Spot welding machine electrode wear prediction method based on time series analysis

Abstract

The present application relates to the technical field of electric digital data processing, in particular to a spot welding machine electrode wear prediction method based on time series analysis, comprising: grouping the welding point response curve according to the end face life cycle and the welding condition, obtaining the welding point response curve sequence in the same end face life cycle; establishing a stable end face reference curve; determining the end face response deviation; determining the deviation retention; converting the welding window and the inter-weld window into the end face wear equivalent duration; and performing electrode life prediction. The present application converts the time axis of electrode wear prediction from natural time or cumulative welding point count into end face wear equivalent duration, compares the stable end face reference curve with the end face response deviation, judges the deviation retention of adjacent welding points, and converts the deviation retention of the welding window and the inter-weld window, so as to establish a corresponding relationship between the electrode life prediction and the real advancement degree of the electrode end face state, and reduce the prediction deviation caused by the production rhythm difference, repair welding and grinding reference change.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and specifically to a method for predicting electrode wear in spot welding machines based on time series analysis. Background Technology

[0002] In resistance spot welding, pressure is applied and welding current is passed through the electrode after it contacts the workpiece. A weld nugget forms in the contact area under thermal pressure. As the number of weld points increases, the electrode end face gradually exhibits end-face enlargement, mushrooming, contamination adhesion, localized pitting, or morphological resetting after grinding. These changes in end-face condition alter the contact area, current density, dynamic resistance, pressure-displacement response, and weld nugget stability. To assess electrode wear and predict remaining lifespan, existing technologies typically establish prediction mechanisms based on the cumulative number of weld points, cumulative production time, dynamic resistance curves, or time series models.

[0003] In existing technologies, the time axis used for electrode wear prediction is usually natural time or cumulative weld count. This time axis lacks a corresponding relationship with the continuous deviation process of the electrode end face state from a relatively stable reference. Welding program switching, changes in plate combination, fixture off-center loading, instantaneous current shunting, resetting of grinding reference, or single-point measurement noise can all cause deviations in the weld response curve. Existing technologies do not distinguish between these interfering factors and the actual wear of the end face, resulting in deviations in the prediction of the electrode's remaining life and maintenance timing. Summary of the Invention

[0004] This invention provides a method for predicting electrode wear in spot welding machines based on time series analysis to solve the above-mentioned problems.

[0005] The present invention provides a method for predicting electrode wear in spot welding machines based on time series analysis, which adopts the following technical solution: One embodiment of the present invention provides a method for predicting electrode wear of a spot welding machine based on time series analysis. The method includes: grouping the weld point response curves according to the end face life cycle and welding conditions to obtain a sequence of weld point response curves within the same end face life cycle; establishing a stable end face reference curve based on the weld point response curve sequence; comparing each weld point response curve in the weld point response curve sequence with the stable end face reference curve to determine the end face response deviation; determining the deviation retention amount based on the end face response deviation of adjacent weld points; converting the welding window and weld gap window into an equivalent end face wear duration based on the deviation retention amount; and predicting the electrode life based on the equivalent end face wear duration.

[0006] Further, the step of comparing each solder joint response curve in the solder joint response curve sequence with the stable end-face reference curve to determine the end-face response deviation includes: normalizing each solder joint response curve into the same phase sequence according to the welding stage, and comparing it phase by phase with the stable end-face reference curve to obtain a phase residual sequence; determining the phase point sequence in the phase residual sequence that continuously exceeds the measurement repeatability range determined based on the stable end-face reference curve and has the same deviation direction as the effective deviation segment of the end face; and determining the end-face response deviation based on the residual area and phase length of each effective deviation segment of the end face.

[0007] Further, determining the deviation retention amount based on the end face response deviation of adjacent solder joints includes: pairing the effective deviation segments of the end faces of adjacent solder joints according to the welding stage and phase interval; when the deviation directions of the paired deviation segments are consistent and the phase intervals overlap, determining the overlapping interval between the paired deviation segments as the deviation retention segment; and determining the deviation retention amount based on the overlapping phase length of each deviation retention segment and the residual area jointly retained by the adjacent solder joints within the overlapping phase length.

[0008] Furthermore, determining the deviation holding amount based on the end face response deviation of adjacent solder joints further includes: organizing records of grinding, cap replacement, program switching, material batch switching, fixture alarm, flow shunting abnormality, and cooling abnormality into an end face breakpoint record sequence; aligning the end face breakpoint record sequence with the adjacent solder joint window according to the timestamp; the adjacent solder joint window is determined from the end time of the previous solder joint to the start time of the next solder joint; if there are records in the end face breakpoint record sequence whose occurrence time falls into the adjacent solder joint window, or if the program number or material batch number of the adjacent solder joint changes, a breakpoint mark is formed; based on the breakpoint mark, the deviation connection between adjacent solder joints is cut off.

[0009] Further, the step of converting the welding window and weld gap window into the equivalent duration of end face wear based on the deviation holding amount includes: determining the welding window duration based on the power-on start record and power-on end record; determining the weld gap window duration based on the time interval between the end time of the previous weld point and the start time of the next weld point in adjacent weld points; converting the weld gap window duration based on the deviation holding amount of the corresponding adjacent weld point pair; adding the converted weld gap window duration to the welding window duration to obtain the equivalent duration of a single window; and accumulating the equivalent durations of each single window within the same end face life cycle in the order of the weld points to obtain the equivalent duration of end face wear.

[0010] Furthermore, the electrode life prediction based on the equivalent duration of end-face wear includes: determining the life boundary based on the cumulative equivalent duration of end-face wear before regrinding or cap replacement in the historical life cycle of similar electrode ends; comparing the cumulative equivalent duration of end-face wear in the current life cycle with the life boundary to determine the remaining life ratio; and outputting an electrode maintenance prompt based on the remaining life ratio.

