner overburning and incomplete penetration of an acceleration section is reduced, and the welding stability and consistency of a high-reflection material are improved; the problems that robot walking welding comprises rapid acceleration and deceleration and corner transition, and the welding quality is unstable under the material high-reflection and high-heat-conduction scene are solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent laser welding technology, and in particular to a laser welding method and intelligent welding system based on adaptive pulse waveform control. Background Technology

[0002] Laser welding offers advantages such as a small heat-affected zone, high energy density, and ease of automation in thin-plate lamination, precision device packaging, and the joining of high thermal conductivity materials. However, the laser welding process is not a simple steady-state energy input process. Especially under pulsed welding conditions, the welding quality is often determined by the uniformity of energy distribution along the path and the stability of the optical-material coupling state.

[0003] In actual working conditions, the combination of the rising edge, peak duration, duty cycle, and base power of the power waveform directly affects the formation of the molten pool, keyhole maintenance, and the probability of spatter generation. This means that the same average power may correspond to drastically different weld formations under different waveforms. For example, when an industrial robot performs thin-plate lap welding on highly reflective and thermally conductive materials (such as copper, aluminum, or coated materials), the trajectory typically includes acceleration, deceleration, and corner transition sections. The robot's end effector speed changes rapidly in a short period, causing the input energy per unit length to fluctuate with the speed. At the same time, highly reflective materials exhibit a significant low absorptivity and strong backlighting in the initial stage of the arc initiation and temperature rise phases, followed by a sudden change in absorptivity due to surface melting and keyhole formation, resulting in drastic changes in backlighting characteristics within a short window. Traditional methods are mostly based on constant pulse parameters or simply closed-loop adjustment of average power. These methods often implicitly assume that the speed is stable and the coupling is stable. Therefore, two typical failures are likely to occur when decelerating or accelerating again at corners: one is that the energy per unit length is too high during the deceleration phase, resulting in overheating, collapse or spatter; the other is that the energy per unit length is insufficient during the acceleration phase, resulting in incomplete penetration, cold solder joints or intermittent weld breaks.

[0004] On the other hand, the coupling uncertainty during the arc initiation stage makes it difficult for traditional control strategies to balance arc initiation reliability and subsequent stable welding. In the presence of highly reflective materials, surface oxide films, or oil contamination, the arc initiation stage results in strong backlighting and weak energy coupling. Directly increasing the peak power may improve the arc initiation probability, but it easily leads to localized transient overheating and spatter. Maintaining a conservative power level may result in delayed arc initiation, intermittent melting, or failure to establish a keyhole. Traditional methods typically lack a quantitative characterization of the difficulty of arc initiation coupling and a mechanism to transfer this characterization to the waveform generation stage. This makes welding parameters more reliant on experience or pre-weld trials, making it difficult to maintain consistent performance on the same production line when facing differences in material batches, surface conditions, and motion states.

[0005] Furthermore, waveform control typically issues power setpoints on the controller side using discrete sampling. When the target power changes too rapidly between adjacent sampling points, the laser's internal power supply and modulation link may experience response lag or overshoot. When the robot is accelerating, if the power setpoint queue is not consistent with the speed change, even if the average energy is sufficient, energy accumulation or energy gaps may occur near corners. Existing methods generally lack slope constraints and acceleration consistency checks for the set waveform queue, making it difficult to guarantee the executability and consistency of control commands under special speed conditions.

[0006] In summary, existing laser pulse welding methods often suffer from problems such as unstable arc initiation coupling, difficulty in maintaining consistent energy per unit length, difficulty in timely identification of coupling fluctuations, and insufficient executability of power queues when facing robotic welding scenarios involving rapid acceleration and deceleration and corner transitions, as well as high reflectivity and thermal conductivity of materials. These problems lead to quality defects such as fluctuating weld formation, increased spatter, increased risk of porosity, or intermittent incomplete penetration. Summary of the Invention

[0007] This invention provides a laser welding method and intelligent welding system based on adaptive pulse waveform control, which at least solves the problem that traditional fixed waveform or average power-based adjustment methods are difficult to simultaneously ensure arc initiation reliability and energy consistency per unit length in scenarios where robot welding involves rapid acceleration and deceleration and corner transitions, and where materials have high reflectivity and high thermal conductivity, thus leading to unstable welding quality.

[0008] To achieve the above objectives, the present invention provides a laser welding method based on adaptive pulse waveform control, the method comprising the following steps: Key sequence data for characterizing energy input, coupling state, and motion state during the welding process are acquired to form discrete power sequences, discrete backlight sequences, and discrete velocity sequences. The power sequence and the backlight sequence are subjected to detection window backlight ratio processing to obtain a difficulty index that reflects the difficulty of arc initiation coupling; A sliding window analysis process is performed on the power sequence, the velocity sequence, and the backlight sequence to obtain energy per unit length, energy error, and risk indicator. Based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, generate the velocity feedforward power, compensation power, and risk suppression coefficient, and determine the target equivalent power; A pulse parameter set is generated based on the target equivalent power, the difficulty index, and the risk indicator. A set waveform queue is generated according to the pulse parameter set. After performing slope constraint and acceleration consistency correction, waveform control data is output to perform laser welding according to the waveform control data.

