A method and system for testing the performance of a medium frequency direct current projection welder

Abstract

The application relates to the technical field of welding quality detection, and discloses a medium-frequency direct-current projection welding machine performance testing method and system, which comprises the following steps: synchronously acquiring an electrical original signal of a welding loop and a mechanical movement original signal of an electrode pressing mechanism of a medium-frequency direct-current projection welding machine in a welding power-on period, the electrical original signal comprising a current signal and a voltage signal, and the mechanical movement original signal comprising a displacement signal; based on the current signal, the voltage signal and the displacement signal, a feature parameter set representing a welding process state is extracted; the feature parameter set is compared and analyzed with a preset reference benchmark; according to the comparison and analysis result, the abnormal state of a welding spot is recognized; and a performance testing evaluation report is output. Through synchronous acquisition of the two types of core signals, namely, electrical signals and mechanical signals, and extraction of multidimensional features such as resistance, power consumption and forming time from the signals, online, lossless and comprehensive state imaging of each welding cycle is realized, and the problems of hysteresis and limitations of traditional sampling detection are solved.

Description

Technical Field

[0001] This application relates to the field of welding quality inspection technology, and more specifically, to a method and system for testing the performance of a medium-frequency DC projection welding machine. Background Technology

[0002] Medium-frequency DC projection welding machines are widely used in fields with stringent welding quality requirements, such as automobile manufacturing and power batteries, due to their stable welding process and precise control. In these high-end manufacturing scenarios, the quality of each weld point is directly related to the safety and performance of the final product. However, current methods for evaluating the performance of projection welding machines are significantly lagging and limited. The industry generally relies on offline sampling inspection, which involves randomly selecting welding samples from the production line and sending them to a laboratory for destructive testing, such as metallographic analysis or tensile testing. First, sampling inspection is essentially a probabilistic check and cannot cover all weld points in the production process. There is a high risk of missing defects caused by occasional fluctuations in equipment performance. Second, destructive testing processes are cumbersome and time-consuming. By the time the test results show problems, a large number of potentially defective products have often already been produced, resulting in huge economic losses. In addition, some on-site assessments rely on the experience of operators, judging by observing phenomena such as welding sparks. This method lacks unified quantitative standards, and the assessment results are highly subjective, making it difficult to accurately diagnose subtle degradation in equipment performance or mismatches in process parameters. This traditional, outdated quality control model can no longer meet the urgent needs of modern automated production lines for full-process, digital, and real-time monitoring of welding quality.

[0003] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a medium-frequency DC projection welding machine performance testing method and system, which can solve the technical problem in existing technologies that rely on offline sampling and subjective judgment for projection welding machine performance evaluation, resulting in detection lag and the inability to achieve real-time monitoring of the entire process.

[0005] In a first aspect, this application provides a method for testing the performance of a medium-frequency DC projection welding machine, including: During the welding energization cycle, the electrical raw signals of the welding circuit of the medium frequency DC projection welding machine and the mechanical motion raw signals of the electrode pressure mechanism are acquired synchronously. The electrical raw signals include current signals and voltage signals, and the mechanical motion raw signals include displacement signals. Based on current, voltage, and displacement signals, a set of characteristic parameters is extracted to characterize the state of the welding process. The set of characteristic parameters includes instantaneous resistance characteristics reflecting the evolution of the welding circuit impedance, power consumption characteristics reflecting the energy input, and forming time characteristics reflecting the crushing time of the bump. The set of characteristic parameters is compared and analyzed with a preset reference benchmark. Based on the comparison and analysis results, abnormal states of solder joints are identified, and a performance test evaluation report is output.

[0006] By simultaneously acquiring two core types of signals—electrical and mechanical—and extracting multi-dimensional features such as resistance, power consumption, and forming time, an online, non-destructive, and comprehensive status profile of each welding cycle is achieved. This fundamentally solves the problems of lag and limitations in traditional sampling inspection and provides an objective basis for real-time quality control.

[0007] Furthermore, the steps for extracting a set of characteristic parameters representing the welding process state based on current, voltage, and displacement signals include: Calculate the ratio of the voltage signal to the current signal as a characteristic of the instantaneous resistance; The product of the current signal and the voltage signal is integrated over time to obtain the total welding energy value, which serves as the power consumption characteristic. The displacement signal is differentiated to obtain the displacement acceleration signal; Identify peak and trough feature points in the displacement acceleration signal, and determine the time corresponding to the peak feature point as the point where the convex point begins to soften, and the time corresponding to the trough feature point as the point where the convex point completely collapses. The time interval between the point where the bump begins to soften and the point where the bump completely collapses is calculated as a molding time feature.

[0008] This technical solution provides a specific calculation path for transforming raw sensor signals into key performance indicators, and clarifies the quantification methods for three core characteristics: instantaneous resistance, power consumption, and molding time. In particular, by using secondary processing of the displacement signal to identify key time nodes in the convex point crushing process, the abstract concept of molding time can be accurately calculated, enhancing the operability of the technical solution and the reliability of the results.

[0009] Furthermore, after the step of calculating the ratio of the voltage signal to the current signal as a characteristic of instantaneous resistance, the process includes: During the welding energization cycle, the pressure signal of the electrode pressurization mechanism is collected; Calculate the first derivative of the pressure signal with respect to time to obtain the pressure change rate sequence; Calculate the first derivative of the instantaneous resistance characteristic with respect to time to obtain the resistance change rate sequence; A sliding window cross-correlation operation was performed on the pressure change rate series and the resistance change rate series to obtain the correlation index; Based on the correlation index, numerical gain compensation is applied to the instantaneous resistance characteristics to eliminate mechanical oscillation interference.

