A method and system for evaluating the durability of an automobile starter

Abstract

The application relates to a kind of automobile starter endurance evaluation method and system.The application includes obtaining the voltage time series data and current time series data corresponding to automobile starter endurance test;Form current time series curve according to current time series data;Traverse current time series curve, determine global maximum current value;Retrieve first effective wave crest and second effective wave crest;Determine the reference wave crest in the time series in front of first effective wave crest and second effective wave crest;With the sampling position of reference wave crest as reference, continuously intercept a preset number of current sampling points backward, obtain steady-state current data;According to the same sampling serial number and original time stamp as steady-state current data, steady-state voltage data is synchronously obtained from voltage time series data, so that steady-state current data and steady-state voltage data correspond one by one;Obtain linear fitting equation and goodness of fit;According to goodness of fit, evaluate the smoothness and whole machine health degree of starter steady-state operation.The application realizes the effect of early prediction of starter failure.

Description

Technical Field

[0001] This invention relates to the field of automotive starter performance testing and durability analysis technology, and in particular to a method and system for evaluating the durability of automotive starters. Background Technology

[0002] Automotive starters require a large electromagnetic torque to be output at the moment of startup, and their current waveform typically exhibits significant transient impact characteristics. In endurance testing, factors such as battery voltage, instantaneous engine mechanical resistance, starter brush wear, commutator condition, bearing friction, and the number of usage cycles affect the current waveform of the same model of starter in different starting events, resulting in no fixed pattern. In actual testing, the applicant found that the current waveform often exhibits two distinct peaks that form sequentially. In some starting processes, the first peak has a larger amplitude, while in others, the second peak has a larger amplitude. Furthermore, due to sampling noise, sensor disturbances, and electromagnetic interference at the scene, the actual curve may also contain scattered waves, small fluctuations, and spikes.

[0003] In existing testing methods, relying solely on a fixed threshold to capture the first peak makes it easy to select the wrong reference position, resulting in errors in the steady-state data extraction. At the same time, traditional testing only relies on the size of the peak to roughly judge whether it is good or bad, and cannot quantify hidden faults such as internal wear of the starter motor, brush aging, and bearing jamming. The evaluation criteria are vague, and it is impossible to predict failure in advance, which cannot meet the precise control requirements of large-scale durability testing. Summary of the Invention

[0004] Therefore, the technical problem to be solved by the present invention is to overcome the problem that relying on a fixed threshold to capture the first peak in the durability test analysis of automobile starters in the prior art makes it easy to select the wrong reference position and cause errors in the steady-state data interception.

[0005] To solve the above-mentioned technical problems, the present invention provides a method for evaluating the durability of an automotive starter motor, comprising: The voltage timing data and current timing data corresponding to the durability test of the automobile starter are obtained. The voltage timing data and the current timing data are both raw sampled data and have the same time base. A current timing curve is generated based on the current timing data; By traversing the current timing curves, the global maximum current value is determined; Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, a first effective peak containing the sampling position corresponding to the global maximum current value is determined. In the time before and time after the first effective peak, a second effective peak adjacent to the first effective peak and separated by a valley is retrieved. The first effective peak and the second effective peak have a continuous rising segment, a peak apex, and a continuous falling segment. The earlier time-series peak between the first effective peak and the second effective peak is determined as the reference peak; Using the sampling position of the reference peak as a reference, a preset number of current sampling points are continuously intercepted to obtain steady-state current data; Steady-state voltage data is synchronously obtained from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one; Using the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis, voltage-current scatter data is constructed. The least squares method is used to perform linear fitting on the voltage-current scatter data to obtain the linear fitting equation and the goodness of fit R². Based on the goodness of fit, the stability of the starter's steady-state operation and the overall health of the machine are evaluated.

[0006] In one embodiment of the present invention, a first valid peak containing the sampling position corresponding to the global maximum current value is determined, and a second valid peak adjacent to the first valid peak and separated by a valley is retrieved in the time before and time after the first valid peak, including: Determine the difference sequence based on the current difference between adjacent sampling points in the current time series data; Based on the median, mean absolute deviation, or quantile statistics of the difference sequence, the noise reference amplitude is determined; A preset multiple of the noise reference amplitude is used as the adaptive judgment amplitude to distinguish between the actual start-up shock wave peak and the instantaneous disturbance. Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, candidate vertices are retrieved forward and backward along the time axis, and the sampling points that satisfy the local maximum value condition are determined as candidate vertices; For any candidate vertex, the rising start point is retrieved before the time of the candidate vertex, and the falling end point is retrieved after the time of the candidate vertex. A candidate peak is then determined based on the rising start point, the candidate vertex, and the falling end point, wherein: When the current increment of the candidate peak relative to the rising start point is not less than the adaptive judgment amplitude, and the current fall amount of the candidate peak relative to the falling end point is not less than the adaptive judgment amplitude, the candidate peak is determined to meet the peak height and fall amount requirements. The peak width of the candidate peak is determined based on the number of sampling points or the time length between the rising start point and the falling end point. When the peak width is not less than the preset minimum width threshold, the candidate peak is determined to meet the peak width requirement. The allowable amplitude of the continuous rising and falling segments of the candidate peak does not exceed the noise reference amplitude. When the proportion of sampling points that satisfy the overall upward trend in the sampling points before the time of the candidate peak reaches a preset proportion, and the proportion of sampling points that satisfy the overall downward trend in the sampling points after the time of the candidate peak reaches a preset proportion, it is determined that the candidate peak has a complete rising segment and a complete falling segment. Candidate peaks that simultaneously meet the requirements for peak height and fallback, peak width, and complete rising and falling segments are identified as independent valid peaks. The independent effective peak containing the sampling position corresponding to the global maximum current value is determined as the first effective peak, and the independent effective peak adjacent to the first effective peak and separated by a valley is selected as the second effective peak in the time before and time after the first effective peak.