[0009] Optionally, key sequence data characterizing the energy input, coupling state, and motion state of the welding process are acquired to form discrete power sequences, discrete backlight sequences, and discrete velocity sequences, specifically including: Establish the correspondence between discrete indices and sampling times, and generate a sequence of sampling times; At each sampling moment, the power monitoring value is read and written into the power sequence, the backlight intensity value is read and written into the backlight sequence, and the motion speed value is read and written into the speed sequence to form a triplet; Triples corresponding to the same discrete index are aggregated and written into the same sequence buffer, and the sequence buffer is used as the input for the detection window backlight ratio processing and sliding window analysis processing.

[0010] Optionally, the power sequence and the return sequence are subjected to a detection window return ratio processing to obtain a difficulty index reflecting the difficulty of arc initiation coupling, specifically including: Determine the set of sampling points corresponding to the detection window, and read the power sequence and backlight sequence within the detection window from the sequence buffer; The backlight ratio is calculated for each sampling point within the detection window to form a backlight ratio sequence, and the mean and standard deviation of the backlight ratio sequence are calculated to obtain the mean and standard deviation of the backlight ratio. The mean of the detection backlight ratio and the standard deviation of the detection backlight ratio are normalized and weighted to obtain the difficulty index, and the difficulty index is written into the sequence buffer.

[0011] Optionally, a sliding window analysis is performed on the power sequence and the velocity sequence to obtain the energy per unit length and the energy error, specifically including: Set a sliding window and read the power and speed data within the sliding window from the sequence buffer; The power data within the sliding window is discretely accumulated to obtain the window energy, and the velocity data within the sliding window is discretely accumulated to obtain the window displacement. A ratio operation is performed between the window energy and the window displacement to obtain the energy per unit length, and a difference operation is performed between the target energy per unit length and the energy per unit length to obtain the energy error. The energy per unit length and the energy error are then written into the sequence buffer.

[0012] Optionally, a sliding window analysis is performed on the backlight sequence and the power sequence to obtain a risk indicator, specifically including: Read the backlight sequence and power sequence from the sequence buffer, and calculate the backlight ratio sequence based on the backlight sequence and the power sequence; The mean and standard deviation of the sliding window are calculated for the backlight ratio sequence to obtain the mean and standard deviation of the window backlight ratio. Based on the mean and standard deviation of the window backlight ratio, normalization and pruning are combined to obtain the coupling stability. The coupling stability is compared with the difficulty index using a threshold to generate a risk flag, and the coupling stability and the risk flag are written into the sequence buffer.

[0013] Optionally, based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, a velocity feedforward power, a compensation power, and a risk suppression coefficient are generated, specifically including: Read the velocity sequence from the sequence buffer, perform a multiplication operation between the velocity sequence and the target unit length energy to obtain the velocity feedforward power, and write the velocity feedforward power into the sequence buffer; The energy error is read from the sequence buffer, a proportional-integral summation operation is performed based on the energy error to obtain the compensation power, and the compensation power is subjected to a limiting process before being written into the sequence buffer. The risk flag is read from the sequence buffer, a hierarchical mapping operation is performed based on the risk flag to obtain the risk suppression coefficient, and then written into the sequence buffer.

[0014] Optionally, the target equivalent power is determined, specifically including: Read the sequence buffer speed feedforward power, compensation power, and risk suppression coefficient; The velocity feedforward power and the compensation power are summed, and the summation result is multiplied by the risk suppression coefficient to obtain the target equivalent power, which is then written into the sequence buffer.

[0015] Optionally, a pulse parameter set is generated based on the target equivalent power, the difficulty index, and the risk indicator, specifically including: Read the target equivalent power, difficulty index, and risk indicator from the sequence buffer; Based on the difficulty index, access the preset template parameter table to obtain the template boundary set, and perform constraint tightening processing on the template boundary set based on the risk flag to obtain the parameter boundary after risk constraint. Within the parameter boundaries after the risk constraints, an equivalent power matching operation is performed to determine the duty cycle, peak power, and base power. The duty cycle, peak power, and base power are then pruned to form a pulse parameter set, which is then written into the sequence buffer.

[0016] Optionally, a set waveform queue is generated based on the pulse parameter set, and waveform control data is output after performing slope constraint and acceleration consistency correction, so as to perform laser welding according to the waveform control data, specifically including: Read the pulse parameter set from the sequence buffer and generate a set waveform queue within the look-ahead window length; The adjacent sampling point differential processing is performed on the set waveform queue, and slope constraint pruning is performed based on the differential result to obtain a queue that meets the power change rate limit; Read the velocity sequence from the sequence buffer and perform velocity difference operation to obtain the velocity change rate, which is used to generate the acceleration status flag; When the acceleration state flag is true, the speed feedforward power is read, and the set waveform queue is aligned with the speed feedforward power to obtain a queue that has passed the acceleration consistency correction. The output waveform queue, after slope constraint and acceleration consistency correction, is used as waveform control data to perform laser welding according to the waveform control data.

