A method and system for determining vehicle headlight circuit anomalies based on ADC feedback

By filtering and limiting the ADC sampling sequence of the vehicle lamp circuit, extracting feature parameters and calculating fault scores, the problem of insufficient stability of traditional vehicle lamp circuit feedback signals in complex environments is solved, and the accuracy and consistency of anomaly detection are improved.

CN123066641APending Publication Date: 2026-11-10CIXI GRAND UNION AUTO PARTS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611579273.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-13
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Traditional vehicle lighting circuit feedback signals are easily affected by periodic disturbances, spike noise, and sampling fluctuations in complex environments, resulting in insufficient stability in anomaly identification and affecting the accuracy and consistency of lighting circuit anomaly judgment.

Method used

By acquiring the ADC sampling sequence of the vehicle headlight circuit, filtering and limiting are performed, and feature parameters such as window average, average deviation, normalized window standard deviation, peak density and window change rate are extracted. Combined with the comprehensive interference amount and interference level status, the initial fault score is calculated and corrected, and finally the anomaly type is determined within the continuous analysis window.

Benefits of technology

It improves the accuracy and consistency of vehicle light path anomaly detection, reduces the impact of isolated spikes and abnormal jumps on the detection, and enhances the reliability of feedback reference quantities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123066641A_ABST
    Figure CN123066641A_ABST
Patent Text Reader

Abstract

This invention relates to the field of automotive electronics and fault detection technology, specifically to a method and system for determining abnormalities in vehicle headlight circuits based on ADC feedback. The method includes: acquiring an ADC sampling sequence and buffering it according to an analysis window; performing median filtering and amplitude limiting on the sampling sequence to obtain a preprocessed sequence; updating a baseline value based on the preprocessed sequence, the interference level value of the previous analysis window, and the average absolute value of the difference between adjacent preprocessed values ​​within the current analysis window; calculating the window average, average deviation, normalized window standard deviation, peak density, window change rate, and comprehensive interference quantity to determine the interference level status; calculating initial fault scores for open circuit, short circuit, and poor contact respectively, and correcting them based on the compensation coefficients corresponding to the comprehensive interference quantity and interference level status; accumulating the corrected fault scores within a continuous analysis window, and determining the corresponding abnormality type when a preset confirmation threshold is reached. This invention can improve the anti-interference capability and accuracy of determining abnormalities in vehicle headlight circuits.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of automotive electronics and fault detection technology, specifically to a method and system for determining abnormalities in vehicle headlight circuits based on ADC feedback. Background Technology

[0002] Determining abnormalities in vehicle lighting circuits is a common technical requirement in vehicle lighting control. It typically requires obtaining feedback signals while the lights are on and judging whether there are open circuits, short circuits, or contact abnormalities in the lighting circuit based on the feedback amplitude or threshold comparison results. It has the characteristics of being relatively straightforward to implement and easy to deploy in the vehicle body control system.

[0003] As the complexity of vehicle electrical systems increases, the feedback signals of vehicle headlight circuits are easily affected by periodic disturbances, spike noise, and sampling fluctuations under actual working conditions. Traditional judgment methods are still prone to insufficient stability in anomaly identification due to their high dependence on instantaneous sampling values ​​or single statistics when facing different interference environments, which in turn affects the accuracy and consistency of headlight circuit anomaly judgment. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for determining vehicle headlight circuit anomalies based on ADC feedback, avoiding insufficient stability in anomaly identification due to reliance on instantaneous sampling values ​​or single statistical quantities, thereby improving the accuracy and consistency of determining vehicle headlight anomaly types in complex interference environments; the purpose of this invention can be achieved through the following technical solutions:

[0005] A method for determining vehicle headlight circuit anomalies based on ADC feedback includes:

[0006] The sampling circuit, which is electrically connected to the vehicle headlight circuit, acquires the ADC sampling sequence of the headlight circuit when it is lit, and caches it according to the analysis window.

[0007] The ADC sampling sequence is filtered and limited to obtain a preprocessed sequence, and the baseline value is updated based on the average absolute value of the difference between the interference level value of the previous analysis window and the difference between adjacent preprocessed values ​​in the current analysis window.