[0011] Furthermore, the electrode life prediction also includes: in the event of at least one of program switching, plate batch switching, fixture alarm or cooling abnormality during the prediction period, forming a breakpoint marker based on the program switching, plate batch switching, fixture alarm or cooling abnormality, and truncating the continuous accumulation of the end face wear equivalent time based on the breakpoint marker.

[0012] Furthermore, the method of predicting electrode life based on the equivalent duration of end-face wear includes: using the equivalent duration of end-face wear as the time index of the time series prediction model; or, using the equivalent duration of a single window as the time step of each solder joint in the time series prediction model.

[0013] Furthermore, the step of establishing a stable end face reference curve based on the weld point response curve sequence includes: selecting weld points from the weld point response curve sequence that do not have grinding breakpoints, fixture alarms, or current shunting abnormalities under the same welding procedure and the same plate combination after grinding or cap replacement, and whose contact state is stable, to form a stable end face reference curve.

[0014] Furthermore, the step of grouping the weld point response curves according to the end face life cycle and welding conditions to obtain a sequence of weld point response curves within the same end face life cycle includes: dividing the end face life cycle according to electrode number, grinding and cap replacement records, welding program, and plate combination; collecting weld point response data, which includes voltage, current, pressure, displacement, welding start time, welding end time, program switching records, plate batch records, fixture alarm records, current shunting abnormality records, and cooling abnormality records; forming a dynamic resistance curve based on the voltage and current, and determining the boundaries of the closing, energizing, effective heating, holding, and falling stages based on the pressure or displacement; organizing the weld point response data into a weld point response curve sequence according to the order of weld point occurrence; forming adjacent weld point response pairs; and organizing grinding, cap replacement, program switching, plate batch switching, fixture alarm, cooling abnormality, and current shunting abnormality into an end face breakpoint record sequence.

[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, the weld joint response curves are grouped according to the end face life cycle and welding conditions to obtain a sequence of weld joint response curves within the same end face life cycle; a stable end face reference curve is established based on the weld joint response curve sequence; the response curves of each weld joint in the weld joint response curve sequence are compared with the stable end face reference curve to determine the end face response deviation; the deviation holding amount is determined based on the end face response deviation of adjacent weld joints; the welding window and weld gap window are converted into the equivalent end face wear duration based on the deviation holding amount; and electrode life is predicted based on the equivalent end face wear duration.

[0016] This invention transforms the time axis for electrode wear prediction from natural time or cumulative weld count to the equivalent duration of end-face wear. By comparing the deviation of the stable end-face reference curve with the end-face response, judging the deviation holding amount of adjacent weld points, and calculating the deviation holding amount of welding windows and weld interval windows, a correspondence is established between the electrode life prediction and the actual progression of the electrode end-face condition. This reduces prediction errors caused by differences in production cycle time, rework welding, and changes in grinding reference. Furthermore, by grouping weld point response data according to end-face lifecycle and welding conditions, weld points under different electrode numbers, grinding and cap replacement records, welding procedures, and plate combinations are aggregated into the same comparable sequence. This avoids the overlap of weld point response curves from different end-face lifecycles and heterogeneous conditions, reducing the impact of variations in end-face lifecycle and welding conditions. The probability of misjudging wear trends caused by working condition switching is reduced. On the other hand, based on the stable end face reference curve, the current weld point response curve is compared phase by phase. By determining the over-limit of the measurement repeatability range and connecting the continuous deviation segments, the real change of the end face contact state is separated from single-point measurement noise and stage switching boundary spikes, improving the characterization accuracy of the end face response deviation amount for wear advancement. On the other hand, by judging the direction consistency of the effective deviation segments of adjacent weld point end faces, the phase interval overlap, and the alignment of the end face breakpoint recording sequence, a deviation holding amount is formed. This distinguishes the continuous advancement characteristics of electrode end face wear from isolated anomalies caused by plate overlap disturbance, instantaneous current shunting, program switching, or grinding reference reset, suppressing the interference of non-wear factors on the life prediction time axis. Attached Figure Description

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

[0018] Figure 1 A flowchart illustrating the method for predicting electrode wear of a spot welding machine based on time series analysis provided in an embodiment of the present invention; Figure 2This is a flowchart illustrating the end-face response deviation determination scheme provided in an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following detailed description, in conjunction with the accompanying drawings and preferred embodiments, provides a detailed account of the specific implementation methods, structures, features, and effects of the method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details a specific scheme for a spot welding machine electrode wear prediction method based on time series analysis provided by the present invention.

[0022] like Figure 1 As shown, this embodiment of the invention provides a method for predicting electrode wear in spot welding machines based on time series analysis, including: Step S110: Group the weld joint response curves according to the end face life cycle and welding conditions to obtain the weld joint response curve sequence within the same end face life cycle.

[0023] Optionally, step S110 includes: dividing the end face lifecycle according to electrode number, grinding and cap replacement record, welding program, and plate assembly; collecting weld point response data, including voltage, current, pressure, displacement, welding start time, welding end time, program switching record, plate batch record, fixture alarm record, current shunting abnormality record, and cooling abnormality record; forming a dynamic resistance curve based on voltage and current, and determining the boundaries of the closing, energizing, effective heating, holding, and falling-back stages based on pressure or displacement; organizing the weld point response data into a weld point response curve sequence according to the order of weld point occurrence; forming adjacent weld points into adjacent weld point response pairs; and organizing grinding, cap replacement, program switching, plate batch switching, fixture alarm, cooling abnormality, and current shunting abnormality into an end face breakpoint record sequence.