[0017] Furthermore, to achieve the above objectives, the present invention also provides an intelligent welding system based on adaptive pulse waveform control, the system comprising: The acquisition module is used to acquire key sequence data that characterizes the energy input, coupling state, and motion state of the welding process, in order to form discrete power sequence, discrete backlight sequence, and discrete velocity sequence. The first processing module is used to perform detection window backlight ratio processing on the power sequence and the backlight sequence to obtain a difficulty index that reflects the difficulty of arc coupling. The second processing module is used to perform sliding window analysis on the power sequence, the velocity sequence and the backlight sequence to obtain energy per unit length, energy error and risk indicator; The determination module is used to generate velocity feedforward power, compensation power, and risk suppression coefficient based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, and to determine the target equivalent power; The execution module is used to generate a pulse parameter set based on the target equivalent power, the difficulty index and the risk indicator, generate a set waveform queue according to the pulse parameter set, and output waveform control data after performing slope constraint and acceleration consistency correction, so as to perform laser welding according to the waveform control data.

[0018] The beneficial effects of this invention are as follows: It proposes a laser welding method and intelligent welding system based on adaptive pulse waveform control. By using the detection backlight ratio to quantify the difficulty index, and constraining the energy consistency per unit length to the power generation stage using energy per unit length and energy error, it also triggers risk suppression when coupling fluctuations occur through coupling stability and risk indicators. This allows the power waveform to not only follow speed changes but also make a forward-looking and constrained response to changes in coupling state. By performing slope constraints and acceleration consistency correction on the set waveform queue, the control data can be stably executed by the laser and kept consistent with the robot's motion state, thereby reducing the probability of corner overburning and incomplete penetration during acceleration, and improving the stability and consistency of welding high-reflectivity materials. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0021] This invention provides a laser welding method based on adaptive pulse waveform control, referring to... Figure 1 , Figure 1 This is a schematic flowchart of a laser welding method based on adaptive pulse waveform control according to an embodiment of the present invention.

[0022] In this embodiment, a laser welding method based on adaptive pulse waveform control includes the following steps: S1: Acquire key sequence data to characterize the energy input, coupling state, and motion state of the welding process, in order to form discrete power sequence, discrete backlight sequence, and discrete velocity sequence.

[0023] Specifically, key sequence data for characterizing the energy input, coupling state, and motion state of the welding process are acquired to form discrete power sequences, discrete backlight sequences, and discrete velocity sequences. This includes: establishing a correspondence between discrete indices and sampling times, and generating a sampling time sequence; reading power monitoring values ​​and writing them into the power sequence, reading backlight intensity values ​​and writing them into the backlight sequence, and reading motion velocity values ​​and writing them into the velocity sequence at each sampling time to form triplet pairs; pooling triplet pairs corresponding to the same discrete index and writing them into the same sequence buffer, and using the sequence buffer as input for backlight ratio processing of the detection window and sliding window analysis processing.

[0024] In one embodiment, a discrete index is used to represent the sampling point number, and the sampling time satisfies: ; in, This represents the nth sampling time, where n represents the discrete index and Δt represents the sampling period. Subsequently, in each... Synchronous reading: power monitoring values ​​are written to the power sequence P[n], backlight intensity values ​​are written to the backlight sequence R[n], and motion speed is written to the speed sequence v[n]. The triple {P[n], R[n], v[n]} is written to the cache with the same index.

[0025] It should be noted that the embodiments of the present invention do not limit the specific hardware form of power monitoring and backlight acquisition: power monitoring can originate from the internal power feedback channel of the laser or from the equivalent sampling output of an external power meter; backlight intensity can originate from a photodetector in the coaxial optical path or from an equivalent sensing channel that can reflect changes in reflected light intensity. As long as the sampling results can form a power sequence P[n] and a backlight sequence R[n] in a time sequence and are aligned with the velocity sequence v[n] on the same time axis, they fall within the processing scope of the embodiments of the present invention.

[0026] S2: Perform detection window backlight ratio processing on the power sequence and the backlight sequence to obtain a difficulty index that reflects the difficulty of arc initiation coupling.

[0027] Specifically, the power sequence and the return sequence are processed to obtain a difficulty index reflecting the difficulty of arc initiation coupling. This includes: determining the set of sampling points corresponding to the detection window; reading the power sequence and the return sequence within the detection window from the sequence buffer; calculating the return ratio for each sampling point within the detection window to form a detection return ratio sequence; calculating the mean and standard deviation of the detection return ratio sequence to obtain the mean and standard deviation of the detection return ratio; performing normalization and weighted synthesis processing on the mean and standard deviation of the detection return ratio to obtain the difficulty index; and writing the difficulty index into the sequence buffer.

[0028] It should be noted that during the arc initiation stage of welding high-reflectivity materials, coupling absorption often exhibits a superposition of weak average absorption and large local fluctuations. If the main welding power is used for ignition directly, transient energy overshoot may occur when the coupling suddenly improves; if the ignition is too conservative, it may lead to arc initiation delay and superimpose with the acceleration phase, resulting in the risk of incomplete fusion. To address this, this embodiment of the invention introduces a short detection window before the actual welding process: by sending a low-energy detection pulse and sampling the power sequence P[n] and the backlight sequence R[n] within this window, the backlight ratio under unit power is calculated and its mean and fluctuation are extracted, thereby obtaining an arc initiation coupling difficulty index that can be used for grading.