[0008] Based on the preprocessed sequence and the benchmark value, feature parameters including window mean, average deviation, normalized window standard deviation, peak density and window change rate are extracted, and the comprehensive interference amount is extracted based on the preset target interference frequency.

[0009] The interference level is determined based on the comprehensive interference quantity, normalized window standard deviation, and peak density.

[0010] For open circuit, short circuit and poor contact anomaly types respectively, the initial fault score is calculated based on the characteristic parameters, and the initial fault score is corrected by the compensation coefficient corresponding to the comprehensive interference quantity and interference level state to obtain the corrected fault score.

[0011] Within a continuous analysis window, the fault correction score is accumulated. When the accumulated value reaches the preset confirmation threshold corresponding to the interference level, the corresponding abnormality type of the vehicle light circuit is determined.

[0012] Optionally, the ADC sampling sequence is filtered and limited, specifically including:

[0013] The median value is taken as the median sample value from multiple adjacent ADC sample values.

[0014] The absolute value of the difference between the median sample value and the previous preprocessed value is limited to a preset maximum allowable range of change, and this value is used as the current preprocessed value.

[0015] Optionally, the baseline value is updated, specifically including:

[0016] When the vehicle headlight circuit does not currently output an anomaly type, and the ratio of the current analysis window's average value to the previous analysis window's baseline value is within a preset normal range, the current baseline value is updated according to the following formula:

[0017]

[0018]

[0019] in, The sampling point number, This is the current baseline value. The previous baseline value, This is the current preprocessed value. As the baseline update coefficient, Based on the updated coefficients, This is the interference level value from the previous analysis window. This is the average of the absolute values ​​of the differences between adjacent preprocessed values ​​within the current analysis window. correspond The weighting coefficients, correspond Weighting coefficients;

[0020] like Then let ;

[0021] like Then let ;in, To update the lower bound of the coefficients, To update the upper limit of the coefficients, and .

[0022] Optionally, feature parameters are extracted, specifically including:

[0023] Calculate the average value of the preprocessed values ​​within the current analysis window as the window average;

[0024] The absolute value of the difference between the window average and the baseline value of the current analysis window is used as the average deviation.

[0025] Calculate the window standard deviation of the preprocessed sequence. When the baseline value is greater than the preset lower limit, divide the window standard deviation by the baseline value to obtain the normalized window standard deviation.

[0026] The frequency at which the absolute value of the difference between adjacent preprocessed values ​​is greater than a preset peak detection threshold is used as the peak density.

[0027] The ratio of the difference between the first and last preprocessed values ​​within the current analysis window to the duration of the current analysis window is calculated as the window change rate.

[0028] Optionally, the overall interference quantity is extracted, specifically including:

[0029] The preprocessed sequence is cross-correlated based on the target interference frequency to obtain the periodic interference amplitude.

[0030] The periodic interference amplitude is dimensionless based on the window standard deviation of the preprocessed sequence, and the maximum value is taken as the comprehensive interference quantity when there are multiple target interference frequencies.

[0031] Optionally, the interference level status is determined, specifically including:

[0032] The average values ​​of the comprehensive interference, normalized window standard deviation, and peak density of the continuous analysis window are normalized and weighted summed to obtain the interference level value.

[0033] Hysteresis comparison logic is used to compare the interference level value with the preset first environmental threshold and second environmental threshold to determine whether the current interference level is low, medium or high.

[0034] Optionally, the initial fault score is calculated and corrected, specifically including:

[0035] When the ratio of the window average value to the baseline value is less than or equal to the preset open circuit judgment threshold, the initial fault score of the open circuit is calculated based on the average deviation.

[0036] When the illumination time of the vehicle headlight circuit exceeds the preset start-up stabilization waiting time and the ratio is greater than or equal to the preset short circuit judgment threshold, the initial fault score of the short circuit is calculated based on the window change rate.

[0037] The initial fault score for poor contact is calculated based on the weighted sum of the normalized window standard deviation, peak density, and window change rate.

[0038] For exception types ,according to Correct the initial fault score, if Then let ;

[0039] in, Exception type Initial fault score, To correct the fault rating, The interference level is classified as low, medium, or high. Exception type In the interference level state The interference compensation coefficient is greater than zero. This is the converted value of the comprehensive interference quantity after being adjusted according to the upper limit of the scoring range.