[0007] In one embodiment of the present invention, it further includes: When there is a lowest valley point between two adjacent candidate vertices, and the current drop of the lowest valley point relative to the candidate vertex with the smaller current value among the two adjacent candidate vertices is not less than the adaptive determination amplitude, it is determined that a valley separation is formed between the two adjacent candidate vertices. When no valley separation is formed between two adjacent candidate vertices, or when the sampling interval between two adjacent candidate vertices is less than the preset minimum interval, it is determined that the two adjacent candidate vertices belong to the top local fluctuation, rising segment local fluctuation or falling segment local fluctuation on the same effective peak, and the two adjacent candidate vertices are merged into the same effective peak, and the candidate vertex with the larger current value is retained as the peak position of the same effective peak.

[0008] In one embodiment of the present invention, using the sampling position of the reference peak as a reference, a preset number of current sampling points are continuously intercepted backward to obtain steady-state current data, including: Using the sampling position of the reference peak as the reference point, and taking it as the steady-state starting position, a preset number of current sampling points are continuously intercepted from the steady-state starting position. The preset number of current sampling points includes another valid peak, so as to obtain the steady-state current data after entering the stable working stage.

[0009] In one embodiment of the present invention, the slope, intercept, and goodness of fit are calculated according to the following formula: ; ; ; ; Where U represents the steady-state voltage, I represents the steady-state current, k represents the linear fitting slope, b represents the linear fitting intercept, and n represents the number of sampling points within the steady-state interval. This represents the i-th steady-state current sample value. This represents the i-th steady-state voltage sample value. This represents the average value of the steady-state current samples. This represents the average value of the steady-state voltage samples. R² represents the i-th predicted voltage value calculated based on the linear fitting equation, and R² represents the goodness of fit.

[0010] In one embodiment of the present invention, the smoothness of the starter's steady-state operation and the overall health of the machine are evaluated based on the goodness-of-fit R², including: When R² ≥ 0.91, it is judged to be in a state of excellent voltage-current linearity; When 0.85 ≤ R² < 0.91, it is considered a normal fluctuation state; When 0.75 ≤ R² < 0.85, it is judged as a slight natural aging state; When 0.65≤R²<0.75, it is determined that the starter motor has significant internal wear and requires close monitoring. When R² < 0.65, it is determined that the linear relationship between voltage and current is disordered and there is an imminent risk of damage.

[0011] The present invention also provides an automotive starter motor durability evaluation system, comprising: The data acquisition module is used to acquire voltage timing data and current timing data corresponding to the durability test of the automobile starter. The voltage timing data and the current timing data are both raw sampled data and have the same time reference. A current timing curve acquisition module is used to generate a current timing curve based on the current timing data. The global maximum current value determination module is used to traverse the current timing curves to determine the global maximum current value; The effective peak acquisition module is used to determine a first effective peak containing the sampling position corresponding to the global maximum current value, using the sampling position corresponding to the global maximum current value as the retrieval anchor point, and to retrieve a second effective peak adjacent to the first effective peak and separated by a valley in the time before and time after the first effective peak, wherein the first effective peak and the second effective peak have a continuous rising segment, a peak apex and a continuous falling segment; The reference peak acquisition module is used to determine the earlier time-series peak between the first valid peak and the second valid peak as the reference peak; The data interception module is used to continuously intercept a preset number of current sampling points based on the sampling position of the reference peak to obtain steady-state current data. The data correspondence module is used to synchronously obtain steady-state voltage data from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one; The linear fitting module is used to construct voltage-current scatter data with the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis, and to perform linear fitting on the voltage-current scatter data using the least squares method to obtain the linear fitting equation and the goodness of fit R². The evaluation module is used to evaluate the smoothness of the starter's steady-state operation and the overall health of the machine based on the goodness of fit.

[0012] The technical solution of the present invention has the following advantages over the prior art: The present invention discloses a method and system for evaluating the durability of an automotive starter motor. Instead of relying on a fixed threshold to capture the first peak, it combines the global maximum current value to locate and identify the first and second peaks. This method can adapt to operating conditions where the amplitude of the first peak is higher and the amplitude of the second peak is higher, thereby reducing the risk of misjudging the double-peak waveform.