[0029] In one embodiment, the detection backlight ratio is defined as follows: ; in, Let R[n] represent the backlight ratio at the nth sampling point, R[n] represent the backlight sequence, and P[n] represent the power sequence. To prevent positive numbers with a denominator of zero.

[0030] Then at the detection window Internal Perform statistical mean and standard deviation extraction, the expressions are as follows: ; ; in, This represents the average backlight ratio within the detection window, used to characterize the average reflectance level; It represents the standard deviation of the backlight ratio within the detection window and is used to characterize the amplitude of coupling fluctuations; This represents the number of sampling points in the detection window.

[0031] In practical applications, A higher value usually means stronger backlighting per unit power and weaker material absorption; A higher value usually indicates an unstable coupling state, such as uneven surface film, slight changes in contact gap, or reflection jitter before the initial molten pool is established. Based on the above two types of information, this embodiment of the invention normalizes and weights them to form a difficulty index. This is to facilitate template segmentation during subsequent generation of pulse waveform structures; difficulty level: The expression is: ; in, The arc coupling difficulty index; , This is a reference range parameter for backlight ratio; For fluctuation reference parameters; , These are weight parameters; This is a range clipping function used to restrict the normalization result to a preset range.

[0032] For example, in one operational scenario of copper lap welding, when the detection window is within... Significantly higher than the reference median and When the value is large, the result is... As the difficulty level approaches, subsequent ignition strategies with smaller peak rise steps, gentler rise edges, and lower duty cycles can be selected; conversely, when... and When both are at a low level, For easier difficulty levels, faster ignition ramp-up and higher effective duty cycle can be used to shorten the arc initiation time. It should be noted that the above-described grading strategy is merely an example; the embodiments of this invention do not limit the specific weights and threshold values, which can be determined through calibration based on material reflection characteristics, spot size, and weld thickness.

[0033] S3: Perform sliding window analysis on the power sequence, the velocity sequence, and the backlight sequence to obtain energy per unit length, energy error, and risk indicator.

[0034] Specifically, a sliding window analysis is performed on the power sequence and the velocity sequence to obtain the energy per unit length and the energy error. This includes: setting a sliding window; reading the power data and velocity data within the sliding window from the sequence buffer; performing discrete accumulation on the power data within the sliding window to obtain the window energy; performing discrete accumulation on the velocity data within the sliding window to obtain the window displacement; performing a ratio operation based on the window energy and the window displacement to obtain the energy per unit length; performing a difference operation based on the target energy per unit length and the energy per unit length to obtain the energy error; and writing the energy per unit length and the energy error into the sequence buffer.

[0035] It should be noted that, considering the scenario described in this application, one of the core contradictions of the integrated arc initiation and acceleration action lies in the rapid change of speed from low to high, which causes strong non-stationarity of heat input per unit length on the time axis. Simply focusing on power cannot reflect the heating situation along the weld. Therefore, this embodiment of the invention adopts a sliding window integration and displacement normalization method to directly construct the energy per unit length. This construction method can unify power changes and speed changes into the same evaluation quantity.

[0036] In one embodiment, the sliding window length is set to... Discrete energy accumulation is performed on the power within the window, expressed as: ; Discrete displacement accumulation is performed on the velocity within the window, as expressed by: ; in, Indicates window energy. Indicates window displacement. , The power and velocity at the i-th discrete sampling point are respectively given, where Δt is the sampling period. Then, the energy per unit length is calculated, expressed as: ; Where H[n] is the energy per unit length, To prevent positive numbers with a denominator of zero.

[0037] Furthermore, the energy error is formed based on the target energy per unit length Ht, and its expression is: ; Where Ht is the energy per unit length of the target target. This represents the energy error.

[0038] In practical applications, when the robot is just starting up and its speed v[n] is very low, if the power is kept at a stable level, then very small and The relative size is too large, causing H[n] to be too large and leading to A negative value indicates a risk of energy overshoot per unit length; this occurs when speed enters a rapid ascent but power fails to keep pace. Increase The relative deficiency causes H[n] to be too small and leads to A positive value indicates a risk of non-fusion. Therefore, As an important input for generating compensation power.

[0039] It should be noted that the window length It is not limited to a fixed value. In some implementations, The window value can be adaptively selected based on the speed level: for example, a shorter window is used in the low-speed phase to improve response speed, and a longer window is used in the stable phase to improve estimation smoothness; as long as subsequent steps are still based on... , Construct H[n] and form All of these fall within the scope of the technical concept of the embodiments of the present invention.

[0040] Specifically, a sliding window analysis is performed on the backlight sequence and the power sequence to obtain a risk indicator. This includes: reading the backlight sequence and the power sequence from the sequence buffer; calculating the backlight ratio sequence based on the backlight sequence and the power sequence; performing sliding window mean and sliding window standard deviation calculations on the backlight ratio sequence to obtain the window backlight ratio mean and window backlight ratio standard deviation; performing normalization and pruning combination processing based on the window backlight ratio mean and window backlight ratio standard deviation to obtain coupling stability; performing threshold comparison processing between the coupling stability and the difficulty index to generate a risk indicator; and writing the coupling stability and the risk indicator into the sequence buffer.