[0040] Optionally, determine the type of anomaly in the vehicle headlight circuit, specifically including:

[0041] Within the continuous analysis window, the fault correction score is integrated and accumulated based on a preset attenuation coefficient, and the accumulated value is limited to a valid range.

[0042] An anomaly is confirmed when the cumulative value is greater than or equal to the product of the preset basic confirmation threshold and the preset environmental compensation coefficient.

[0043] Fast short circuit confirmation is performed when the ratio of the window average value to the baseline value is consistently greater than or equal to the preset severe short circuit threshold.

[0044] The recovery score is calculated based on the average deviation, the normalized window standard deviation, and the overall disturbance.

[0045] Once an anomaly is confirmed, if the recovery scores of multiple consecutive analysis windows meet the preset recovery threshold and the cumulative value drops to zero, the anomaly is resolved.

[0046] A vehicle headlight circuit anomaly detection system based on ADC feedback includes:

[0047] The sampling circuit is electrically connected to the vehicle lamp path and is used to acquire the ADC sampling sequence of the vehicle lamp path when it is lit.

[0048] The acquisition module is used to read the ADC sampling sequence and cache it according to the analysis window;

[0049] The preprocessing module is used to perform median filtering and amplitude limiting on the ADC sampling sequence to obtain a preprocessed sequence. The baseline value is updated based on the preprocessed sequence, the interference level value of the previous analysis window, and the average absolute value of the difference between adjacent preprocessed values ​​in the current analysis window.

[0050] The feature extraction module is used to extract feature parameters including window mean, average deviation, normalized window standard deviation, peak density and window change rate based on the preprocessed sequence and benchmark value, and extract the comprehensive interference amount based on the target interference frequency.

[0051] The evaluation module is used to determine the interference level status based on the comprehensive interference amount, normalized window standard deviation, and peak density.

[0052] The scoring module calculates the initial fault score for open circuit, short circuit and poor contact anomaly types respectively, and corrects it based on the compensation coefficient corresponding to the comprehensive interference amount and interference level status to obtain the corrected fault score.

[0053] The judgment module is used to accumulate the fault correction score within a continuous analysis window. When the accumulated value is greater than or equal to the preset confirmation threshold corresponding to the interference level, the abnormality type of the vehicle light circuit is determined.

[0054] Compared with the prior art, the present invention has the following technical effects:

[0055] This invention acquires and caches ADC sampling sequences by analyzing windows while the vehicle lights are on. Combined with median filtering and amplitude limiting, it effectively reduces the impact of isolated spikes and abnormal jumps on subsequent judgments. This ensures that the sampled data has stability that meets preset standards before entering the analysis link. Furthermore, it combines a benchmark recursion mechanism that only allows updates when there are no output anomalies and the average value of the window is within a preset normal range. The benchmark following speed is adjusted based on the interference level value of the previous analysis window and the absolute average of the differences between adjacent preprocessed values ​​in the current window. This avoids the reverse impact of interference enhancement or anomaly maintenance phases on the normal benchmark, thereby improving the reliability of the feedback reference. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort:

[0057] Figure 1 The flowchart of a method for determining abnormalities in vehicle headlight circuits based on ADC feedback according to an embodiment of the present disclosure is shown.

[0058] Figure 2 This is a structural diagram of a vehicle headlight circuit anomaly determination system based on ADC feedback, according to an embodiment.

[0059] Figure 3 This is a schematic diagram of the applicable lamp path and ADC feedback sampling circuit in the embodiment;

[0060] Figure 4 This is a state transition diagram of anomaly confirmation, maintenance, and recovery in an embodiment;

[0061] Figure 5 This is a window-by-window decision timing diagram under typical operating conditions of the embodiment. Detailed Implementation

[0062] To make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments will be described in detail below with reference to the accompanying drawings.

[0063] Please see Figure 1 A method for determining abnormalities in vehicle headlight circuits based on ADC feedback, comprising:

[0064] The sampling circuit, which is electrically connected to the vehicle headlight circuit, acquires the ADC sampling sequence of the headlight circuit when it is lit, and caches it according to the analysis window.