[0024] After grinding or cap replacement, a new contact reference is formed on the electrode end face. The complete usage period from the reconstruction of this reference to the next grinding or cap replacement constitutes an end face life cycle. Within the same life cycle, the wear trend of the electrode end face is continuous and comparable. If weld response data from different life cycles or different operating conditions are directly spliced, the differences in reference before and after grinding, the changes in current density caused by switching welding programs, and the different heat dissipation conditions caused by differences in plate combinations will all cause the response curve to deviate from the end face wear. Therefore, it is necessary to group the weld response data according to electrode number, grinding and cap replacement records, welding programs, and plate combinations to ensure that the data within the same group has a comparable reference.

[0025] The weld joint response data includes timestamps collected during the welding process, such as voltage, current, pressure, displacement, welding start time, and welding end time, as well as program switching records, sheet batch records, fixture alarm records, current shunt anomaly records, and cooling anomaly records. Voltage and current data are used to calculate dynamic resistance; the dynamic resistance curve reflects the impedance change characteristics of the electrode-sheet contact area. Pressure or displacement data is used to identify the boundaries of the closing, energizing, effective heating, holding, and falling-back phases during welding, thus providing a basis for subsequent normalization of the curves into a unified phase sequence.

[0026] The collected weld point response data are organized into a weld point response curve sequence according to the time sequence of weld point occurrence, enabling subsequent analysis to track end-face state changes along the time axis. Adjacent weld points are grouped into adjacent weld point response pairs, providing a data pairing basis for determining whether end-face response deviations are maintained between consecutive weld points. Grinding, cap replacement, program switching, sheet metal batch switching, fixture alarms, cooling anomalies, and flow shunting anomalies are organized into an end-face breakpoint record sequence. This sequence is aligned with the adjacent weld point window by timestamp, used to truncate trend interruptions caused by non-wear factors in deviation maintenance judgment, avoiding interference from abnormal operating conditions on the continuity of wear trends.

[0027] In addition to using the initial stable solder joints after grinding or cap replacement within the current end-face's lifespan, the establishment of a stable end-face reference curve can also utilize the response curves of qualified solder joints under the same program and material combination in history, or use the first batch of stable solder joints in the calibration data of the new electrode cap as the reference source. Furthermore, besides using dynamic resistance curves, the main response curves from voltage drop curves, welding energy curves, or electrode displacement curves can also be used. These curves still need to be compared with the stable end-face reference at the same welding phase.

[0028] Step S120: Establish a stable end face reference curve based on the weld point response curve sequence.

[0029] Optionally, step S120 above includes: selecting from the weld point response curve sequence weld points that do not have grinding breakpoints, fixture alarms, or current shunting abnormalities under the same welding procedure and the same plate combination after grinding or cap replacement, and whose contact state is stable, to form a stable end face reference curve.

[0030] The stable end-face reference curve reflects the contact impedance characteristics of the electrode end face before wear progression occurs after reference reconstruction. This curve serves as the reference origin for subsequent end-face response deviation calculations, and its data purity directly determines the reliability of the wear trend judgment. In the solder joint response curve sequence, only those solder joint response data that can represent the initial state of the end face and are not contaminated by external disturbances are qualified to be included in the reference construction range.

[0031] The welding process and sheet metal assembly directly determine the electrical parameters, heat dissipation path, and current distribution environment. Dynamic resistance curves under different processes or sheet metal assemblies exhibit inherent differences. If weld points from heterogeneous operating conditions are included in the reference, the reference curve will carry information about these operating condition differences rather than purely about the initial state of the end face. Therefore, the selection of reference weld points is limited to the same welding process and the same sheet metal assembly, ensuring that all reference samples are under the same process constraints.

[0032] Grinding or cap replacement reconstructs the geometry and surface condition of the electrode end face. The solder joint response after this point best represents the starting state of the end face's life cycle. Simultaneously, grinding breakpoints, fixture alarms, and shunt anomaly records indicate the presence of reference damage or external disturbances within the corresponding time period. If such solder joint response curves are included in the reference set, they will solidify abnormal fluctuations as reference characteristics, causing deviations caused by subsequent normal wear to be masked or misjudged. The requirement for stable contact further eliminates response distortions caused by poor electrode-board contact or momentary loosening, ensuring that the reference curve reflects the true impedance level under conditions of full contact between the end face and the board.

[0033] In actual production line deployment, if the number of initial solder joints within the current end face's lifecycle is insufficient or the stability is poor, the response curves of qualified solder joints under the same program and material combination in the historical database can serve as an alternative benchmark source. Furthermore, the calibration data before the new electrode cap is put into use records the response characteristics of the electrode cap in its brand-new state. The first batch of stable solder joints in this calibration data can also serve as a data source for the benchmark curve, without being limited to on-site sampling after current grinding or cap replacement.

[0034] After establishing a stable end-face reference curve, the current solder joint response curve is normalized into the same phase sequence according to the stages of closure, energization, effective heating, holding, and fallback. This sequence is then compared phase-by-phase with the reference curve to generate a residual phase sequence. The phase distribution characteristics of the reference curve determine the statistical boundary of the measurement repeatability range. Whether the residual phases of each phase point of the current solder joint fall within this range becomes the primary basis for judging the effective deviation point of the end face.

[0035] Step S130: Compare the response curves of each solder joint in the solder joint response curve sequence with the stable end face reference curve to determine the end face response deviation.