[0041] It should be noted that, in addition to energy per unit length, highly reflective materials also suffer from coupling absorption fluctuations during the arc initiation stage. To ensure that waveform control not only follows speed but also suppresses transient boiling caused by abrupt coupling changes, this embodiment of the invention also statistically analyzes the backlight ratio within a sliding window, constructs coupling stability, and then generates a risk indicator by combining it with a difficulty index.

[0042] First, the backlight ratio sequence is calculated using the backlight sequence and the power sequence. Then, the mean and standard deviation are calculated within a sliding window, with the following expressions: ; ; in, This represents the average backlight ratio within the window, used to characterize the average level of reflection. This indicates the fluctuation in backlight ratio within the window and is used to characterize coupling stability.

[0043] In one embodiment, the coupling stability can be constructed as follows: ; in, For coupling stability, , This is the reference threshold parameter for backlight ratio. As a reference parameter for fluctuations, , For weighting parameters. The meaning of this construction is: when the average backlight is high or the fluctuation is large, the value within the parentheses increases, making... This reduces the instability of coupling, thus expressing it in a more intuitive way.

[0044] Subsequently, the risk indicator F[n] is generated through threshold comparison: for example, when Below the threshold And the difficulty index Above the threshold When F[n] is high risk, mark it as high risk; when only Below The time frame is marked as medium risk; all other situations are marked as normal. It should be noted that this classification method can trigger different structural protection actions during waveform generation, such as limiting the peak rise rate, reducing duty cycle, and inserting a window, thereby reducing overheating and boiling caused by coupling abrupt changes.

[0045] It should be emphasized that the risk rating rules can be adjusted according to the characteristics of materials and equipment, and are not limited to a three-tier structure; as long as it does not violate the principles of... and Under the premise of forming a risk classification and providing it for subsequent waveform control, adopting a two-level or multi-level classification can be regarded as an equivalent substitution of the embodiments of the present invention.

[0046] S4: Based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, generate the velocity feedforward power, compensation power, and risk suppression coefficient, and determine the target equivalent power.

[0047] In this embodiment of the invention, the energy conservation requirement and coupling risk constraint are unified into an equivalent power target that can be directly used for waveform matching. First, velocity feedforward is used to synchronize power and velocity changes to maintain the energy reference per unit length. Second, proportional-integral accumulation is used to compensate for energy errors to correct deviations caused by coupling changes, assembly fluctuations, etc. Finally, a risk suppression coefficient is used to conservatively compress the overall target, so that the system prioritizes overshoot suppression during high-risk phases.

[0048] Specifically, generating velocity feedforward power, compensation power, and risk suppression coefficient based on the target unit length energy, the velocity sequence, the energy error, and the risk flag includes: reading the velocity sequence from the sequence buffer; performing a multiplication operation between the target unit length energy and the velocity sequence to obtain the velocity feedforward power; and writing the velocity feedforward power into the sequence buffer; reading the energy error from the sequence buffer; performing a proportional-integral summation operation based on the energy error to obtain the compensation power; performing a limiting operation on the compensation power; and writing the compensation power into the sequence buffer; and reading the risk flag from the sequence buffer; performing a hierarchical mapping operation based on the risk flag to obtain the risk suppression coefficient; and writing the risk flag into the sequence buffer.

[0049] In one embodiment, the velocity feedforward power is defined as follows: ; in, For speed feedforward power, Energy per unit length of target length It is a velocity sequence.

[0050] It should be noted that in practical applications, the energy per unit length under the continuous limit is... Approximately proportional to v, this ensures that P varies proportionally with v, maintaining consistent energy input along the acceleration phase. Therefore, this feedforward approach can address the problem of sudden energy shortage per unit length caused by a sudden increase in speed, and can also automatically reduce power during low-speed starts to avoid energy buildup.

[0051] For further revision The deviation under actual coupled fluctuations introduces energy error. The proportional-integral summation method is used to calculate the compensation power, and the expression is: ; in, To compensate for power, This is the proportionality coefficient. The integral coefficient is... The value represents the energy error, and Δt represents the sampling period.

[0052] In this embodiment of the invention, the proportional term is used to quickly respond to instantaneous deviations, and the integral term is used to eliminate persistent deviations, thereby making the energy per unit length closer to the target in the stable segment. It should be noted that the compensation power can be further limited to avoid waveform abrupt changes caused by excessive single compensation; the limiting threshold can be determined through calibration, and this embodiment of the invention does not impose any limitation on it.

[0053] Furthermore, the risk indicator F[n] is converted into a risk mitigation coefficient. In one embodiment, a hierarchical mapping method is adopted: a smaller coefficient is used when F[n] is high-risk, a medium coefficient is used when F[n] is medium-risk, and 1 is used when F[n] is normal. Furthermore, when the coupling is unstable, the overall equivalent power is preferentially reduced to avoid transient boiling caused by a sudden improvement in coupling.