[0065] The ADC sampling sequence is filtered and limited to obtain a preprocessed sequence, and the baseline value is updated based on the average absolute value of the difference between the interference level value of the previous analysis window and the difference between adjacent preprocessed values ​​in the current analysis window.

[0066] Based on the preprocessed sequence and the benchmark value, feature parameters including window mean, average deviation, normalized window standard deviation, peak density and window change rate are extracted, and the comprehensive interference amount is extracted based on the preset target interference frequency.

[0067] The interference level is determined based on the comprehensive interference quantity, normalized window standard deviation, and peak density.

[0068] For open circuit, short circuit and poor contact anomaly types respectively, the initial fault score is calculated based on the characteristic parameters, and the initial fault score is corrected by the compensation coefficient corresponding to the comprehensive interference quantity and interference level state to obtain the corrected fault score.

[0069] The fault correction score is accumulated within a continuous analysis window. When the accumulated value reaches the preset confirmation threshold corresponding to the interference level, the corresponding abnormality type of the vehicle light circuit is determined.

[0070] The ADC sampling sequence is filtered and limited, specifically including:

[0071] The median value is taken as the median sample value from multiple adjacent ADC sample values.

[0072] The absolute value of the difference between the median sample value and the previous preprocessed value is limited to a preset maximum allowable range of change, and this value is used as the current preprocessed value.

[0073] Updating the baseline value specifically includes:

[0074] When the vehicle headlight circuit does not currently output an anomaly type, and the ratio of the current analysis window's average value to the previous analysis window's baseline value is within a preset normal range, the current baseline value is updated according to the following formula:

[0075]

[0076]

[0077] in, The sampling point number, This is the current baseline value. The previous baseline value, This is the current preprocessed value. As the baseline update coefficient, Based on the updated coefficients, This is the interference level value from the previous analysis window. This is the average of the absolute values ​​of the differences between adjacent preprocessed values ​​within the current analysis window. correspond The weighting coefficients, correspond Weighting coefficients;

[0078] like Then let ;

[0079] like Then let ;

[0080] in, To update the lower bound of the coefficients, To update the upper limit of the coefficients, and .

[0081] When the headlight branch is under the illumination command, the body controller periodically reads the ADC code value output by the feedback sampling circuit and organizes the continuous sampling results into an analysis window of fixed length. This window is used to perform preprocessing, benchmark update and anomaly detection. The detection process is performed in units of analysis windows.

[0082] As a preferred option, such as Figure 3As shown, the sampling circuit includes a sampling resistor and a current detection amplification branch connected in series between the vehicle lamp path and the driver, or a voltage divider sampling branch connected in parallel across the lamp load. The physical object sampled by the ADC is the analog feedback signal converted by the conditioning circuit. The conditioning circuit includes filtering, gain amplification, and level shifting in sequence. The ADC uses a fixed reference voltage VREF as the full-scale reference, and the sampling range is 0 to VREF. The feedback signal is mapped into the sampling range after being conditioned by the gain G. Under normal lighting conditions, the code value of the ADC sampling sequence is within the normal value range calibrated by the normal lighting reference value and the preset fluctuation width. Open circuit of the lamp path, short circuit of the lamp load to the power supply, short circuit of the lamp load to ground, and poor contact of the lamp path are all within the coverage of the sampling circuit and the ADC feedback. When the lamp is open, the feedback signal drops away from the reference value. The two types of short circuit feedback signals are opposite in direction, and after the drive protection is activated, the feedback signal is clamped to the protection level consistent with the corresponding fault type. Based on this, the polarity coefficient s is pre-calibrated to unify the direction of short circuit scoring and open circuit judgment.

[0083] The preferred analysis window length is one that can cover at least one complete target interference period and include no less than 8 valid sampling points; when the sampling frequency is adjusted, the window length is adjusted accordingly to maintain the comparability of statistics and correlations within the window;

[0084] Furthermore, the preferred method for caching by analysis window is a circular caching method that advances unidirectionally with the sampling clock: when a new sampling point enters, it is written to the current writing position and overwrites the earliest released sampling position; when the cumulative number of nominal samples required for an analysis window is reached, the continuous sample segment is frozen as a window to be analyzed. The frozen samples will no longer be overwritten by subsequent sampling until the window is preprocessed, feature extracted and scored before being released as a whole.