[0013] This invention identifies scattered waves and small fluctuations that lack a complete rise and fall pattern, have insufficient peak height or fall amount, and lack a clear valley separation from the main peak as interference signals, thus ensuring effective peak identification results.

[0014] This invention combines goodness of fit for starter health evaluation, which can transform traditional experience-based judgment into quantitative judgment and identify risks such as poor contact, abnormal commutation, inconsistent load, and impending failure in advance.

[0015] This invention relies on the TDMS standard test data format and the LabVIEW industrial testing platform, and can directly connect to existing starter durability test benches without hardware modification. It has low implementation costs and good batch testing adaptability. Attached Figure Description

[0016] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0017] Figure 1 This is a flowchart of the automobile starter durability evaluation method of the present invention.

[0018] Figure 2 This is an interface diagram of the test results (500 sampling points extracted backwards) for test number 000002 in this embodiment of the invention.

[0019] Figure 3 This is an interface diagram of the test results (500 sampling points extracted backwards) for test number 000003 in this embodiment of the invention.

[0020] Figure 4 This is an interface diagram of the test results (1000 sampling points are extracted backwards) for test number 000002 in this embodiment of the invention.

[0021] Figure 5 This is an interface diagram of the test results (1000 sampling points are extracted backwards) for test number 000003 in this embodiment of the invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0023] Example 1 Reference Figure 1 As shown, a method for evaluating the durability of an automotive starter motor includes: S1. Obtain voltage timing data and current timing data corresponding to the durability test of the automobile starter. The voltage timing data and the current timing data are both raw sampled data and have the same time reference.

[0024] It should be noted that this method is applicable to automotive starter motor durability test benches. The test bench synchronously acquires voltage, current, and other auxiliary channel data during each start-up test event and stores it in TDMS format. The analysis system can be deployed on a LabVIEW industrial test platform or on other industrial control test platforms capable of reading TDMS data.

[0025] Read the TDMS format durability test file stored in the durability testing equipment. Operators can select any set of start test events using interactive controls in the waveform display module, such as data name, curve list, previous, and next. After reading the selected start test event, send the raw channel data.

[0026] The waveform display module is used to display the current and voltage timing curves throughout the startup process, allowing operators to visually observe startup impact changes, such as... Figures 2 to 5 As shown. It should be noted that in this embodiment, the waveform display module displays the median-filtered or smoothed curve for interface observation, but this visualized curve does not participate in subsequent peak identification, steady-state truncation, and health rating calculations. That is to say, the identification process in this embodiment is based on the original sampled data to avoid distortion of the true peak value, peak position, and waveform timing information caused by filtering processing.

[0027] Voltage and current timing data are extracted separately from the same test launch event. The voltage and current channels originate from the same test bench, use the same sampling frequency, and are synchronously triggered for acquisition; therefore, they share a unified time axis and sampling sequence number. During data parsing, the system directly establishes the dual-channel correspondence between voltage and current data based on the shared sampling sequence number and original timestamp.

[0028] S2. Generate a current timing curve based on the current timing data.

[0029] S3. Traverse the current timing curves to determine the global maximum current value.

[0030] S4. Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, determine the first effective peak containing the sampling position corresponding to the global maximum current value. In the time before and time after the first effective peak, retrieve the second effective peak that is adjacent to the first effective peak and separated by a valley. The first effective peak and the second effective peak have a continuous rising segment, a peak apex and a continuous falling segment.

[0031] Step S4 specifically includes: Determine the difference sequence based on the current difference between adjacent sampling points in the current time series data; Based on the median, mean absolute deviation, or quantile statistics of the difference sequence, the noise reference amplitude is determined; A preset multiple of the noise reference amplitude is used as the adaptive judgment amplitude to distinguish between the actual start-up shock wave peak and the instantaneous disturbance. Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, candidate vertices are retrieved forward and backward along the time axis, and sampling points that satisfy the local maximum condition are determined as candidate vertices; For any candidate vertex, the rising start point is retrieved before the time of the candidate vertex, and the falling end point is retrieved after the time of the candidate vertex. A candidate peak is then determined based on the rising start point, the candidate vertex, and the falling end point, wherein: When the current increment of the candidate peak relative to the rising start point is not less than the adaptive judgment amplitude, and the current fall amount of the candidate peak relative to the falling end point is not less than the adaptive judgment amplitude, the candidate peak is determined to meet the peak height and fall amount requirements. The peak width of the candidate peak is determined based on the number of sampling points or the time length between the rising starting point and the falling ending point. When the peak width is not less than the preset minimum width threshold, the candidate peak is determined to meet the peak width requirement. The allowable amplitude of the continuous rising and falling segments of the candidate peak does not exceed the noise reference amplitude. When the proportion of sampling points that satisfy the overall upward trend in the sampling points before the time of the candidate peak reaches a preset proportion, and the proportion of sampling points that satisfy the overall downward trend in the sampling points after the time of the candidate peak reaches a preset proportion, it is determined that the candidate peak has a complete rising segment and a complete falling segment. Candidate peaks that simultaneously meet the requirements for peak height and fallback, peak width, and complete rising and falling segments are identified as independent valid peaks. The independent effective peak containing the sampling position corresponding to the global maximum current value is determined as the first effective peak, and the independent effective peak adjacent to the first effective peak and separated by a valley is selected as the second effective peak in the time before and time after the first effective peak.