[0054] Specifically, determining the target equivalent power includes: reading the velocity feedforward power, compensation power, and risk suppression coefficient from the sequence buffer; performing a summation operation on the velocity feedforward power and the compensation power, and multiplying the summation result with the risk suppression coefficient to obtain the target equivalent power, and writing the target equivalent power into the sequence buffer.

[0055] In one embodiment, the synthesized target equivalent power is expressed as follows: ; in, For target equivalent power, This is the risk mitigation coefficient. For speed feedforward power, To compensate for power.

[0056] In practical applications, the detection difficulty index is high in the early stages of arc initiation. High and coupling stability If the value frequently falls below the threshold, causing F[n] to repeatedly enter the high-risk state, then... Will make Relatively conservative; once the molten pool is established and the coupling stabilizes, F[n] returns to normal. The value is restored to 1, so that the equivalent power returns to the main path of speed feedforward and energy error compensation, thus balancing stability and efficiency.

[0057] S5: Generate a pulse parameter set based on the target equivalent power, the difficulty index, and the risk indicator; generate a set waveform queue based on the pulse parameter set; and output waveform control data after performing slope constraint and acceleration consistency correction, so as to perform laser welding according to the waveform control data.

[0058] In this embodiment of the invention, the equivalent power target is transcribed into an executable pulse structure and continuously updated within a short look-ahead window using a rolling queue. Considering that adjusting the average power alone is often insufficient to suppress transient risks, this structured output can directly influence the molten pool establishment and spatter suppression during the arc initiation stage of highly reactive materials by utilizing waveform shape parameters such as pulse rise edge, duty cycle, and peak step size.

[0059] Specifically, generating a pulse parameter set based on the target equivalent power, the difficulty index, and the risk flag includes: reading the target equivalent power, the difficulty index, and the risk flag from the sequence buffer; accessing a preset template parameter table based on the difficulty index to obtain a template boundary set, and performing constraint tightening processing on the template boundary set based on the risk flag to obtain a risk-constrained parameter boundary; performing equivalent power matching operations within the risk-constrained parameter boundary to determine the duty cycle, peak power, and base power; performing pruning processing on the duty cycle, peak power, and base power to form a pulse parameter set; and writing the pulse parameter set into the sequence buffer.

[0060] In one embodiment, the pulse parameter set is defined as: ; in, Peak power, Base value power, Duty cycle, This is the rising edge parameter.

[0061] Based on this, first read: target equivalent power Difficulty Index 1. Risk flag F[n], and perform template classification and risk constraints: (1) Based on the difficulty index Access the preset template parameter table to obtain the parameter boundary set for the ignition and transition phases (e.g., the range of d[n], the upper limit of the peak step size, etc.). (e.g., lower limit) (2) Based on F[n], perform constraint tightening on the parameter boundary set: reduce the allowable duty cycle limit, reduce the peak rise speed, and increase the rise time in the high-risk stage to weaken the transient energy impact; relax the constraints in the normal stage to improve efficiency.

[0062] Then, equivalent power matching is performed within the boundary, satisfying: ; in, Let d[n] be the target equivalent power, and d[n] be the duty cycle. Peak power, This is the base power value.

[0063] It should be noted that this equivalent power matching formula is used to map the average equivalent power target to a pulse peak-based structure. For example, in one implementation, it can be first fixed... A certain level within the allowable range is used to maintain the continuity of the molten pool, and then... Inverse solution Alternatively, d[n] can be used; the duty cycle can be fixed first, and then the peak value can be solved. It should be noted that the specific order of matching solution is not a restriction, as long as the equivalent power consistency is ultimately satisfied and a pulse parameter set is formed. All of these can be considered equivalent implementations of the embodiments of the present invention.

[0064] Specifically, a set waveform queue is generated based on the pulse parameter set. After performing slope constraint and acceleration consistency correction, waveform control data is output to perform laser welding based on the waveform control data. This includes: reading the pulse parameter set from the sequence buffer and generating a set waveform queue within the look-ahead window length; performing adjacent sampling point differential processing on the set waveform queue and performing slope constraint pruning based on the differential result to obtain a queue that meets the power change rate limit; reading the velocity sequence from the sequence buffer and performing velocity differential operation to obtain the velocity change rate, which is used to generate an acceleration status flag; when the acceleration status flag is true, reading the velocity feedforward power and performing queue correction processing to align the set waveform queue with the velocity feedforward power to obtain a queue that passes the acceleration consistency correction; and outputting the set waveform queue after slope constraint and acceleration consistency correction as waveform control data to perform laser welding based on the waveform control data.

[0065] Obtaining the pulse parameter set Then, a set waveform queue is generated within the look-ahead window length M. This is used to characterize the discrete setpoint sequence of power over a short period of time, addressing the issue of frequent control updates during arc initiation acceleration while avoiding output jitter. The specific generation method is based on the pulse parameter set. The pulse sequence is expanded within the window, and the peak segment and the base segment are arranged according to the duty cycle. The peak edge is discretized and expanded according to the rising edge parameter to obtain continuous discrete power setpoints.