[0085] If a timestamp is missing, a lighting command is canceled, or the sampling interval exceeds the preset step loss limit during sampling, the window currently being formed will not enter the subsequent abnormal judgment, and will be marked as an incomplete window and released directly. At the same time, a new window will be formed again from the next valid sampling point.

[0086] Therefore, the status of the sampled data in the system can be divided into at least four categories: writing, pending analysis, low confidence, and completed release. The subsequent control flow determines the permission of the corresponding window to enter the interference level link and the fault accumulation link based on the status.

[0087] After sampling enters the preprocessing stage, the median of three adjacent sampled values ​​is taken first to eliminate the influence of isolated spikes on subsequent calculations; the difference between the median sampled value and the previous preprocessed value is taken and the absolute value is compared with the preset maximum allowable change;

[0088] If the change exceeds the upper limit, the median sampled value is no longer used, and the current preprocessed value is advanced along the direction of change by a limited step size equal to the maximum allowable change. hour:

[0089]

[0090] in, This is the current median sample value. This is the previous preprocessed value. The maximum allowable variation is set at the median sample value. If the upper limit is not exceeded, the median sample value is used directly. If there is no previous preprocessed value at the beginning of the analysis window, the first median sample value in the window can be used as the first preprocessed value, and subsequent sample points are processed according to the above amplitude limit rules.

[0091] After the above processing, abnormal jumps in the sampling sequence will not be carried into the window statistics as is, while the actual changes below the preset rate of change can still be retained; the maximum allowable change is preferably determined according to the statistical upper bound of the difference between adjacent samples under normal lighting and fault-free conditions. Specifically, it can be taken as the high quantile value of the distribution of the absolute values ​​of adjacent differences within the most recent consecutive effective lighting window, and a lower limit of not less than 1 ADC code value is retained to take into account the change range of different lamp types in the stable and transition sections.

[0092] Furthermore, preprocessing is preferably performed point by point in chronological order within the window, and only the previous preprocessed value required for the current point is retained, without rewriting the old points that have been processed; if more than a preset number of restricted step size replacement points appear consecutively in the same window, an abnormal fluctuation mark is added to the window, but it is still allowed to continue to complete feature extraction so that subsequent diversion judgment can be made by interference level or fault scoring link, rather than directly discarding the entire window of data in the preprocessing stage;

[0093] Regarding the update of the baseline value, this implementation method only allows the current baseline value to be updated when the light path does not output an abnormal type and the average value of the current analysis window is still within the preset normal range relative to the baseline value of the previous analysis window.

[0094] The window average is formed by the arithmetic mean of all valid preprocessed values ​​in the current analysis window to form a temporary statistical result. This temporary statistical result is only used to determine whether the baseline value can be updated. If the update is allowed, the baseline recursion continues, and the same arithmetic mean result is used as the formal window average for subsequent fault scoring and environmental determination after the window ends, so that another set of mean objects will not be generated again.

[0095] If the baseline value of the previous analysis window is not greater than the preset extremely low baseline threshold, or if the corresponding light path has just entered the lighting state and has not yet formed a usable reference, the baseline value saved at the end of the last effective lighting cycle will be used first; if there is no saved effective baseline value, the normal lighting baseline value that has been pre-calibrated and stored for the current light type and light path will be used.

[0096] The current window average value will only be used as the initialization benchmark when neither of the above two types of benchmark values ​​is available and the current window average value is within the pre-calibrated range of normal lighting absolute values.

[0097] If the current average value of the window is not within the range of the normal lighting absolute value, then the window will not be used as the initial reference, and the window will be treated as an abnormal candidate window to avoid forming an incorrect reference with the feedback value in the open circuit or short circuit state.

[0098] The benchmark value used for ratio calculation should be greater than the preset lower limit for ratio calculation, which is a fixed value greater than zero. If the benchmark value is not greater than the lower limit, the ratio calculation between the window average and the benchmark value is not performed, nor is the benchmark value updated or the ratio judgment condition triggered based on the ratio. Instead, the valid benchmark value is obtained first in the above order. As a result, all comparisons of the ratio between the window average and the benchmark value have non-zero and stable denominators.