[0036] It is understandable that electrode wear is not manifested as an instantaneous jump at an isolated sampling point, but rather as a continuous residual in the response curve relative to the stable end-face reference within the same welding stage. This continuous residual corresponds to a gradual change in the end-face contact area, contact resistance, or indentation rebound characteristics. However, current disturbances, measurement noise, or stage switching boundary effects can also cause a single sampling point to exceed the normal fluctuation range. If such isolated exceedances are directly regarded as wear signals, it is impossible to distinguish between normal process fluctuations and the actual changes in the end-face state, leading to the wear trend being easily drowned out by noise or misjudged. The end-face response deviation determined in step S130 above can solidify the continuous phase point sequence that exceeds the measurement repeatability range and has the same deviation direction into an effective end-face deviation segment by comparing the current weld point response curve with the stable end-face reference curve phase by phase. Based on the residual area and phase length of each deviation segment, the curve deviation is converted into a measurable numerical index, thereby filtering out the interference of single-point noise and boundary peaks, and quantifying the true gradual change in the end-face contact state. The following introduces an optional determination scheme for the end-face response deviation: like Figure 2 As shown, optionally, step S130 above includes: Step S131: Normalize the response curves of each weld point into the same phase sequence according to the welding stage, and compare them phase by phase with the reference curve of the stable end face to obtain the phase residual sequence; Step S132: The phase point sequence that continuously exceeds the measurement repeatability range determined based on the stable end face reference curve and has the same deviation direction in the phase residual sequence is determined as the effective deviation segment of the end face; Step S133: Determine the end face response deviation based on the residual area and phase length of the effective deviation segment of each end face.

[0037] The original dynamic resistance curves of different solder joints differ in sampling length and stage distribution along the time axis, making it difficult to align feature points when directly superimposed for comparison. By dividing the response curves of each solder joint into stages of closure, energization, effective heating, holding, and fallback, and mapping each stage to a unified normalized phase sequence, the influence of time scale differences between different solder joints on curve comparison can be eliminated, ensuring that the response characteristics within the same welding stage have the same phase coordinate reference.

[0038] The aforementioned phase residual can be defined as the difference between the dynamic resistance value of the current solder joint at a normalized phase point and the corresponding phase value of the stable end-face reference curve. This difference directly reflects the degree of deviation of the end-face contact state relative to the reference. The measurement repeatability range is obtained by statistically analyzing the phase residual fluctuations of several qualified solder joints under the same welding procedure and the same plate combination after grinding or cap replacement. Its boundary value is used to distinguish between normal process fluctuations and abnormal deviations. If the phase residual of the current solder joint falls within this range, it is considered as sampling repeatability or normal process fluctuation; if it exceeds this range, it needs to be further judged in conjunction with the deviation status of adjacent phase points.

[0039] The selection of effective deviation segments on the end face follows the principle of continuity. A single sampling point spike may originate from current disturbance or measurement noise. Only when multiple consecutive phase points exceed the measurement repeatability range and maintain the same deviation direction does it indicate that the end face contact state has undergone a continuous change within the corresponding welding stage. Isolated out-of-limit points (i.e., neither adjacent phase points exceed the limit, or adjacent points exceed the limit but in opposite directions) and boundary spikes that only occur at stage switching boundaries and do not continue to the next phase point are considered as instantaneous fluctuations caused by non-wear factors and are not included in subsequent calculations. The selection mechanism retains not individual noise points, but continuous deviation segments corresponding to changes in the end face contact state.

[0040] The deviation of the end face response is obtained by normalizing the residual density within the effective deviation segment, and its calculation formula is as follows:

[0041] In the formula, For the first The first end face lifecycle The deviation of the end face response of each solder joint, after normalization... It is a dimensionless parameter; This represents the effective deviation segment of the end face formed by the weld point. For the first The residual area of ​​the effective deviation segment of each end face; For the first The phase length of each effective offset segment on the end face represents the normalized phase range covered by the offset segment. To avoid extremely small positive numbers with a denominator of zero, Units and The units are the same; This represents the normalization function. The above end-face response deviation... The calculation formula only measures the residual density within the effective deviation segment. The screening of the original sampling points and the connection of continuous segments have been completed in the previous operation.

[0042] Through the above processing, the deviation of the end face response is no longer affected by isolated noise points or transient phase switching, but instead reflects the continuous changes in the end face contact state during the welding stages, such as initial contact, effective heating, or holding and falling back. The closed contact deviation segment typically reflects changes in the initial contact area or surface contamination of the end face, the effective heating deviation segment reflects changes in contact resistance and current density, and the holding and falling back deviation segments reflect changes in pressing and cooling springback. This staged continuous deviation identification mechanism provides a reliable single-point deviation measurement basis for judging the deviation holding between subsequent adjacent weld points.

[0043] Step S140: Determine the deviation holding amount based on the deviation of the end face response of adjacent solder joints.

[0044] It is understandable that electrode end face wear is continuous in actual welding cycles. If the previous weld point has shown an increase in contact area or a deterioration in contact condition, the subsequent weld point will often continue to show a response offset in a similar direction at the same welding stage. Conversely, if the offset stage of the subsequent weld point changes significantly or the offset direction reverses, it is more likely to be due to occasional disturbances caused by plate overlap gaps, local adhesion of plating, fixture off-center loading, or instantaneous current shunting, rather than the actual advancement of end face wear. The deviation of the end face response of a single weld point alone cannot distinguish between a continuous wear trend and an isolated anomaly. Therefore, this embodiment of the invention uses the deviation holding amount as an indicator to quantify the degree of deviation acceptance between adjacent weld points, thereby separating the continuous advancement characteristics of wear from ordinary curve anomalies, and providing an accumulative time axis input basis for subsequently converting the actual wear process into the equivalent end face wear duration. An optional scheme for determining the deviation holding amount is described below: Optionally, step S140 includes: pairing the effective deviation segments of the end faces of adjacent solder joints according to the welding stage and phase interval; when the deviation directions of the paired deviation segments are consistent and the phase intervals overlap, determining the overlapping interval between the paired deviation segments as the deviation retention segment; and determining the deviation retention amount based on the overlapping phase length of each deviation retention segment and the residual area jointly retained by adjacent solder joints within the overlapping phase length.