[0010] This technical solution addresses the problem that mechanical vibrations generated by electrode pressing during welding can interfere with resistance signals. It innovatively introduces a pressure signal as a reference and analyzes the correlation between pressure changes and resistance changes to accurately identify and compensate for resistance artifacts caused by mechanical factors. This results in a purer and more realistic instantaneous resistance curve that reflects the formation of the weld nugget, significantly improving the accuracy of resistance characteristics.

[0011] Furthermore, the steps for identifying peak and trough feature points in the displacement acceleration signal include: Read the pre-stored first mechanical noise sequence. The first mechanical noise sequence is the acceleration fluctuation data generated by the electrode pressurization mechanism itself, which is collected and stored before the welding energization cycle when the simulated pressing action is performed. During the welding energizing cycle, the displacement acceleration signal and the first mechanical noise sequence are differentially processed point by point to obtain the displacement acceleration component caused by the heat softening and crushing of the protrusion. The moment when the rate of decrease of the instantaneous resistance characteristic in the displacement acceleration component first exceeds the preset slope threshold value within a preset time window centered on the displacement acceleration peak is identified as the peak feature point. The moment when the fluctuation amplitude of the pressure signal enters a preset stable range within a preset time window centered on the displacement acceleration trough in the displacement acceleration component is identified as the trough feature point.

[0012] By pre-recording and subtracting the mechanical noise of the equipment itself, the subtle acceleration changes caused by welding itself are effectively separated. At the same time, by combining the instantaneous resistance change rate and pressure stability as auxiliary judgment criteria, the misjudgment that may occur based solely on the peaks and troughs of the acceleration signal is avoided, which greatly improves the accuracy of identifying feature points during the forming time.

[0013] Furthermore, the steps for identifying abnormal states of solder joints by comparing and analyzing the set of characteristic parameters with a preset reference standard and based on the comparison and analysis results include: If the deviation between the instantaneous resistance characteristic and the preset reference resistance characteristic exceeds the first preset threshold, the abnormal state is identified as abnormal electrode contact or abnormal melt nugget formation. If the power consumption characteristic exceeds the preset benchmark energy value, the abnormal state is identified as an abnormal energy input. If the molding time characteristic exceeds the preset benchmark time period, the abnormal state is identified as abnormal bump molding quality.

[0014] This technical solution establishes a mapping relationship from quantitative characteristics to specific fault types, making test results no longer simply pass or fail, but providing diagnostically valuable information. Deviations in resistance, power consumption, and molding time are correlated with physical anomalies such as electrode contact, energy input, and molding quality, providing clear guidance for operators to quickly pinpoint the root cause of problems and adjust process parameters.

[0015] Furthermore, prior to the step of extracting a set of characteristic parameters representing the welding process state based on current signals, voltage signals, and displacement signals, the following steps are included: The instantaneous heat generation is calculated based on the current and voltage signals; Based on the instantaneous heat generation and the preset material thermal expansion coefficient, the expansion correction value reflecting the upward expansion height of the material under heat is estimated in real time; The displacement signal is corrected based on the expansion correction value to obtain the net crushing displacement value, which reflects the actual downward collapse of the workpiece material under pressure.

[0016] By establishing a thermal expansion model and correcting the displacement signal in real time, the interference of thermal expansion was successfully eliminated, and the net crushing displacement value that purely reflects the collapse process of the molten core was obtained, providing more accurate basic data for subsequent displacement-based analysis.

[0017] Furthermore, the step of comparing and analyzing the set of characteristic parameters with a preset reference standard, and identifying the abnormal state of the solder joint based on the comparison and analysis results, also includes: Monitor the direction of the rate of change of net crush displacement; When the rate of change of the net crushing displacement value is detected to change from increasing to decreasing, a softening critical trigger signal is generated, which reflects that the workpiece material has overcome the thermal expansion resistance and officially entered the substantial crushing stage. Record the time difference between the start of welding energization and the moment when the softening critical trigger signal is generated, as the material softening response time; If the material softening response time exceeds the preset benchmark softening response time range, the abnormal state is identified as an abnormal material thermal response.

[0018] By capturing the inflection point where the net crush displacement changes from expansion-dominant to collapse-dominant, the response speed of the welded material to heat input can be directly reflected. This allows for the effective identification of abnormal thermal response caused by material composition, surface condition, or coating issues, thus increasing the dimensions of fault diagnosis.

[0019] Furthermore, the step of comparing and analyzing the set of characteristic parameters with a preset reference standard, and identifying the abnormal state of the solder joint based on the comparison and analysis results, also includes: When a positive pulse exceeding the second preset threshold appears in the displacement acceleration signal, and the instantaneous resistance characteristic decreases beyond the third preset threshold within a preset time period, the abnormal state is identified as welding spatter abnormality.

[0020] This technical solution provides a direct and reliable method for online identification of welding spatter defects. When spatter occurs, the ejection of molten metal causes the electrode to rebound instantaneously, generating a positive acceleration pulse and a sudden drop in welding circuit resistance. By simultaneously monitoring these two phenomena, the spatter event can be accurately captured, overcoming the shortcomings of traditional methods in detecting spatter in real time.

[0021] Furthermore, the steps for generating a performance test evaluation report include: When an abnormal condition is identified, the solder joint is marked as a defective solder joint; When a preset number of solder joints are marked as unqualified, an alarm signal is output and a stop command is sent. Record the set of characteristic parameters for each solder joint and the identification results of abnormal states, generate and output a performance test evaluation report.

[0022] This technical solution establishes a complete closed-loop quality control and response mechanism. It can not only identify individual defective solder joints but also, by setting continuous defect alarm thresholds, automatically warn and shut down equipment in case of serious malfunctions or systematic deviations in the process, effectively preventing batch quality incidents. Furthermore, the detailed reports provide valuable data archives for subsequent quality traceability and process optimization.