[0099] The update process uses a weighted combination of the previous baseline value and the current preprocessed value. The weighting coefficient is determined by the average absolute value of the difference between the interference level value of the previous analysis window and the difference between adjacent preprocessed values ​​in the current analysis window. If the calculated baseline update coefficient exceeds the upper and lower limits, it is truncated according to the upper and lower limits before participating in the update. This process ensures that the baseline value follows slowly only within the range of normal illumination and acceptable fluctuations, avoiding being skewed when interference increases.

[0100] Furthermore, Used to indicate the update sequence number of the preprocessed value participating in the baseline recursion within the current analysis window; after an analysis window ends, it is used to indicate the result of one recursion within that window. This serves as the current baseline value for the current analysis window and is used as the previous baseline value at the start of the next analysis window; correspondingly... At the window level, the interference level value that has been determined in the previous analysis window is taken, so that the reference of the sampling point level recursive quantity and the window level state quantity are consistent;

[0101] For the first valid analysis window that does not yet have a previous interference level value, The initial value of the low interference level is taken as a pre-calibrated value. This initial value and the subsequent interference level values ​​use the same dimension and are used in a fixed manner under the same lamp type or the same lamp circuit configuration.

[0102] The normal range of the ratio of the window average value to the baseline value of the previous analysis window is calibrated using normal lighting samples, and the allowable proportion range is taken after covering the normal lamp path brightness drift and temperature rise drift. Take a constant between 0 and 1. and Take a non-negative value to characterize the degree to which the reference following speed is slowed down when the interference is enhanced, and calibrate according to the reference drift speed of the target light path in the stable section;

[0103] Keep it as a dimensionless quantity. It is preferable to use the dimensionless interference level value calibrated using the same dimensional system. It is preferable to first normalize the constants according to the upper bound of the differential reference corresponding to the current lamp type or current lamp circuit configuration before substituting them in.

[0104] Optionally, Calibrated as The coefficients that match the reciprocal dimensions make It is a dimensionless quantity; the boundary conditions of the benchmark update coefficients satisfy... This ensures that the truncated baseline recursion is always an effective weighted combination between the previous baseline value and the current preprocessed value;

[0105] Furthermore, the average absolute value of the difference between adjacent preprocessed values ​​within the current analysis window is calculated in a fixed order after the window is formed: first, the absolute value of the difference between each adjacent preprocessed value is calculated in chronological order, and then the arithmetic mean of these absolute values ​​is taken; if there are fewer than two valid preprocessed values ​​within the window, the average value is treated as zero, and the result is set to zero. Only by and Decide;

[0106] The preferred execution order for baseline updates is as follows: first, complete the full window preprocessing, then calculate the average value of the temporary window and the average value of the absolute values ​​of adjacent differences, then determine whether the normal range conditions are met, and perform baseline recursion on each preprocessed value in the window in sequence; if the normal range conditions are not met, the current analysis window directly uses the previous baseline value as the current baseline value, and does not perform recursion.

[0107] Furthermore, to distinguish between benchmark following and slow failures, a benchmark following speed limit is pre-calibrated. According to the formula Calculation, where The number of valid sampling points within an analysis window. The unit is ADC code value per analysis window; when the relative deviation of the window average value relative to the baseline value of the previous analysis window in multiple consecutive analysis windows continuously exceeds the preset upper limit of the normal range, and the relative deviation increases with the window size by more than [amount missing], the analysis window is considered to be in a state of flux. When a slow fault candidate window is identified, the baseline update is frozen and the fault scoring process begins directly; thus, faults below this window are considered slow fault candidates. Brightness drift and temperature rise drift are normally followed by the reference and are not judged as faults; drift speed is greater than The persistent faults will not be masked by the benchmark;

[0108]

[0109] Furthermore, Within a single analysis window, a fixed value is taken, that is, the absolute value of the difference between all adjacent preprocessed values ​​in that window is averaged, and then each value within the window is... Used uniformly; therefore, Follow The change only reflects the current recursive step. Enter update mode instead of importing into the same window. Ambiguity caused by repeated changes;

[0110] Furthermore, if the current window has been marked as a low-confidence window, an incomplete window, or the current light path is in the abnormal confirmation and holding phase, the benchmark update link is directly closed, and the window only inherits the previous benchmark value to enter the subsequent feature calculation; benchmark recursion is only allowed again when the window state is restored to validity and the abnormal confirmation is lifted; this can avoid the abnormal feedback during the fault holding period from having a reverse effect on the normal benchmark.