[0032] Specifically, it also includes: When there is a lowest valley point between two adjacent candidate vertices, and the current drop of the lowest valley point relative to the candidate vertex with the smaller current value among the two adjacent candidate vertices is not less than the adaptive determination amplitude, it is determined that a valley separation is formed between the two adjacent candidate vertices. When no valley separation is formed between two adjacent candidate vertices, or when the sampling interval between two adjacent candidate vertices is less than the preset minimum interval, it is determined that the two adjacent candidate vertices belong to the top local fluctuation, rising segment local fluctuation or falling segment local fluctuation on the same effective peak, and the two adjacent candidate vertices are merged into the same effective peak, and the candidate vertex with the larger current value is retained as the peak position of the same effective peak.

[0033] S5. The peak with the earlier time sequence among the first effective peak and the second effective peak is determined as the reference peak.

[0034] It should be noted that the baseline peak corresponds to the initial impact stage of the armature stall, while the subsequent peak corresponds to the starting load change stage.

[0035] S6. Using the sampling position of the reference peak as a reference, continuously extract a preset number of current sampling points to obtain steady-state current data. This includes: Using the sampling position of the reference peak as the reference point, as the steady-state starting position, and continuously extracting a preset number of current sampling points (such as 500 or 1000) from the steady-state starting position, the preset number of current sampling points includes the sampling position of another valid peak and a predetermined number of sampling points thereafter, to obtain the steady-state current data after entering the stable working stage.

[0036] It should be noted that selecting the earlier peak in the time sequence as the cutoff point is to unify the analysis standards under different starting conditions. Specifically, during the durability test, the starter motor may exhibit two types of conditions: one with a larger amplitude peak in the first wave and the other with a larger amplitude peak in the second wave. If the peak with the largest amplitude is fixed as the cutoff point, the cutoff point may vary between the earlier and later peaks in different batches and samples of test data. This would lead to inconsistencies in the starting point for analysis across samples, hindering cross-sectional comparisons of batch test data and potentially introducing misjudgments due to inconsistent cutoff points. Therefore, by uniformly selecting the earlier peak in the time sequence as the cutoff point, all samples are subjected to data extraction and feature analysis according to consistent time positioning rules, thereby improving the comparability, stability, and accuracy of durability assessment results.

[0037] In this embodiment, let This represents the i-th current sample value, and the current difference between adjacent sample points is: ; The noise reference amplitude can be determined based on the median of the difference sequence: ; ; The noise reference amplitude can also be determined based on the mean absolute deviation or quantile statistics of the difference sequence. This represents the difference between the i-th and i-th adjacent sampling points. This represents the i-th current sample value. This represents the (i-1)th current sample value. Indicates the noise reference amplitude. This represents the median calculation. The noise reference amplitude is used to characterize the typical amplitude of field sampling noise, small jitter, and instantaneous sensor disturbances.

[0038] After determining the noise reference amplitude, the system generates an adaptive judgment amplitude based on the noise reference amplitude: ; in, Indicates the adaptive judgment range. This represents a preset multiplier coefficient greater than 1. This represents the noise reference amplitude. The adaptive judgment amplitude is not a fixed current threshold, but varies with the sampling accuracy, sampling frequency, and ambient noise level of the durability testing equipment. It is used to determine whether candidate peaks are distinct from scattered waves and small fluctuations.

[0039] Candidate vertices are searched forward and backward along the time axis around the location of the global maximum current value. For any candidate vertex j, the system searches for the rising start point l ahead of its time axis and the falling end point r behind its time axis. The left peak height of candidate vertex j relative to the rising start point l is... and the rightward pullback relative to the descent endpoint r They are respectively: ; ; in, This represents the current sample value corresponding to candidate vertex j. This represents the current sampling value corresponding to the starting point l. This represents the current sampling value corresponding to the endpoint r of the descent. Indicates the height of the left peak. This indicates the volume of the pullback on the right side. When... Not less than ,and Not less than This indicates that the candidate vertex has a sufficiently significant peak height and decline amplitude relative to the preceding and following lower intervals.

[0040] To avoid misidentifying single-point spikes or localized small waves as valid peaks, the system further calculates the percentage of upward trends. and the proportion of downward trends In one implementation, the percentage of upward trends... It can be represented as: ; percentage of downward trends It can be represented as: ; in, This represents the number of sampling points that satisfy the overall upward trend from the starting point l to the candidate vertex j. This represents the number of sampling points satisfying the overall downward trend from candidate vertex j to the descent endpoint r, where jl represents the sampling interval of the rising segment and rj represents the sampling interval of the falling segment. Both the overall upward trend and the overall downward trend are allowed to have amplitudes not exceeding the noise reference amplitude. .when and When both reach the preset ratio, the candidate vertex is determined to have a complete ascending segment and a complete descending segment.