[0023] Secondly, this application also discloses a medium-frequency DC projection welding machine performance testing system for performing the medium-frequency DC projection welding machine performance testing method described in any of the preceding claims. The system includes: The signal acquisition module synchronously acquires the original electrical signals of the welding circuit of the medium-frequency DC projection welding machine and the original mechanical motion signals of the electrode pressure mechanism during the welding energization cycle. The original electrical signals include current signals and voltage signals, and the original mechanical motion signals include displacement signals. The feature extraction module extracts a set of feature parameters characterizing the state of the welding process based on current signals, voltage signals, and displacement signals. The set of feature parameters includes instantaneous resistance characteristics reflecting the evolution of the welding circuit impedance, power consumption characteristics reflecting the energy input, and forming time characteristics reflecting the crushing time of the bump. The performance evaluation module compares and analyzes the set of characteristic parameters with a preset reference benchmark. Based on the comparison and analysis results, it identifies abnormal states of solder joints and outputs a performance test evaluation report.

[0024] This application achieves online, real-time monitoring of the welding process at each weld point by simultaneously acquiring electrical signals from the welding circuit and mechanical signals from the electrode pressurization mechanism during the welding energization cycle. This fundamentally overcomes the shortcomings of traditional offline sampling inspection methods, such as detection lag and inability to comprehensively cover all weld points, enabling immediate detection of welding quality issues. Based on the acquired raw electrical and mechanical signals, this application extracts a multi-dimensional set of characteristic parameters, including instantaneous resistance, power consumption, and forming time characteristics. This transforms the complex physicochemical changes during welding into a set of quantifiable objective data for evaluating the welding status, replacing subjective judgments based on operator experience and significantly improving the accuracy and consistency of performance evaluation. Furthermore, this application not only determines whether a weld point is qualified but also identifies specific abnormal states such as abnormal electrode contact, abnormal energy input, and abnormal bump forming quality through analysis of different characteristic parameters. This provides precise data support for quickly diagnosing the root cause of faults and optimizing the welding process, achieving in-depth performance diagnosis. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a performance testing method for a medium-frequency DC projection welding machine provided in an embodiment of this application.

[0026] Figure 2 This is a schematic diagram of the structure of a medium-frequency DC projection welding machine performance testing system provided in an embodiment of this application.

[0027] Labeling explanation: 210, signal acquisition module; 220, feature extraction module; 230, performance evaluation module. Detailed Implementation

[0028] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0029] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] In precision industrial settings such as modern automobile body manufacturing, projection welding of high-strength steel plates is a critical process, and the quality of the welds directly determines the structural safety of the vehicle body. However, on high-speed automated production lines, after thousands of consecutive cycles, the electrodes of medium-frequency DC projection welding machines may experience minor wear or surface oxidation, or there may be subtle fluctuations in the power supply system. This can lead to insufficient weld nugget size or internal defects in dozens or even hundreds of subsequent welds. Traditional offline sampling inspection methods, due to their lag, cannot detect this gradual quality decline in a timely manner. By the time problems are exposed through destructive testing, a large number of potentially defective products have often already been produced, resulting in significant safety hazards and economic losses.

[0031] To solve this technical problem, firstly, see [reference needed] Figure 1 This application provides a method for testing the performance of a medium-frequency DC projection welding machine, the method comprising: S1. During the welding energizing cycle, the electrical raw signals of the welding circuit of the medium frequency DC projection welding machine and the mechanical motion raw signals of the electrode pressure mechanism are acquired synchronously. The electrical raw signals include current signals and voltage signals, and the mechanical motion raw signals include displacement signals. S2. Based on current signal, voltage signal and displacement signal, extract a set of characteristic parameters that characterize the state of the welding process. The set of characteristic parameters includes instantaneous resistance characteristics that reflect the evolution of the welding circuit impedance, power consumption characteristics that reflect the energy input, and forming time characteristics that reflect the crushing time of the bump. S3. Compare and analyze the set of characteristic parameters with the preset reference benchmark. Based on the comparison and analysis results, identify the abnormal state of the solder joint and output a performance test evaluation report.

[0032] Among them, the instantaneous resistance characteristic refers to the total resistance of the welding circuit at various moments during the entire welding energization period. It comprehensively reflects the contact resistance between the electrode and the workpiece, the volume resistance of the workpiece itself, and the resistance changes of the material in the weld area during the solid-state, plastic-state, liquid-state, and solid-state phase transitions. It is one of the most core electrical characteristics of the welding physical process.

[0033] Power consumption characteristics typically refer to the total electrical energy input into the welding circuit during a complete welding energization cycle.

[0034] Molding time characteristics refer to the time it takes from the moment the protrusions on the workpiece begin to soften under heat and undergo plastic deformation until they are completely crushed and form a stable melt nugget. It reflects the material's response speed to heat input and the speed of the plastic flow process.

[0035] In one specific embodiment, the above method is applied to an automated welding production line for automotive side panel assemblies. The robotic welding workstation on the production line is equipped with a medium-frequency DC projection welding machine for welding two or three layers of high-strength galvanized steel sheets. In order to achieve real-time quality monitoring of each weld point, a performance testing system is added to the projection welding machine.

[0036] Before performing the welding task, when the welding robot clamps the workpiece with the welding clamp and issues the welding start command, the command signal is simultaneously sent as a synchronous trigger signal to the data acquisition unit of the performance testing system. The data acquisition unit can be a high-speed multi-channel data acquisition card.

[0037] During the signal acquisition phase, the data acquisition card simultaneously begins acquiring analog signals from multiple channels at a sampling frequency of, for example, 50 kHz per second. Specifically, electrical signals are acquired by measuring the welding current through a high-precision Rogowski coil or Hall effect current sensor connected in series in the secondary output circuit of the welding transformer, while voltage probes are directly connected to the upper and lower electrode arms near the electrode caps to measure the voltage drop between the electrodes. Mechanical motion signals are acquired by installing a high-precision linear displacement sensor, such as a wire-type displacement sensor or a laser displacement sensor, on the movable electrode arm of the welding clamp to measure the vertical displacement of the electrode in real time.