[0045] The physical mechanisms underlying electrode wear differ fundamentally across different welding stages. The response offset in the initial contact stage primarily reflects changes in the initial contact area and surface contamination state between the electrode and the plate. The offset in the effective heating stage corresponds to changes in contact resistance and current density, while the hold-and-fall stage is related to thermal expansion, indentation depth, and cooling rebound. If deviation segments from different welding stages are paired across stages, the resistance increase in the effective heating stage may be incorrectly appended to the displacement offset in the hold-and-fall stage, leading to curve changes driven by different physical mechanisms being mistakenly interpreted as a continuation of the same wear pattern. Therefore, deviation-and-fall judgment is limited to within the same welding stage; only deviation segments within the initial contact, effective heating, or hold-and-fall stages are included in the pairing process.

[0046] The establishment of a deviation holding segment requires the simultaneous fulfillment of three conditions: directional consistency, phase interval overlap, and deviation persistence. Directional consistency requires that the offset directions of the two consecutive solder joints on the paired deviation segment be the same; if the directions reverse, it indicates that the end-face condition may have undergone a transient change due to non-wear factors. Phase interval overlap requires that the two deviation segments intersect on the normalized phase axis, ensuring that wear occurs within the same process phase window. Deviation persistence requires that the subsequent solder joint still exceeds the measurement repeatability range within the overlap interval, meaning that the end-face condition change caused by the previous solder joint is not restored to the baseline level by process fluctuations in the subsequent solder joint. Only when all three conditions are met is the overlap interval solidified as a deviation holding segment, characterizing the true transmission of end-face wear trends between consecutive solder joints.

[0047] Optionally, step S140 further includes: organizing the records of grinding, cap replacement, program switching, sheet metal batch switching, fixture alarm, flow shunting abnormality, and cooling abnormality into an end face breakpoint record sequence; aligning the end face breakpoint record sequence with the adjacent solder joint window according to the timestamp; the adjacent solder joint window is determined from the end time of the previous solder joint to the start time of the next solder joint; if there are records in the end face breakpoint record sequence whose occurrence time falls into the adjacent solder joint window, or if the program number or sheet metal batch number of the adjacent solder joint changes, a breakpoint mark is formed; based on the breakpoint mark, the deviation between adjacent solder joints is cut off.

[0048] In actual welding processes, grinding and cap replacement operations reconstruct the geometric reference of the electrode end face; program switching and batch changes in sheet metal alter the energizing parameters and heat dissipation paths; and fixture alarms and shunt anomalies introduce external mechanical or electrical disturbances. The differences in weld joint responses caused by these events are unrelated to progressive end-face wear. If these events are indiscriminately included in the deviation maintenance judgment, reference reconstruction and operational disturbances will be mistakenly identified as accelerated wear. Therefore, these events are organized into an end-face breakpoint record sequence, and trend interruptions caused by non-wear factors are identified by aligning the timestamps with adjacent weld joint windows.

[0049] The adjacent weld point window is defined by the end time of the previous weld point to the start time of the next weld point. This time interval covers all process intervals and state transitions between the two weld points. When the occurrence time of a record in the end face breakpoint record sequence falls into this window, or when the program number or batch number of the adjacent weld point changes, a breakpoint mark is formed. This mark is only used to truncate the deviation connection between adjacent weld points. That is, adjacent weld point pairs with breakpoint marks do not form a deviation holding segment, thereby avoiding the inclusion of grinding datum reset and sudden changes in working conditions in the wear trend calculation.

[0050] The deviation retention amount transforms the aforementioned spatial overlap information into a density index of deviation inheritance between adjacent solder joints, and its calculation formula is as follows:

[0051] In the formula, For the first From the first solder joint to the... The deviation between individual solder joints, after normalization... It is a dimensionless parameter; This refers to the number of deviation holding segments formed by the adjacent solder joints; For the first The common remaining area of ​​the deviation-maintaining segments; For the first The overlap phase length of each offset holding segment; To avoid extremely small positive numbers with a denominator of zero, Units and The units are the same. The above deviation holding amount The calculation formula only measures the residual density of the deviation holding segment in the same stage and direction and not cut off by the breakpoint. The deviation of the end face response of the previous weld point has been indirectly included in the calculation by participating in the pairing of deviation holding segments and is no longer used as an independent multiplier for repeated weighting.

[0052] It should be noted that the boundary of the role of the breakpoint marker in the deviation retention calculation needs to be strictly defined. The breakpoint marker is only used to sever the deviation connection between adjacent solder joints, and is not weighted together with the curve residual, nor does it participate in the numerical correction of the jointly retained residual area or overlapping phase length within the deviation retention segment. This decoupling design ensures that the end-face breakpoint suppression only acts on the trend continuity judgment level, without changing the quantitative indicators within the established deviation retention segment, keeping the wear trend calculation and the abnormal working condition isolation mechanism independent, and avoiding numerical coupling between the breakpoint suppression strength and the wear degree.

[0053] Step S150: Based on the deviation holding amount, convert the welding window and weld gap window into the equivalent duration of end face wear.