[0041] In this embodiment, local fluctuations that do not meet the requirements of a complete rise and fall trend are excluded. Specifically, if a local maximum only rises without a significant fall, or only falls without a complete rise, it is determined to be an incomplete rise and fall pattern; if the candidate vertex... or Less than If the peak height or fall is insufficient, it is determined that the peak height or fall is insufficient. If the peak width of the candidate peak is less than the preset minimum width threshold that matches the sampling frequency, it is determined to be a single-point spike or short-term jitter. If the candidate peak is located in the rising segment, top plateau segment, or falling segment of a larger peak, and there is no valley separating it from the larger peak, it is determined to be a local small fluctuation on the same candidate peak. The above local maxima are not used as independent valid peaks in the determination of the first and second valid peaks.

[0042] In determining the independence of adjacent candidate peaks, let pa and pb be two adjacent candidate peaks, and v be the lowest valley point between them. Then the valley separation value is... It can be represented as: ; in, This represents the current sample value at candidate vertex pa. This represents the current sample value at candidate vertex pb. This represents the current sample value at the lowest valley point v between two candidate vertices. This represents the valley separation between two candidate vertices. Less than If the sampling interval between two candidate vertices is less than the minimum interval matching the sampling frequency, they are determined to belong to the top local fluctuation, rising segment local fluctuation or falling segment local fluctuation on the same candidate peak, and the two are merged into the same candidate peak, retaining the candidate vertex with the larger current value as the peak position of the same candidate peak.

[0043] Based on the verification results above, two independent valid peaks are determined. The system first identifies the independent valid peak containing the sampling position corresponding to the global maximum current value and designates this independent valid peak as the first valid peak. Then, in the time range before and after the first valid peak, another independent valid peak is retrieved that is adjacent to the first valid peak, has a valley separation, and also meets the requirements of a complete rising segment, a complete falling segment, peak height, fallback amount, and peak width. This other independent valid peak is designated as the second valid peak. Thus, regardless of whether the global maximum current value is located in an earlier or later valid peak in the time sequence, the system can complete the dual-peak identification according to a unified rule.

[0044] It should be noted that at the moment of power-on, the starter armature is in a near-locked state, with the mechanical load at its maximum, which easily forms an initial inrush current peak. Subsequently, as the starter drives the engine, the load state changes dynamically, and the current curve may rise again, forming a secondary peak. This method, through effective peak identification, can determine two adjacent independent effective peaks under different operating conditions, such as when the first peak amplitude is large, the second peak amplitude is large, or the two peak amplitudes are close.

[0045] Furthermore, since the effective shock wave peak at the moment of starter motor start-up must belong to the main energy fluctuation in the entire starting current curve, using the position of the global maximum current value as the retrieval anchor point can avoid local maxima such as pre-power-on jitter, sampling glitches, sensor noise, and instantaneous contact disturbances being directly misjudged as effective peaks.

[0046] S7. Obtain steady-state voltage data synchronously from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one.

[0047] Since the voltage and current channels are synchronously triggered and acquired by the same rack, and the sampling frequency is consistent, the system does not need to use nearest neighbor matching, linear interpolation or other secondary correction methods to achieve strict one-to-one matching of voltage and current data.

[0048] S8. Construct voltage-current scatter data with the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis. Use the least squares method to perform linear fitting on the voltage-current scatter data to obtain the linear fitting equation and the goodness of fit R².

[0049] Specifically, the slope, intercept, and goodness of fit are calculated using the following formulas: ; ; ; ; Where U represents the steady-state voltage, I represents the steady-state current, k represents the linear fitting slope, b represents the linear fitting intercept, and n represents the number of sampling points within the steady-state interval. This represents the i-th steady-state current sample value. This represents the i-th steady-state voltage sample value. This represents the average value of the steady-state current samples. This represents the average value of the steady-state voltage samples. R² represents the i-th predicted voltage value calculated based on the linear fitting equation, and R² represents the goodness of fit.

[0050] In engineering terms, the k-value is closely related to the change in the equivalent internal resistance of the starter motor and directly corresponds to the equivalent internal resistance characteristics of the starter motor. A stable k-value indicates that the starter motor brushes, armature, coil, and contact resistance are in normal condition; abnormal fluctuations, increases, or decreases in the k-value indicate potential faults such as internal wear, poor contact, coil aging, and mechanical jamming.

[0051] The intercept b represents the starter's no-load reference voltage characteristic, reflecting the base voltage level of the power supply circuit. It can be used to determine whether the voltage environment under test conditions is normal and to eliminate the interference of power supply fluctuations on the test results. R² is used to quantify the strength of the linear correlation between voltage and current, directly reflecting the stability of the starter's steady-state operation and the overall health of the machine. It is a core quantitative indicator for judging starter aging, failure, and impending damage. The closer the R² value is to 1, the better the starter's working linearity, the more stable the operation, and the better the equipment condition; the lower the R² value, the more disordered the operating conditions, the larger the internal resistance fluctuations, and the more serious the equipment wear and aging.