[0066] To prevent uncontrolled waveform jumps during abrupt coupling changes, embodiments of the present invention generate a predefined waveform queue. Apply slope constraints. A slope constraint can be expressed as: ; in, To set the power setting value for the k-th sampling point in the waveform queue, The maximum permissible power change slope is given by Δt, where Δt is the sampling period. It should be noted that this constraint provides an upper limit to the power change, which can suppress instantaneous power surges when coupling suddenly improves, thereby reducing the risk of splashing and boiling over.

[0067] Meanwhile, to ensure that the energy per unit length does not drop precipitously during the acceleration phase, this embodiment of the invention also requires acceleration consistency correction: calculating the rate of change of velocity based on the velocity sequence v[n] and determining whether it is in an acceleration state; when in an acceleration state, reading the velocity feedforward power. and will In general trend towards Alignment helps avoid the problem of insufficient power output due to excessive risk suppression or compensation lag, which can lead to a speed increase that is not followed by a cooling effect.

[0068] For example, in an arc-starting acceleration instance, if the acceleration phase velocity increases from 0.2 m / s to 0.8 m / s within 100 ms, then This will increase significantly; if the coupling stability at this time... To reduce triggering risk, slope constraints will suppress excessively rapid power increases, while accelerated consistency correction will ensure that power continues to rise and gradually approaches feedforward requirements, thus creating an operable trade-off between preventing overshoot and preventing undershoot.

[0069] It should be noted that the embodiments of the present invention do not limit the specific value of the look-ahead window length M; M can take different sizes under different laser response characteristics and control cycle conditions. A shorter window is beneficial to improving responsiveness, while a longer window is beneficial to improving smoothness. Those skilled in the art can select the appropriate value based on the system bandwidth and welding speed range.

[0070] Reference Figure 2 , Figure 2This is a schematic diagram of the structure of an intelligent welding system based on adaptive pulse waveform control according to an embodiment of the present invention.

[0071] like Figure 2 As shown, the intelligent welding system based on adaptive pulse waveform control proposed in this embodiment of the invention includes: The acquisition module 10 is used to acquire key sequence data that characterizes the energy input, coupling state, and motion state of the welding process, so as to form discrete power sequence, discrete backlight sequence, and discrete velocity sequence. The first processing module 20 is used to perform detection window backlight ratio processing on the power sequence and the backlight sequence to obtain a difficulty index that reflects the difficulty of arc coupling. The second processing module 30 is used to perform sliding window analysis on the power sequence, the velocity sequence and the backlight sequence to obtain energy per unit length, energy error and risk indicator; The determination module 40 is used to generate velocity feedforward power, compensation power and risk suppression coefficient based on the target unit length energy, the velocity sequence, the energy error and the risk indicator, and to determine the target equivalent power; The execution module 50 is used to generate a pulse parameter set based on the target equivalent power, the difficulty index and the risk indicator, generate a set waveform queue according to the pulse parameter set, and output waveform control data after performing slope constraint and acceleration consistency correction, so as to perform laser welding according to the waveform control data.

[0072] Other embodiments or specific implementations of the intelligent welding system based on adaptive pulse waveform control of the present invention can be found in the above-described method embodiments, and will not be repeated here.

[0073] It is understood that in the description of this specification, references to terms such as "one embodiment," "another embodiment," "other embodiments," or "first embodiment to Nth embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0074] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0075] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A laser welding method based on adaptive pulse waveform control, characterized in that, The method includes the following steps: Key sequence data for characterizing energy input, coupling state, and motion state during the welding process are acquired to form discrete power sequence, discrete backlight sequence, and discrete velocity sequence. The power sequence and the backlight sequence are processed by the detection window backlight ratio to obtain a difficulty index that reflects the difficulty of arc initiation coupling; A sliding window analysis process is performed on the power sequence, the velocity sequence, and the backlight sequence to obtain energy per unit length, energy error, and risk indicator. Based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, generate the velocity feedforward power, compensation power, and risk suppression coefficient, and determine the target equivalent power; A pulse parameter set is generated based on the target equivalent power, the difficulty index, and the risk indicator. A set waveform queue is generated according to the pulse parameter set. After performing slope constraint and acceleration consistency correction, waveform control data is output to perform laser welding according to the waveform control data.

2. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, Key sequence data for characterizing energy input, coupling state, and motion state during the welding process are acquired to form discrete power sequences, discrete backlight sequences, and discrete velocity sequences, specifically including: Establish the correspondence between discrete indices and sampling times, and generate a sequence of sampling times; At each sampling moment, the power monitoring value is read and written into the power sequence, the backlight intensity value is read and written into the backlight sequence, and the motion speed value is read and written into the speed sequence to form a triplet; Triples corresponding to the same discrete index are aggregated and written into the same sequence buffer, and the sequence buffer is used as the input for the detection window backlight ratio processing and sliding window analysis processing.

3. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, The power sequence and the return sequence are subjected to a detection window return ratio processing to obtain a difficulty index reflecting the difficulty of arc initiation coupling, specifically including: Determine the set of sampling points corresponding to the detection window, and read the power sequence and backlight sequence within the detection window from the sequence buffer; The backlight ratio is calculated for each sampling point within the detection window to form a backlight ratio sequence, and the mean and standard deviation of the backlight ratio sequence are calculated to obtain the mean and standard deviation of the backlight ratio. The mean of the detection backlight ratio and the standard deviation of the detection backlight ratio are normalized and weighted to obtain the difficulty index, and the difficulty index is written into the sequence buffer.

4. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, Perform sliding window analysis on the power sequence and the velocity sequence to obtain the energy per unit length and the energy error, specifically including: Set a sliding window and read the power and speed data within the sliding window from the sequence buffer; The power data within the sliding window is discretely accumulated to obtain the window energy, and the velocity data within the sliding window is discretely accumulated to obtain the window displacement. A ratio operation is performed between the window energy and the window displacement to obtain the energy per unit length, and a difference operation is performed between the target energy per unit length and the energy per unit length to obtain the energy error. The energy per unit length and the energy error are then written into the sequence buffer.

5. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, Performing sliding window analysis on the backlight sequence and the power sequence to obtain risk indicators specifically includes: Read the backlight sequence and power sequence from the sequence buffer, and calculate the backlight ratio sequence based on the backlight sequence and the power sequence; The mean and standard deviation of the sliding window are calculated for the backlight ratio sequence to obtain the mean and standard deviation of the window backlight ratio. Based on the mean and standard deviation of the window backlight ratio, normalization and pruning are combined to obtain the coupling stability. The coupling stability is compared with the difficulty index using a threshold to generate a risk flag, and the coupling stability and the risk flag are written into the sequence buffer.

6. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, Based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, a velocity feedforward power, a compensation power, and a risk suppression coefficient are generated, specifically including: Read the velocity sequence from the sequence buffer, perform a multiplication operation between the velocity sequence and the target unit length energy to obtain the velocity feedforward power, and write the velocity feedforward power into the sequence buffer; The energy error is read from the sequence buffer, a proportional-integral summation operation is performed based on the energy error to obtain the compensation power, and the compensation power is subjected to a limiting process before being written into the sequence buffer. Read the risk flag from the sequence buffer, perform a hierarchical mapping operation based on the risk flag to obtain the risk suppression coefficient, and write the coefficient into the sequence buffer.

7. The laser welding method based on adaptive pulse waveform control as described in claim 6, characterized in that, Determining the target equivalent power specifically includes: Read the sequence buffer speed feedforward power, compensation power, and risk suppression coefficient; The velocity feedforward power and the compensation power are summed, and the summation result is multiplied by the risk suppression coefficient to obtain the target equivalent power, which is then written into the sequence buffer.

8. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, A pulse parameter set is generated based on the target equivalent power, the difficulty index, and the risk indicator, specifically including: Read the target equivalent power, difficulty index, and risk indicator from the sequence buffer; Based on the difficulty index, access the preset template parameter table to obtain the template boundary set, and perform constraint tightening processing on the template boundary set based on the risk flag to obtain the parameter boundary after risk constraint. Within the parameter boundaries after the risk constraints, an equivalent power matching operation is performed to determine the duty cycle, peak power, and base power. The duty cycle, peak power, and base power are then pruned to form a pulse parameter set, which is then written into the sequence buffer.

9. The laser welding method based on adaptive pulse waveform control as described in claim 1, characterized in that, A waveform queue is generated based on the pulse parameter set. After performing slope constraint and acceleration consistency correction, waveform control data is output to perform laser welding based on the waveform control data. Specifically, this includes: Read the pulse parameter set from the sequence buffer and generate a set waveform queue within the look-ahead window length; The adjacent sampling point differential processing is performed on the set waveform queue, and slope constraint pruning is performed based on the differential result to obtain a queue that meets the power change rate limit; Read the velocity sequence from the sequence buffer and perform velocity difference operation to obtain the velocity change rate, which is used to generate the acceleration status flag; When the acceleration state flag is true, the speed feedforward power is read, and the set waveform queue is aligned with the speed feedforward power to obtain a queue that has passed the acceleration consistency correction. The output waveform queue, after slope constraint and acceleration consistency correction, is used as waveform control data to perform laser welding according to the waveform control data.

10. An intelligent welding system based on adaptive pulse waveform control, characterized in that, The system includes: The acquisition module is used to acquire key sequence data that characterizes the energy input, coupling state, and motion state of the welding process, in order to form discrete power sequence, discrete backlight sequence, and discrete velocity sequence. The first processing module is used to perform detection window backlight ratio processing on the power sequence and the backlight sequence to obtain a difficulty index that reflects the difficulty of arc coupling. The second processing module is used to perform sliding window analysis on the power sequence, the velocity sequence and the backlight sequence to obtain energy per unit length, energy error and risk indicator; The determination module is used to generate velocity feedforward power, compensation power, and risk suppression coefficient based on the target energy per unit length, the velocity sequence, the energy error, and the risk indicator, and to determine the target equivalent power; The execution module is used to generate a pulse parameter set based on the target equivalent power, the difficulty index and the risk indicator, generate a set waveform queue according to the pulse parameter set, and output waveform control data after performing slope constraint and acceleration consistency correction, so as to perform laser welding according to the waveform control data.

Images