[0038] During the feature extraction stage, the data acquisition card transmits the acquired raw data stream to an industrial control computer or embedded processor in real time. The analysis software running inside the processor first preprocesses this data, such as performing digital filtering to remove high-frequency noise. Then, according to a preset algorithm, it calculates the aforementioned set of feature parameters from this data. For example, the instantaneous resistance feature is obtained by dividing the voltage sample value at each moment by the current sample value at the same moment; the power consumption feature is obtained by accumulating the instantaneous power (the product of voltage and current at each moment) and then multiplying it by the sampling time interval to obtain the total welding energy; while the forming time feature needs to be determined by analyzing the displacement signal.

[0039] During the comparison analysis and report output phase, the newly calculated set of characteristic parameters is compared with a reference benchmark stored in the local database. This benchmark was established during the process debugging phase by welding a large number of qualified samples and conducting destructive testing, extracting the average value and allowable fluctuation range of its characteristic parameters; it can be called the gold welding curve or standard process fingerprint. The comparison analysis process checks whether the various characteristic parameters of the current solder joint fall within the preset acceptable range. If all parameters are within the range, the solder joint is deemed qualified. If any parameter exceeds the range, it is deemed abnormal, and the cause of the abnormality is preliminarily determined based on the type of parameter exceeding the range and the magnitude of the deviation. Finally, regardless of whether the solder joint is qualified or not, its complete set of characteristic parameters, original signal waveform, comparison results, and timestamps are recorded to form a log, which is ultimately compiled into a detailed performance test evaluation report.

[0040] In this way, the original quality control model, which relied on manual sampling and post-event analysis, has been replaced by an online, comprehensive, and automated real-time evaluation of every solder joint. Production managers can intuitively see the quality status of each solder joint on a screen in the central control room. Once a series of abnormal alarms occur, production can be immediately suspended for investigation, thereby preventing large-scale quality accidents.

[0041] Furthermore, the steps for extracting a set of characteristic parameters representing the welding process state based on current, voltage, and displacement signals include: Calculate the ratio of the voltage signal to the current signal as a characteristic of the instantaneous resistance; The product of the current signal and the voltage signal is integrated over time to obtain the total welding energy value, which serves as the power consumption characteristic. The displacement signal is differentiated to obtain the displacement acceleration signal; Identify peak and trough feature points in the displacement acceleration signal, and determine the time corresponding to the peak feature point as the point where the convex point begins to soften, and the time corresponding to the trough feature point as the point where the convex point completely collapses. The time interval between the point where the bump begins to soften and the point where the bump completely collapses is calculated as a molding time feature.

[0042] Specifically, the calculation of instantaneous resistance is a point-by-point application of Ohm's law. That is, at each sampling point of the time series, the operation R(t) = V(t) / I(t) is performed, where R is the instantaneous resistance, V is the voltage signal, I is the current signal, and t is time, thus obtaining a complete curve showing the resistance changing with time. The calculation of power consumption characteristics, in discrete digital signal processing, is represented by the summation of the instantaneous power sequence, i.e., E_total = Σ[V(i)*I(i)]*Δt, where i represents the sampling point number, Δt is the sampling time interval, and E_total is the instantaneous power sequence.

[0043] The underlying physical logic for extracting the forming time characteristics is as follows: when the bump begins to soften under heat, its downward collapse speed increases under constant electrode pressure. This is represented by a positive peak in the second derivative of the displacement signal, i.e., the acceleration signal. As the melt nucleus forms and expands, the collapse process reaches its fastest point. Subsequently, due to the support of the liquid metal and the expansion of the deformation area, the collapse speed begins to slow down, which is represented by a negative trough in the acceleration signal, marking the end of the main crushing process. Therefore, after obtaining the acceleration signal by numerically differentiating the original displacement signal, a simple peak detection algorithm can be used, such as finding the local maximum and minimum values ​​in the signal sequence, to locate these two key time points. The difference between these two time points is the forming time. Although this method is intuitive, in practical applications, factors such as the inertia of the robotic arm movement and slight vibrations of the workpiece may introduce many interfering spikes or spurious peaks into the acceleration signal, making the positioning based solely on peak and trough values ​​unstable and inaccurate.

[0044] Furthermore, after the step of calculating the ratio of the voltage signal to the current signal as a characteristic of instantaneous resistance, the process includes: During the welding energization cycle, the pressure signal of the electrode pressurization mechanism is collected; Calculate the first derivative of the pressure signal with respect to time to obtain the pressure change rate sequence; Calculate the first derivative of the instantaneous resistance characteristic with respect to time to obtain the resistance change rate sequence; A sliding window cross-correlation operation was performed on the pressure change rate series and the resistance change rate series to obtain the correlation index; Based on the correlation index, numerical gain compensation is applied to the instantaneous resistance characteristics to eliminate mechanical oscillation interference.

[0045] Fluctuations in electrode pressure during welding, especially mechanical vibrations caused by rapid movements of servo motors or cylinders, directly alter the contact state between the electrode and the workpiece, leading to fluctuations in contact resistance. This resistance change caused by mechanical factors, combined with resistance changes caused by material melting, can interfere with accurate assessment of the weld nugget formation process. To address this, this embodiment introduces a fourth synchronously acquired signal: the electrode pressure signal. This can be achieved by installing a high-frequency response force sensor, such as a piezoelectric force sensor, between the electrode head and the electrode arm.