[0052] S9. Based on the goodness of fit, evaluate the stability of the starter's steady-state operation and the overall health of the machine. Specifically, this includes: When R²≥0.91, it is judged to be in a state of excellent voltage and current linearity, that is, there is no obvious wear inside the starter, the operation is extremely stable, and it is in a state of excellent new / near new condition. When 0.85 ≤ R² < 0.91, it is considered a normal fluctuation state; When 0.75≤R²<0.85, it is judged to be in a state of slight natural aging, with a slight decrease in performance, stable working conditions, and can be used normally in vehicles for a long time. When 0.65≤R²<0.75, it is determined that there is significant internal wear, large fluctuation in internal resistance, and decreased stability of the operating condition, requiring key monitoring and advance planning of maintenance; When R² < 0.65, it is determined that the voltage-current linear relationship is severely disordered, indicating hidden faults such as severe brush wear, bearing jamming, and coil aging. The equipment is nearing damage and it is recommended to repair or scrap it.

[0053] This method also displays the analysis results on the interface. The output may include the peak value and sampling location of the first peak (after filtering), the peak value and sampling location of the second peak, a UI scatter plot, a fitted trend line, a linear fitting equation, R², and the corresponding health rating result. Through the output results, operators can intuitively see the changes in current surge throughout the startup process, the dual-peak identification results, the fitted parameters, and the starter's health status.

[0054] Example 2

[0055] This embodiment provides the experimental application process. An automotive starter motor durability test bench performs continuous starting durability cycles on the starter motor under test, simultaneously acquiring raw timing data from the voltage and current channels and generating a TDMS file. After reading the selected TDMS file, the analysis system automatically extracts channel data, identifies the first and second peaks, determines the reference peak, extracts steady-state data, and completes UI linear fitting.

[0056] Reference Figure 2As shown, in the startup event of test number 000002 (higher first peak condition), the first peak current was 1809.89A and the second peak current was 1714.80A, with the first peak having a larger amplitude. The system identifies the first and second peaks according to the time sequence, and uses the first peak, which is earlier in the time sequence, as the reference peak. It then shifts backward by 500 sampling points and extracts steady-state data to complete the fitting. The fitting equations are as follows: ; The R² value for this group is 0.9252, which falls within the R² ≥ 0.91 range. This indicates that there is no significant wear inside the starter motor, and it runs extremely smoothly, indicating that it is in excellent condition, either brand new or nearly new.

[0057] Reference Figure 3 As shown, in the startup event of test number 000003 (higher peak condition), the first peak current is 1655.92A, and the second peak current is 1722.68A, with the second peak having a larger amplitude. Under this reverse peak condition, the system can still accurately identify the two valid peaks and use the earlier peak as the unified reference peak to complete steady-state truncation. It then shifts backward by 500 sampling points and extracts the steady-state data to complete the fitting. The fitting equations are: ; The R² value of this group is 0.9340, indicating good steady-state operation and that the overall machine is in a state of slight natural aging and stable performance.

[0058] As can be seen from the above, the analysis system can stably distinguish between the first and second peaks of the time series, complete steady-state truncation and UI fitting based on a unified benchmark, and the output linear formula and R² value show stable trends without identification errors or calculation anomalies. It can meet the requirements of long-term operation stability and batch detection adaptation.

[0059] For comparison, refer to Figure 4 As shown, in the startup event of test number 000002 (higher first peak condition), the first peak current was 1809.89A and the second peak current was 1714.80A, with the first peak having a larger amplitude. The system identifies the first and second peaks according to the time sequence, and uses the first peak, which is earlier in the time sequence, as the reference peak. It then shifts backward by 1000 sampling points and extracts steady-state data to complete the fitting. The fitting equations are as follows: ; The R² for this group is 0.9071.

[0060] For comparison, refer to Figure 5As shown, in the startup event of test number 000003 (higher peak condition), the first peak current is 1655.92A and the second peak current is 1722.68A, with the second peak having a larger amplitude. Under this reverse peak condition, the system can still accurately identify the two valid peaks and uses the first peak, which is earlier in the time sequence, as the unified reference peak to complete steady-state truncation. It then shifts backward by 1000 sampling points and extracts the steady-state data to complete the fitting. The fitting equations are: ; The R² value for this group is 0.9140.

[0061] As can be seen from the above, the goodness of fit R² exhibits a certain variation pattern under different truncation lengths.

[0062] When the truncation length is 500 sampling points, the extracted data is mainly concentrated in the optimal stable range during the initial steady-state phase of the starter motor. Within this range, the starter motor speed has not yet fully established itself into the later fluctuation stage, and the current and voltage signals are less affected by factors such as downstream commutation noise and battery voltage drop disturbances. Therefore, the overall goodness of fit R² is high, and the numerical fluctuation range is small. In this embodiment, for large-scale durability test data, using 500 sampling points as the preferred truncation length ensures good consistency and comparability between different samples, which is beneficial for improving the stability of lateral comparisons and reducing misjudgments introduced by later disturbances.