[0054] Optionally, step S150 includes: determining the welding window duration based on the power-on start record and power-on end record; determining the welding window duration based on the time interval between the end time of the previous welding point and the start time of the next welding point in adjacent welding points; calculating the welding window duration based on the deviation holding amount of the corresponding adjacent welding point pairs; adding the calculated welding window duration to the welding window duration to obtain the equivalent duration of a single window; and accumulating the equivalent durations of each single window within the life cycle of the same end face in the order of welding points to obtain the equivalent duration of end face wear.

[0055] The welding window characterizes the time span during which the electrode end face directly experiences thermo-pressure contact in a single welding process. It is determined by recording the start and end of energization. During this period, there is continuous current flow and mechanical pressure between the electrode and the workpiece, representing the direct impact period for electrolytic erosion, thermal fatigue, and mechanical wear of the end face material. The weld interval window, on the other hand, characterizes the process interval between two adjacent weld points. It is determined by the time interval between the end of the previous weld point and the start of the next weld point. During this period, the electrode end face is removed from direct thermo-pressure contact and is in a state of cooling, oxidation, and stress recovery. Natural time treats these two time intervals with vastly different physical mechanisms as equal, failing to reflect the true difference between intensive continuous welding and long periods of downtime waiting for end face wear to progress. Therefore, it cannot be used as a time axis for wear prediction.

[0056] The contribution of the weld window to the wear trend is not constant, but depends on the degree of recovery of the electrode end face between two welds. If there is a deviation in holding between adjacent weld points, it indicates that the end face response offset caused by the previous weld point is inherited by the subsequent weld point, and the end face has not returned to the vicinity of the stable reference within the weld window. This weld window period still has a driving effect on the subsequent wear trend. If no deviation in holding is formed, or if there are breakpoints such as grinding, cap replacement, or program switching between adjacent weld points, the end face state has been recovered or the reference has been rebuilt. The corresponding weld window should not continue to accumulate according to the same wear trend. This differentiated measurement of weld window time decouples the actual change in the end face state from the fixed cycle time.

[0057] The single-window equivalent duration integrates the direct hot-pressing contact time of the welding window and the deviation holding time of the weld window into a unified wear time metric. Its calculation formula is as follows:

[0058] In the formula, For the first The equivalent duration of a single window for each wear time window; The welding window duration for this weld joint is determined by the records at the start and end of power-on. The duration of the welding window between this weld point and the next weld point is determined by the timestamp interval between adjacent weld points. The deviation retention amount corresponding to adjacent weld point pairs characterizes the degree of inheritance of the end face deviation state within the weld window. The above-mentioned single-window equivalent duration... The calculation formula weights the weld window based on the deviation retention amount, while the weld window, as the direct hot-pressing contact time of the end face, is directly included without weighting.

[0059] The equivalent duration of a single window within the same end face's lifespan is accumulated according to the order in which the weld points occur, forming the cumulative equivalent duration of end face wear. The calculation formula is as follows:

[0060] In the formula, For the first The cumulative equivalent duration of end face wear over the life cycle of each end face; This represents the number of effective solder joints within the lifespan of this end face. The above refers to the cumulative end face wear equivalent duration. The scattered welding actions and process intervals on the natural time axis are transformed into a continuous time axis corresponding to the degree of end face wear, replacing the natural time or weld point number horizontal axis in traditional time series prediction.

[0061] Through the aforementioned conversion and accumulation mechanism, the equivalent duration of end face wear absorbs the differences caused by dense welding, intermittent waiting, re-welding, and datum reconstruction after grinding. Within the same natural time span, periods with dense, continuous weld points and consistently high deviations from the holding amount correspond to a larger accumulated equivalent duration, reflecting that the end face is in a state of continuous thermo-pressurized contact and insufficient recovery; while the accumulated equivalent duration during parts replacement waiting or shutdown cooling periods is smaller, reflecting that the end face has returned to near a stable datum.

[0062] Step S160: Based on the equivalent duration of end face wear, predict the electrode lifetime.

[0063] Optionally, step S160 includes: determining the lifespan boundary based on the equivalent duration of cumulative end-face wear before regrinding or cap replacement in the historical lifespan of similar electrode end faces; comparing the equivalent duration of cumulative end-face wear in the current lifespan of the end face with the lifespan boundary to determine the remaining lifespan ratio; and outputting an electrode maintenance prompt based on the remaining lifespan ratio.

[0064] The equivalent life boundary characterizes the wear equivalent time limit experienced by similar electrode end faces under the same welding procedures and plate combinations, from baseline reconstruction to loss of usability. The equivalent life boundary value is derived from the statistical analysis of the cumulative equivalent wear time of the end face before regrinding or cap replacement throughout its historical life cycle, rather than a simple count of natural days or the total number of weld points. Since the equivalent wear time of the end face has absorbed the differences caused by intensive welding, intermittent waiting, rework welding, and baseline reconstruction after regrinding, the life boundary established based on this time can reflect the process quantity of the end face's actual state progressing to the failure threshold, making the life limits under different production rhythms comparable. In addition to obtaining the equivalent life boundary through statistical analysis of the equivalent duration before historical grinding or cap replacement, it can also be calibrated using electrode cap end-face image detection, indentation detection, weld nugget quality records, or manual maintenance records. End-face image detection maps the geometric failure threshold to the equivalent duration limit by visually measuring the end-face diameter, roundness, and surface roughness. Indentation detection indirectly infers the degree of end-face deformation by measuring the depth and diameter of electrode indentations on the plate surface. Weld nugget quality records correlate events of insufficient weld nugget diameter or excessive spatter with the corresponding equivalent duration to establish a quality failure boundary. Manual maintenance records integrate the experience of field engineers in judging end-face failure states, serving as an auxiliary basis for data calibration. The boundary calibration results from multiple sources can be cross-validated, improving the reliability of life limit estimation.