[0046] During data processing, the analysis software first calculates the first derivatives of the pressure signal and the original instantaneous resistance signal, i.e., their respective rates of change. Then, a sliding window cross-correlation method is employed. Specifically, the software extracts a short segment of pressure rate of change data, such as 5 milliseconds, and the resistance rate of change data within the same time period, calculating the cross-correlation coefficient between these two short sequences. This coefficient ranges from -1 to 1, and its absolute value reflects the degree of similarity between the two signals. This calculation window slides along the entire time axis, resulting in a correlation index curve that varies over time.

[0047] When the correlation index is very high at a certain moment, such as an absolute value greater than 0.8, it means that at that moment, the change in resistance is highly synchronized with the change in pressure, which is very likely caused by mechanical oscillation. After identifying this situation, the instantaneous resistance characteristics at that moment are compensated. A simple compensation method is to apply an inverse, proportional attenuation to the instantaneous resistance characteristics at that point based on the magnitude of the correlation, or to use the instantaneous resistance characteristics from moments before and after it for smoothing filtering, thereby weakening the influence of this mechanical interference. After such processing, the resulting instantaneous resistance characteristic curve can more purely reflect the impedance evolution caused by the welding thermophysical process itself, providing a more reliable data foundation for subsequent analysis.

[0048] Furthermore, the method includes: Read the pre-stored first mechanical noise sequence. The first mechanical noise sequence is the acceleration fluctuation data generated by the electrode pressurization mechanism itself, which is collected and stored before the welding energization cycle when the simulated pressing action is performed. During the welding energizing cycle, the displacement acceleration signal and the first mechanical noise sequence are differentially processed point by point to obtain the displacement acceleration component caused by the heat softening and crushing of the protrusion. The moment when the rate of decrease of the instantaneous resistance characteristic in the displacement acceleration component first exceeds the preset slope threshold value within a preset time window centered on the displacement acceleration peak is identified as the peak feature point. The moment when the fluctuation amplitude of the pressure signal enters a preset stable range within a preset time window centered on the displacement acceleration trough in the displacement acceleration component is identified as the trough feature point.

[0049] Before formal production begins, or after equipment maintenance, the welding machine is controlled to perform several no-load pressurization actions, i.e., pressurization without power. During this process, displacement signals are collected and acceleration is calculated. This acceleration curve represents the inherent vibration mode of the welding machine's mechanical system during movement, i.e., the first mechanical noise sequence.

[0050] During actual welding, the displacement acceleration signal calculated in real time is subtracted point by point from the pre-stored noise sequence, which theoretically eliminates the influence of the equipment's own vibration. The remaining acceleration component can more accurately reflect the displacement changes caused by the softening and collapse of the workpiece protrusions due to heat.

[0051] Even so, to further improve the accuracy of key point identification, this solution also introduces a multi-signal joint criterion. When searching for the point where the convex begins to soften, instead of simply taking the highest peak of the post-cleaning acceleration curve, the solution examines the changes in the instantaneous resistance characteristic curve within a small time window around this peak, such as 2 milliseconds before and after the peak. Because physically, the softening of the convex is inevitably accompanied by an increase in contact area and a decrease in resistance, the precise moment when the slope of the resistance curve first exceeds a preset threshold within this time window is finally confirmed as the point where the convex begins to soften.

[0052] Similarly, when searching for the point where the convex point completely collapses, the lowest point of the acceleration curve is no longer simply chosen. Instead, the stability of the pressure signal is checked within a time window near this trough. This is because once the convex point completely collapses and the melt nugget forms, the entire mechanical system reaches a new state of force equilibrium, and the violent pressure fluctuations cease. Therefore, the moment when the fluctuation amplitude of the pressure signal first enters a preset stable range within this time window—for example, when the fluctuation is less than 1% of the set pressure—is ultimately identified as the point where the convex point completely collapses. This cross-validation method, combining electrical, mechanical, and force signal characteristics, can significantly improve the accuracy and robustness of identifying the start and end points of the molding time.

[0053] Furthermore, the steps for identifying abnormal states of solder joints by comparing and analyzing the set of characteristic parameters with a preset reference standard and based on the comparison and analysis results include: If the deviation between the instantaneous resistance characteristic and the preset reference resistance characteristic exceeds the first preset threshold, the abnormal state is identified as abnormal electrode contact or abnormal melt nugget formation. If the power consumption characteristic exceeds the preset benchmark energy value, the abnormal state is identified as an abnormal energy input. If the molding time characteristic exceeds the preset benchmark time period, the abnormal state is identified as abnormal bump molding quality.

[0054] For example, when comparing instantaneous resistance characteristics, it's not just about comparing a single value, but the entire curve's shape. The resistance curve is divided into several key stages, such as the preheating period, melting period, and stabilization period, and the average value, peak value, and slope of each stage are calculated and compared with the corresponding parameters of the reference curve. If the resistance in the initial stage is much higher than the reference value, this usually indicates oxides or dirt on the electrode surface, leading to poor contact and triggering an alarm for abnormal electrode contact. If the resistance decreases too quickly or too slowly during the melting period, it may be related to the growth rate of the melt nugget, triggering an alarm for abnormal melt nugget formation.

[0055] For power consumption characteristics, an energy window is typically set. For example, for a certain process, the standard energy is 5 kJ, so the energy window might be set to 4.8 to 5.2 kJ. If the actual energy is 5.5 kJ, exceeding the upper limit, an alarm will sound indicating abnormal energy input, and the possible causes will be excessive welding current or time settings, or abnormal fluctuations in the mains voltage.

[0056] For the molding time characteristic, there is also a time window. For example, the standard molding time is 60 milliseconds, and the time window may be set to 55 to 65 milliseconds. If the actual time is 80 milliseconds, it indicates that the bump softening and collapse is too slow, triggering an alarm for abnormal bump molding quality. Possible causes include insufficient welding current or an excessively thick plating layer on the workpiece surface affecting heat concentration. In this way, not only can the quality of the weld be determined, but also possible causes can be provided, greatly facilitating on-site engineers to quickly troubleshoot problems.