[0063] When a 1000-sampling-point length is selected, the truncated data further covers the latter half of the waveform. This portion of the data easily incorporates noise accumulated during motor commutation, minor battery voltage drop disturbances, and other non-fault-related fluctuations during operation, leading to increased data dispersion. Under normal operating conditions, the goodness-of-fit R² of the same prototype may also show a significant decrease, further amplifying the range of numerical fluctuations between normal samples.

[0064] Based on the above analysis, 500 sampling points can be used as the preferred truncation length for routine batch durability screening to avoid the influence of noise in the later part of the waveform and improve the consistency of evaluation and the stability of horizontal comparison; 1000 sampling points can be used as the extended truncation length for prototype deep aging analysis to observe waveform attenuation, disturbance accumulation and aging trend changes over a longer time range.

[0065] Example 3 Based on the same inventive concept, this embodiment provides an automobile starter durability assessment system. The principle of solving the problem is similar to that of the automobile starter durability assessment method described above, and the repeated parts will not be repeated.

[0066] This embodiment provides a vehicle starter motor durability evaluation system, including: The data acquisition module is used to acquire voltage timing data and current timing data corresponding to the durability test of the automobile starter. The voltage timing data and the current timing data are both raw sampled data and have the same time reference. A current timing curve acquisition module is used to generate a current timing curve based on the current timing data. The global maximum current value determination module is used to traverse the current timing curves to determine the global maximum current value; The effective peak acquisition module is used to determine a first effective peak containing the sampling position corresponding to the global maximum current value, using the sampling position corresponding to the global maximum current value as the retrieval anchor point, and to retrieve a second effective peak adjacent to the first effective peak and separated by a valley in the time before and time after the first effective peak, wherein the first effective peak and the second effective peak have a continuous rising segment, a peak apex and a continuous falling segment; The reference peak acquisition module is used to determine the earlier time-series peak between the first valid peak and the second valid peak as the reference peak; The data interception module is used to continuously intercept a preset number of current sampling points based on the sampling position of the reference peak to obtain steady-state current data. The data correspondence module is used to synchronously obtain steady-state voltage data from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one; The linear fitting module is used to construct voltage-current scatter data with the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis, and to perform linear fitting on the voltage-current scatter data using the least squares method to obtain the linear fitting equation and the goodness of fit R². The evaluation module is used to evaluate the smoothness of the starter's steady-state operation and the overall health of the machine based on the goodness of fit.

[0067] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0071] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the durability of an automotive starter motor, characterized in that, include: The voltage timing data and current timing data corresponding to the durability test of the automobile starter are obtained. The voltage timing data and the current timing data are both raw sampled data and have the same time base. A current timing curve is generated based on the current timing data; By traversing the current timing curves, the global maximum current value is determined; Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, a first effective peak containing the sampling position corresponding to the global maximum current value is determined. In the time before and time after the first effective peak, a second effective peak adjacent to the first effective peak and separated by a valley is retrieved. The first effective peak and the second effective peak have a continuous rising segment, a peak apex, and a continuous falling segment. The earlier time-series peak between the first effective peak and the second effective peak is determined as the reference peak; Using the sampling position of the reference peak as a reference, a preset number of current sampling points are continuously intercepted to obtain steady-state current data; Steady-state voltage data is synchronously obtained from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one; Using the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis, voltage-current scatter data is constructed. The least squares method is used to perform linear fitting on the voltage-current scatter data to obtain the linear fitting equation and the goodness of fit R². Based on the goodness of fit, the stability of the starter's steady-state operation and the overall health of the machine are evaluated.

2. The method for evaluating the durability of an automobile starter motor according to claim 1, characterized in that, Determine the first valid peak containing the sampling position corresponding to the global maximum current value, and in the time before and time after the first valid peak, retrieve the second valid peak adjacent to the first valid peak and separated by a valley, including: Determine the difference sequence based on the current difference between adjacent sampling points in the current time series data; Based on the median, mean absolute deviation, or quantile statistics of the difference sequence, the noise reference amplitude is determined; A preset multiple of the noise reference amplitude is used as the adaptive judgment amplitude to distinguish between the actual start-up shock wave peak and the instantaneous disturbance. Using the sampling position corresponding to the global maximum current value as the retrieval anchor point, candidate vertices are retrieved forward and backward along the time axis, and sampling points that satisfy the local maximum condition are determined as candidate vertices; For any candidate vertex, the rising start point is retrieved before the time of the candidate vertex, and the falling end point is retrieved after the time of the candidate vertex. A candidate peak is then determined based on the rising start point, the candidate vertex, and the falling end point, wherein: When the current increment of the candidate peak relative to the rising start point is not less than the adaptive judgment amplitude, and the current fall amount of the candidate peak relative to the falling end point is not less than the adaptive judgment amplitude, the candidate peak is determined to meet the peak height and fall amount requirements. The peak width of the candidate peak is determined based on the number of sampling points or the time length between the rising starting point and the falling ending point. When the peak width is not less than the preset minimum width threshold, the candidate peak is determined to meet the peak width requirement. The allowable amplitude of the continuous rising and falling segments of the candidate peak does not exceed the noise reference amplitude. When the proportion of sampling points that satisfy the overall upward trend in the sampling points before the time of the candidate peak reaches a preset proportion, and the proportion of sampling points that satisfy the overall downward trend in the sampling points after the time of the candidate peak reaches a preset proportion, it is determined that the candidate peak has a complete rising segment and a complete falling segment. Candidate peaks that simultaneously meet the requirements for peak height and fallback, peak width, and complete rising and falling segments are identified as independent valid peaks. The independent effective peak containing the sampling position corresponding to the global maximum current value is determined as the first effective peak, and the independent effective peak adjacent to the first effective peak and separated by a valley is selected as the second effective peak in the time before and time after the first effective peak.