[0065] The remaining lifespan ratio is obtained by comparing the cumulative wear equivalent duration of the front face over its lifetime with the equivalent lifespan boundary. The calculation formula is as follows:

[0066] In the formula, For the first The remaining lifetime percentage of each end face after normalization is calculated. It is a dimensionless parameter; This is the equivalent lifetime boundary for the same type of electrode under the same welding procedure and plate combination. This represents the cumulative wear-equivalent duration of the end face throughout its current lifespan. To avoid extremely small positive numbers with a denominator of zero, Units and The units are the same. Remaining lifespan percentage The current wear state is normalized and mapped to the historical failure limit. The higher the value, the greater the remaining usability of the end face. The lower the value, the closer the wear is to the limit.

[0067] The comparison between the remaining lifespan ratio and a preset threshold drives maintenance decision generation. When the remaining lifespan ratio is higher than the preset safety range, it indicates that the end face wear is in the normal progress range, and a prompt to continue production is output. When the remaining lifespan ratio enters the grinding warning range, a grinding prompt is output to arrange planned maintenance based on the current number of welding tasks and historical grinding records. When the remaining lifespan ratio is lower than the cap replacement boundary, or the previous end face breakpoint record shows that the reference cannot be restored after grinding, it indicates that the end face geometry has exceeded the grinding range, and a cap replacement prompt is output.

[0068] Optionally, step S160 further includes: if at least one of program switching, sheet metal batch switching, fixture alarm, or cooling abnormality occurs within the prediction period, a breakpoint marker is formed based on the program switching, sheet metal batch switching, fixture alarm, or cooling abnormality, and the continuous accumulation of the equivalent duration of end face wear is cut off based on the breakpoint marker.

[0069] When program switching, sheet metal batch switching, fixture alarms, or cooling anomalies occur within the prediction period, these events alter the welding conditions or introduce external disturbances. The resulting response changes should not be included in the current end-face wear trend. At this point, a breakpoint is marked based on the aforementioned anomalies, and the continuous accumulation of the equivalent end-face wear duration is truncated. This anomaly serves as the basis for breakpoint suppression or sequence reconstruction. The truncation operation stops the accumulation of equivalent duration at the anomaly point, preventing sudden changes in operating conditions and external disturbances from being misinterpreted as wear acceleration signals, and ensuring that each accumulated segment on the time axis corresponds to the same end-face state progression process.

[0070] Optionally, step S160 may further include: using the equivalent duration of end face wear as the time index of the time series prediction model; or, using the equivalent duration of a single window as the time step of each weld point in the time series prediction model.

[0071] For deployed time series forecasting models, there is no need to reconstruct the main model structure; simply replacing the time axis is sufficient to adapt to the equivalent duration of end-face wear. One possible implementation is to directly replace the model's original time index with the cumulative equivalent duration of end-face wear, transforming the model's input horizontal axis from natural time or weld point number to an equivalent time axis corresponding to the degree of end-face state progression. Another possible implementation is to use the single-window equivalent duration as the time step input for each weld point, allowing the model to perceive the differential effects of welding cycle time and end-face recovery state as it progresses point by point. Both methods can correct the time scale of wear trends without altering the prediction structure of the trend model, threshold model, or remaining life model.

[0072] This invention is now complete.

[0073] In summary, in this embodiment of the invention, the weld joint response curves are grouped according to the end face life cycle and welding conditions to obtain a sequence of weld joint response curves within the same end face life cycle; a stable end face reference curve is established based on the weld joint response curve sequence; the response curves of each weld joint in the weld joint response curve sequence are compared with the stable end face reference curve to determine the end face response deviation; the deviation holding amount is determined based on the end face response deviation of adjacent weld joints; the welding window and weld gap window are converted into the equivalent end face wear duration based on the deviation holding amount; and electrode life is predicted based on the equivalent end face wear duration.

[0074] This invention transforms the time axis for electrode wear prediction from natural time or cumulative weld count to the equivalent duration of end-face wear. By comparing the deviation of the stable end-face reference curve with the end-face response, judging the deviation holding amount of adjacent weld points, and calculating the deviation holding amount of welding windows and weld interval windows, a correspondence is established between the electrode life prediction and the actual progression of the electrode end-face condition. This reduces prediction errors caused by differences in production cycle time, rework welding, and changes in grinding reference. Furthermore, by grouping weld point response data according to end-face lifecycle and welding conditions, weld points under different electrode numbers, grinding and cap replacement records, welding procedures, and plate combinations are aggregated into the same comparable sequence. This avoids the overlap of weld point response curves from different end-face lifecycles and heterogeneous conditions, reducing the impact of variations in end-face lifecycle and welding conditions. The probability of misjudging wear trends caused by working condition switching is reduced. On the other hand, based on the stable end face reference curve, the current weld point response curve is compared phase by phase. By determining the over-limit of the measurement repeatability range and connecting the continuous deviation segments, the real change of the end face contact state is separated from single-point measurement noise and stage switching boundary spikes, improving the characterization accuracy of the end face response deviation amount for wear advancement. On the other hand, by judging the direction consistency of the effective deviation segments of adjacent weld point end faces, the phase interval overlap, and the alignment of the end face breakpoint recording sequence, a deviation holding amount is formed. This distinguishes the continuous advancement characteristics of electrode end face wear from isolated anomalies caused by plate overlap disturbance, instantaneous current shunting, program switching, or grinding reference reset, suppressing the interference of non-wear factors on the life prediction time axis.