[0057] Furthermore, prior to the step of extracting a set of characteristic parameters representing the welding process state based on current signals, voltage signals, and displacement signals, the following steps are included: The instantaneous heat generation is calculated based on the current and voltage signals; Based on the instantaneous heat generation and the preset material thermal expansion coefficient, the expansion correction value reflecting the upward expansion height of the material under heat is estimated in real time; The displacement signal is corrected based on the expansion correction value to obtain the net crushing displacement value, which reflects the actual downward collapse of the workpiece material under pressure.

[0058] In a specific implementation example, the correction process is as follows: First, based on the real-time collected current I(t) and voltage V(t), the instantaneous heating power P(t) = I(t) × V(t) is calculated. Since the temperature rise of the material is proportional to the total heat absorbed, and the thermal expansion is also proportional to the temperature rise, a simplified model can be established, assuming that the total thermal expansion height is approximately proportional to the total energy input to the solder joint.

[0059] Based on the material and thickness of the workpiece to be welded, such as DP980 high-strength steel, the coefficient of thermal expansion K_exp corresponding to a unit energy input is obtained through experimental calibration or finite element simulation. This coefficient integrates multiple factors such as the material's coefficient of thermal expansion, specific heat capacity, density, and heat dissipation.

[0060] During the welding process, the instantaneous power is integrated over time in real time to obtain the cumulative input energy E(t) = ∫P(τ)dτ. Then, by E(t) × K_exp, the upward expansion height caused by thermal expansion from the start of energization to time t is estimated in real time, i.e., the expansion correction value ΔL_exp(t).

[0061] Finally, the raw displacement value ΔL_raw(t) measured by the displacement sensor, assuming downward is positive, is subtracted from the calculated expansion correction value to obtain the net crushing displacement value ΔL_net(t) = ΔL_raw(t) - (-ΔL_exp(t)) = ΔL_raw(t) + ΔL_exp(t). This corrected net crushing displacement value, free from the interference of thermal expansion, more realistically and clearly shows the complete process of plastic collapse of the workpiece under electrode pressure, providing a high-quality data foundation for subsequent more refined feature extraction and state analysis.

[0062] Based on the purer net crushing displacement value obtained above, new characteristic parameters can be further extracted to diagnose anomalies related to the material's inherent properties. Therefore, the step of comparing and analyzing the set of characteristic parameters with a preset reference standard, and identifying the abnormal state of the weld joint based on the comparison and analysis results, may further include: Monitor the direction of the rate of change of net crush displacement; When the rate of change of the net crushing displacement value is detected to change from increasing to decreasing, a softening critical trigger signal is generated, which reflects that the workpiece material has overcome the thermal expansion resistance and officially entered the substantial crushing stage. Record the time difference between the start of welding energization and the moment when the softening critical trigger signal is generated, as the material softening response time; If the material softening response time exceeds the preset benchmark softening response time range, the abnormal state is identified as an abnormal material thermal response.

[0063] In the initial stage of energization, thermal expansion dominates, and even after correction, the downward velocity of net crushing displacement is slow and may even plateau briefly. As heat continues to be input, once the material reaches its softening temperature, plastic deformation accelerates dramatically, at which point the downward velocity of net crushing displacement reaches a peak. This turning point from slow sinking to accelerated sinking marks the material's entry into the effective plastic forming stage.

[0064] In practical implementation, the first derivative of the net crushing displacement value ΔL_net(t) is taken to obtain the net crushing velocity v_net(t). Then, the change of v_net(t) is continuously monitored. When v_net(t) reaches its maximum value, that is, when the rate of change of v_net(t) (i.e., acceleration) turns from positive to negative, a softening critical trigger signal is generated. The time interval from the start of welding energization (t=0) to the generation of this trigger signal is defined as the material softening response time.

[0065] This time parameter directly reflects the material's sensitivity to heat input. For example, the softening response time of steel plates from the same batch should be very stable. If a batch of steel plates experiences changes in its thermal conductivity or endothermic properties due to compositional deviations or abnormal surface coatings, its softening response time will deviate from the standard range. If the softening response time of a weld joint is detected to be 25 milliseconds, while the standard range is 15 ± 3 milliseconds, an alarm for abnormal material thermal response will be output, prompting operators to check for problems with the incoming material quality.

[0066] As a preferred embodiment, the step of comparing and analyzing the set of feature parameters with a preset reference benchmark, and identifying the abnormal state of the solder joint based on the comparison and analysis results, may further include: When a positive pulse exceeding the second preset threshold appears in the displacement acceleration signal, and the instantaneous resistance characteristic decreases beyond the third preset threshold within a preset time period, the abnormal state is identified as welding spatter abnormality.

[0067] Welding spatter is the phenomenon of molten liquid metal being ejected at high speed from the weld pool. This process physically produces two significant, almost simultaneous signal characteristics. First, there is a violent mechanical rebound. After a portion of the liquid metal is ejected, the material supporting the electrode suddenly disappears, causing the electrode to rebound slightly under pressure. This rapid rebound is reflected in the displacement acceleration signal as an instantaneous, violent positive pulse. Second, there is an electrical path change. The ejection of liquid metal alters the geometry and composition of the welding circuit, typically leading to a momentary drop in circuit resistance or violent oscillations.