3. The method for evaluating the durability of an automobile starter motor according to claim 2, characterized in that, Also includes: When there is a lowest valley point between two adjacent candidate vertices, and the current drop of the lowest valley point relative to the candidate vertex with the smaller current value among the two adjacent candidate vertices is not less than the adaptive determination amplitude, it is determined that a valley separation is formed between the two adjacent candidate vertices. When no valley separation is formed between two adjacent candidate vertices, or when the sampling interval between two adjacent candidate vertices is less than the preset minimum interval, it is determined that the two adjacent candidate vertices belong to the top local fluctuation, rising segment local fluctuation or falling segment local fluctuation on the same effective peak, and the two adjacent candidate vertices are merged into the same effective peak, and the candidate vertex with the larger current value is retained as the peak position of the same effective peak.

4. The method for evaluating the durability of an automobile starter motor according to claim 1, characterized in that, Using the sampling position of the reference peak as a reference, a preset number of current sampling points are continuously intercepted to obtain steady-state current data, including: Using the sampling position of the reference peak as the reference point, and taking it as the steady-state starting position, a preset number of current sampling points are continuously intercepted from the steady-state starting position. The preset number of current sampling points includes another valid peak, so as to obtain the steady-state current data after entering the stable working stage.

5. The method for evaluating the durability of an automobile starter motor according to claim 1, characterized in that, The slope, intercept, and goodness of fit are calculated using the following formulas: ； ； ； ； Where U represents the steady-state voltage, I represents the steady-state current, k represents the linear fitting slope, b represents the linear fitting intercept, and n represents the number of sampling points within the steady-state interval. This represents the i-th steady-state current sample value. This represents the i-th steady-state voltage sample value. This represents the average value of the steady-state current samples. This represents the average value of the steady-state voltage samples. R² represents the i-th predicted voltage value calculated based on the linear fitting equation, and R² represents the goodness of fit.

6. The method for evaluating the durability of an automobile starter motor according to claim 1, characterized in that, Based on the goodness-of-fit R², the stability of the starter motor's steady-state operation and the overall health of the machine are evaluated, including: When R² ≥ 0.91, it is judged to be in a state of excellent voltage-current linearity; When 0.85 ≤ R² < 0.91, it is considered a normal fluctuation state; When 0.75 ≤ R² < 0.85, it is judged as a slight natural aging state; When 0.65≤R²<0.75, it is determined that the starter motor has significant internal wear and requires close monitoring. When R² < 0.65, it is determined that the linear relationship between voltage and current is disordered and there is an imminent risk of damage.

7. A durability evaluation system for automotive starters, characterized in that, include: The data acquisition module is used to acquire voltage timing data and current timing data corresponding to the durability test of the automobile starter. The voltage timing data and the current timing data are both raw sampled data and have the same time reference. A current timing curve acquisition module is used to generate a current timing curve based on the current timing data. The global maximum current value determination module is used to traverse the current timing curves to determine the global maximum current value; The effective peak acquisition module is used to determine a first effective peak containing the sampling position corresponding to the global maximum current value, using the sampling position corresponding to the global maximum current value as the retrieval anchor point, and to retrieve a second effective peak adjacent to the first effective peak and separated by a valley in the time before and time after the first effective peak, wherein the first effective peak and the second effective peak have a continuous rising segment, a peak apex and a continuous falling segment; The reference peak acquisition module is used to determine the earlier time-series peak between the first valid peak and the second valid peak as the reference peak; The data interception module is used to continuously intercept a preset number of current sampling points based on the sampling position of the reference peak to obtain steady-state current data. The data correspondence module is used to synchronously obtain steady-state voltage data from the voltage time series data according to the same sampling sequence number and original timestamp as the steady-state current data, so that the steady-state current data and the steady-state voltage data correspond one-to-one; The linear fitting module is used to construct voltage-current scatter data with the steady-state current data as the horizontal axis and the steady-state voltage data as the vertical axis, and to perform linear fitting on the voltage-current scatter data using the least squares method to obtain the linear fitting equation and the goodness of fit R². The evaluation module is used to evaluate the smoothness of the starter's steady-state operation and the overall health of the machine based on the goodness of fit.

Images