[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting electrode wear in spot welding machines based on time series analysis, characterized in that, The method includes: The weld joint response curves are grouped according to the end face life cycle and welding conditions to obtain the weld joint response curve sequence within the same end face life cycle. Based on the weld joint response curve sequence, a stable end face reference curve is established; The response curve of each solder joint in the solder joint response curve sequence is compared with the stable end face reference curve to determine the end face response deviation. Based on the deviation of the end face response of adjacent solder joints, determine the deviation holding amount; Based on the aforementioned deviation holding amount, the welding window and weld gap window are converted into the equivalent duration of end face wear; Electrode lifetime is predicted based on the equivalent duration of end-face wear.

2. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 1, characterized in that, The step of comparing each solder joint response curve in the solder joint response curve sequence with the stable end face reference curve to determine the end face response deviation includes: The response curves of each weld point are normalized to the same phase sequence according to the welding stage, and compared with the stable end face reference curve phase by phase to obtain the phase residual sequence. The sequence of phase points in the residual phase sequence that continuously exceed the measurement repeatability range determined based on the stable end face reference curve and whose deviation directions are consistent are defined as the effective deviation segments of the end face. The deviation amount of the end face response is determined based on the residual area and phase length of the effective deviation segment of each end face.

3. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 2, characterized in that, The determination of the deviation holding amount based on the end face response deviation of adjacent solder joints includes: The effective deviation segments of the end faces of adjacent weld points are paired according to the welding stage and phase interval; When the deviation directions of the paired deviation segments are consistent and their phase intervals overlap, the overlapping interval between the paired deviation segments is determined as the deviation holding segment. The deviation holding amount is determined based on the overlap phase length of each of the deviation holding segments and the residual area jointly retained by the adjacent solder joints within the overlap phase length.

4. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 3, characterized in that, The method of determining the deviation holding amount based on the end face response deviation of adjacent solder joints further includes: The records of grinding, cap replacement, program switching, sheet metal batch switching, fixture alarm, flow abnormality, and cooling abnormality are compiled into an end face breakpoint record sequence. The end face breakpoint recording sequence is aligned with the adjacent solder joint window according to the timestamp; the adjacent solder joint window is determined from the end time of the previous solder joint to the start time of the next solder joint. A breakpoint marker is formed when there is a record in the end face breakpoint record sequence where the occurrence time falls within the window of the adjacent solder joint, or when the program number or batch number of the adjacent solder joint changes. Based on the breakpoint markings, the deviation between adjacent solder joints is cut off.

5. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 1, characterized in that, The step of converting the welding window and weld gap window into equivalent end-face wear duration based on the deviation holding amount includes: The welding window duration is determined based on the power-on start record and power-on end record; The welding window duration is determined based on the time interval between the end time of the previous welding point and the start time of the next welding point in adjacent welding points. The welding window duration is calculated based on the deviation retention amount of the corresponding adjacent welding point pairs. The calculated welding window duration is added to the welding window duration to obtain the equivalent duration of a single window. The equivalent duration of each single window within the same end face lifecycle is accumulated in the order of the solder joints to obtain the equivalent duration of end face wear.

6. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 5, characterized in that, The electrode lifetime prediction based on the equivalent duration of end-face wear includes: In the historical life cycle of similar electrode end faces, the life boundary is determined based on the equivalent duration of cumulative end face wear before grinding or cap replacement. The cumulative end-face wear equivalent duration of the current end-face life cycle is compared with the life boundary to determine the remaining life ratio; Based on the remaining lifespan percentage, output electrode maintenance prompts are provided.

7. The method for predicting electrode wear of a spot welding machine based on time series analysis according to claim 6, characterized in that, The electrode lifetime prediction process also includes: If at least one of the following occurs during the prediction period: program switching, sheet metal batch switching, fixture alarm, or cooling anomaly, a breakpoint marker is formed based on the program switching, sheet metal batch switching, fixture alarm, or cooling anomaly, and the continuous accumulation of the end face wear equivalent duration is truncated based on the breakpoint marker.

8. The method for predicting electrode wear of a spot welding machine based on time series analysis according to any one of claims 1-7, characterized in that, The electrode lifetime prediction based on the equivalent duration of end-face wear includes: The equivalent duration of end-face wear is used as the time index of the time series prediction model; Alternatively, the equivalent duration of a single window can be used as the time step for each solder joint in the time series prediction model.

9. The method for predicting electrode wear of a spot welding machine based on time series analysis according to any one of claims 1-7, characterized in that, The process of establishing a stable end-face reference curve based on the solder joint response curve sequence includes: From the weld point response curve sequence, select weld points that do not have grinding breakpoints, fixture alarms, or current shunting abnormalities under the same welding procedure and the same plate combination after grinding or cap replacement, and whose contact state is stable, to form a stable end face reference curve.

10. The method for predicting electrode wear of a spot welding machine based on time series analysis according to any one of claims 1-7, characterized in that, The process of grouping weld joint response curves according to end-face lifecycle and welding conditions to obtain a sequence of weld joint response curves within the same end-face lifecycle includes: The life cycle of the end face is defined according to the electrode number, grinding and cap replacement record, welding procedure, and plate assembly. Collect weld joint response data, which includes voltage, current, pressure, displacement, welding start time, welding end time, program switching record, plate batch record, fixture alarm record, current shunting abnormality record, and cooling abnormality record; A dynamic resistance curve is formed based on the voltage and the current, and the boundaries of the closing, energizing, effective heating, holding and falling-back stages are determined based on the pressure or the displacement. The solder joint response data are organized into a solder joint response curve sequence according to the order in which the solder joints occur. Form adjacent solder joints into adjacent solder joint response pairs; The grinding, cap replacement, program switching, sheet metal batch switching, fixture alarm, cooling abnormality, and flow diversion abnormality were organized into an end face breakpoint record sequence.

Images