[0068] In practical implementation, the second preset threshold can be set to 5, far exceeding the mechanical vibration during normal welding. The third preset threshold can be set to the decrease in resistance within a very short time, such as 1 millisecond, exceeding 15% of the current value. Throughout the welding process, both the acceleration signal and the instantaneous resistance signal are monitored simultaneously. Once a positive spike exceeding 5 is detected in the acceleration signal, and within a time window of no more than 0.5 milliseconds from the spike, the instantaneous resistance also experiences a sudden drop exceeding 15%, it is highly certain that welding spatter has occurred, and the weld point is immediately marked as welding spatter abnormal. This logic of joint judgment based on dual conditions can greatly improve the accuracy of spatter identification and effectively avoid false alarms caused by simple electrical interference or mechanical impact.

[0069] Furthermore, the steps for generating a performance test evaluation report include: When an abnormal condition is identified, the solder joint is marked as a defective solder joint; When a preset number of solder joints are marked as unqualified, an alarm signal is output and a stop command is sent. Record the set of characteristic parameters for each solder joint and the identification results of abnormal states, generate and output a performance test evaluation report.

[0070] Specifically, each solder joint's data record contains a quality status field. Once any anomaly, such as energy anomaly or spatter anomaly, is identified, this field will be marked as unqualified.

[0071] Meanwhile, the system maintains a continuous non-conforming counter. For example, the preset number can be set to 3. When the first non-conforming solder joint appears, the counter becomes 1; if the next solder joint is qualified, the counter is reset to zero; if the next solder joint is still non-conforming, the counter becomes 2; when the counter reaches 3, it indicates that a systemic problem may have occurred, such as severe electrode wear, cooling water interruption, or workpiece batch issues. At this time, a chain reaction is immediately triggered: on the one hand, an audible and visual alarm is triggered, such as lighting up the red alarm light on the workstation and emitting a buzzer to notify the on-site personnel; on the other hand, a predefined pause production command is sent to the main PLC or robot controller of the production line via industrial Ethernet, so that the production line automatically stops after completing the current cycle and waits for manual intervention.

[0072] At the same time, regardless of whether the solder joint is qualified, all process data, including but not limited to the original waveforms of current, voltage, displacement, and pressure, the calculated instantaneous resistance, power consumption, forming time, softening response time and other characteristic parameters, as well as the final diagnostic results, will be structured and stored in a database or local file. This data constitutes a complete performance test evaluation report, providing a traceable digital archive of the quality of each product.

[0073] Secondly, see Figure 2 This application also provides a medium-frequency DC projection welding machine performance testing system, which is used to perform the medium-frequency DC projection welding machine performance testing method described in any of the preceding claims. The system includes: The signal acquisition module 210 is used to simultaneously acquire the electrical raw signals of the welding circuit of the medium frequency DC projection welding machine and the mechanical motion raw signals of the electrode pressure mechanism during the welding energizing cycle. The electrical raw signals include current signals and voltage signals, and the mechanical motion raw signals include displacement signals. The feature extraction module 220 is used to extract a set of feature parameters characterizing the state of the welding process based on current signal, voltage signal and displacement signal. The set of feature parameters includes instantaneous resistance characteristics reflecting the evolution of the welding circuit impedance, power consumption characteristics reflecting the energy input and forming time characteristics reflecting the crushing time of the bump. The performance evaluation module 230 is used to compare and analyze the set of characteristic parameters with a preset reference benchmark, identify abnormal states of solder joints based on the comparison and analysis results, and output a performance test evaluation report. The instantaneous resistance characteristic compensation module, used after calculating the ratio of the voltage signal to the current signal as the instantaneous resistance characteristic, includes: During the welding energization cycle, the pressure signal of the electrode pressurization mechanism is collected; Calculate the first derivative of the pressure signal with respect to time to obtain the pressure change rate sequence; Calculate the first derivative of the instantaneous resistance characteristic with respect to time to obtain the resistance change rate sequence; A sliding window cross-correlation operation was performed on the pressure change rate series and the resistance change rate series to obtain the correlation index; Based on the correlation index, numerical gain compensation is performed on the instantaneous resistance characteristics to eliminate mechanical oscillation interference; The displacement signal correction module, used before extracting the set of characteristic parameters representing the welding process state based on current, voltage, and displacement signals, includes: The instantaneous heat generation is calculated based on the current and voltage signals; Based on the instantaneous heat generation and the preset material thermal expansion coefficient, the expansion correction value reflecting the upward expansion height of the material under heat is estimated in real time; The displacement signal is corrected based on the expansion correction value to obtain the net crushing displacement value, which reflects the actual downward collapse of the workpiece material under pressure.

[0074] This technical solution provides a physical device capable of executing the above method, which functionalizes and modularizes each step in the method, providing a clear hardware architecture for the industrial application and equipment integration of this technology, and has strong engineering practical significance.

[0075] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for testing the performance of a medium-frequency DC projection welding machine, characterized in that, The method includes: During the welding energization cycle, the electrical raw signals of the welding circuit of the medium-frequency DC projection welding machine and the mechanical motion raw signals of the electrode pressure mechanism are acquired synchronously. The electrical raw signals include current signals and voltage signals, and the mechanical motion raw signals include displacement signals. Based on the current signal, the voltage signal, and the displacement signal, a set of characteristic parameters characterizing the welding process state is extracted. The set of characteristic parameters includes instantaneous resistance characteristics reflecting the evolution of the welding circuit impedance, power consumption characteristics reflecting the energy input, and forming time characteristics reflecting the crushing time of the bump. The set of characteristic parameters is compared and analyzed with a preset reference benchmark. Based on the comparison and analysis results, abnormal states of solder joints are identified, and a performance test evaluation report is output. The step of extracting a set of feature parameters characterizing the welding process state based on the current signal, the voltage signal, and the displacement signal includes: The ratio of the voltage signal to the current signal is calculated and used as the instantaneous resistance characteristic. The product of the current signal and the voltage signal is integrated over time to obtain the total welding energy value, which is used as the power consumption characteristic. The displacement signal is differentiated to obtain the displacement acceleration signal; Identify the peak and trough feature points in the displacement acceleration signal, and determine the time corresponding to the peak feature point as the point where the convex point begins to soften, and determine the time corresponding to the trough feature point as the point where the convex point completely collapses. The time interval between the point where the protrusion begins to soften and the point where the protrusion completely collapses is calculated as the molding time feature. After the step of calculating the ratio of the voltage signal to the current signal as a characteristic of the instantaneous resistance, the method includes: During the welding energizing cycle, the pressure signal of the electrode pressurizing mechanism is collected; Calculate the first derivative of the pressure signal with respect to time to obtain the pressure change rate sequence; Calculate the first derivative of the instantaneous resistance characteristic with respect to time to obtain the resistance change rate sequence; A sliding window cross-correlation operation is performed on the pressure change rate sequence and the resistance change rate sequence to obtain a correlation index. Based on the correlation index, numerical gain compensation is applied to the instantaneous resistance characteristics to eliminate mechanical oscillation interference.

2. The performance testing method for a medium-frequency DC projection welding machine according to claim 1, characterized in that, The step of identifying peak and trough feature points in the displacement acceleration signal includes: Read the pre-stored first mechanical noise sequence, which is the acceleration fluctuation data generated by the electrode pressurization mechanism itself, collected and stored before the welding energizing cycle when a simulated pressing action is performed; During the welding energizing cycle, the displacement acceleration signal and the first mechanical noise sequence are subjected to point-by-point differential operation to obtain the displacement acceleration component caused by the heat softening and crushing of the protrusion; The moment when the rate of decrease of the instantaneous resistance feature first exceeds a preset slope threshold value within a preset time window centered on the displacement acceleration peak in the displacement acceleration component is identified as the peak feature point. The moment when the fluctuation amplitude of the pressure signal enters a preset stable range within a preset time window centered on the displacement acceleration trough in the displacement acceleration component is identified as the trough feature point.

3. The performance testing method for a medium-frequency DC projection welding machine according to claim 1, characterized in that, The step of comparing and analyzing the set of feature parameters with a preset reference benchmark, and identifying the abnormal state of the solder joint based on the comparison and analysis results, includes: If the deviation between the instantaneous resistance characteristic and the preset reference resistance characteristic exceeds a first preset threshold, the abnormal state is identified as an abnormal electrode contact or an abnormal melt nugget formation. If the power consumption characteristic exceeds a preset benchmark energy value, the abnormal state is identified as an abnormal energy input. If the molding time characteristic exceeds the preset benchmark time period, the abnormal state is identified as an abnormality in the convex point molding quality.

4. The performance testing method for a medium-frequency DC projection welding machine according to claim 3, characterized in that, Before the step of extracting a set of feature parameters characterizing the welding process state based on the current signal, the voltage signal, and the displacement signal, the method includes: The instantaneous heat generation is calculated based on the current signal and the voltage signal; Based on the instantaneous heat generation and the preset material thermal expansion coefficient, the expansion correction value reflecting the upward expansion height of the material under heat is estimated in real time; The displacement signal is corrected based on the expansion correction value to obtain the net crushing displacement value, which reflects the actual downward collapse of the workpiece material under pressure.

5. The performance testing method for a medium-frequency DC projection welding machine according to claim 4, characterized in that, The step of comparing and analyzing the set of feature parameters with a preset reference benchmark, and identifying the abnormal state of the solder joint based on the comparison and analysis results, further includes: Monitor the direction of the rate of change of the net crush displacement value; When the rate of change of the net crush displacement value is detected to change from increasing to decreasing, a softening critical trigger signal is generated, reflecting that the workpiece material has overcome the thermal expansion resistance and officially entered the substantial crushing stage. Record the time difference between the start of welding energization and the moment when the softening critical trigger signal is generated, as the material softening response time; If the material softening response time exceeds the preset benchmark softening response time range, the abnormal state is identified as an abnormal material thermal response.

6. The performance testing method for a medium-frequency DC projection welding machine according to claim 3, characterized in that, The step of comparing and analyzing the set of feature parameters with a preset reference benchmark, and identifying the abnormal state of the solder joint based on the comparison and analysis results, further includes: When the displacement acceleration signal produces a positive pulse exceeding the second preset threshold, and the instantaneous resistance characteristic decreases by more than the third preset threshold within a preset time period, the abnormal state is identified as welding spatter abnormality.

7. The performance testing method for a medium-frequency DC projection welding machine according to claim 1, characterized in that, The steps for generating the performance test evaluation report include: When the abnormal state is identified, the solder joint is marked as a defective solder joint; When a preset number of solder joints are marked as defective solder joints, an alarm signal is output and a stop command is sent. Record the set of characteristic parameters for each solder joint and the identification results of the abnormal state, generate the performance test evaluation report and output it.

8. A medium-frequency DC projection welding machine performance testing system, used to perform the medium-frequency DC projection welding machine performance testing method as described in any one of claims 1 to 7, characterized in that, The system includes: The signal acquisition module is used to simultaneously acquire the original electrical signals of the welding circuit of the medium-frequency DC projection welding machine and the original mechanical motion signals of the electrode pressure mechanism during the welding energizing cycle. The original electrical signals include current signals and voltage signals, and the original mechanical motion signals include displacement signals. The feature extraction module is used to extract a set of feature parameters characterizing the state of the welding process based on the current signal, the voltage signal and the displacement signal. The set of feature parameters includes instantaneous resistance characteristics reflecting the impedance evolution process of the welding circuit, power consumption characteristics reflecting the energy input and forming time characteristics reflecting the crushing time of the bump. The performance evaluation module is used to compare and analyze the set of feature parameters with a preset reference benchmark, identify abnormal states of solder joints based on the comparison and analysis results, and output a performance test evaluation report.

Images