[0111] The statistical features within the window are extracted sequentially; the window mean is used to characterize the DC feedback level of the window; the mean deviation is used to characterize the degree of deviation of the window mean from the benchmark value; and the normalized window standard deviation is used to characterize the relative relationship between the fluctuation intensity and the benchmark amplitude.

[0112] Peak density is obtained by counting the number of times the difference between adjacent preprocessed values ​​exceeds the peak detection threshold; window change rate is obtained by dividing the difference between the preprocessed values ​​at the beginning and end of the window by the window duration; these quantities are collectively included in the subsequent calculation of interference and fault score, so that fluctuation information that is not covered by a single mean can be included in the judgment.

[0113] The peak detection threshold is preferably strictly less than the maximum allowable change and higher than the upper quantile boundary of the absolute value distribution of the difference between adjacent preprocessed values ​​in the normal lighting window. This ensures that the ability to distinguish abnormal rapid changes is retained after the amplitude limiting process, and that general noise fluctuations do not pass through the threshold, while local jumps close to the maximum allowable change can be included in the peak statistics.

[0114] The extraction of the comprehensive interference quantity is based on the preset target interference frequency. The system performs correlation operations on the preprocessed sequence to obtain the sinusoidal correlation component and the cosine correlation component, and then synthesizes the periodic interference intensity at the target frequency. This intensity is then combined with the window standard deviation for dimensionless processing to form the comprehensive interference quantity.

[0115] If there are multiple target interference frequencies, the periodic interference quantity corresponding to each frequency is calculated separately, and the larger value is taken as the current comprehensive interference quantity. After the above processing, the periodic disturbance under the known interference frequency will not be mistaken for the actual change in the state of the lamp path. The interference level value is composed of the comprehensive interference quantity, the normalized window standard deviation and the peak density, and the interference level state is determined by comparing it with the environmental threshold.

[0116] To avoid frequent switching of environmental levels near the threshold, the comparison process adopts a state-holding method with hysteresis: when the interference level value is below the low threshold, it is determined to be a low interference level; when it is above the high threshold, it is determined to be a high interference level; and when it is between the two, it is determined to be a medium interference level. Hysteresis intervals are set at each threshold boundary to retain the previous state. This state participates in the fault scoring correction to keep the judgment intensity consistent under different environments.

[0117] For three types of anomalies—open circuit, short circuit, and poor contact—the system generates an initial fault score for each. The open circuit type mainly depends on the deviation of the window average value from the baseline value; the short circuit type combines the window change rate and the stabilization time after the light circuit is started; and the poor contact type uses a weighted combination of the normalized window standard deviation, peak density, and window change rate.

[0118] After the initial fault score is formed, it is corrected according to the comprehensive interference amount and the compensation coefficient corresponding to the current interference level. If the correction result is less than zero, it is set to zero directly to avoid negative scores from being included in the accumulation.

[0119] The corrected fault score is not directly used as the final fault result, but is included in the cumulative process of continuous windows. After each analysis window ends, the current cumulative value is added to the corrected fault score and a preset attenuation amount is deducted. At the same time, the cumulative value is limited to between zero and the upper limit of the integral.

[0120] The corresponding abnormality type is only output when the cumulative value reaches or exceeds the preset confirmation threshold corresponding to the interference level. This limits the direct triggering of fault conclusions by single-window disturbances, and determines a continuous abnormality when the cumulative value reaches the confirmation threshold.

[0121] As a preferred option, feature parameters are extracted, specifically including:

[0122] Calculate the average value of the preprocessed values ​​within the current analysis window as the window average;

[0123] The absolute value of the difference between the window average and the baseline value of the current analysis window is used as the average deviation.

[0124] Calculate the window standard deviation of the preprocessed sequence. When the baseline value is greater than the preset lower limit, divide the window standard deviation by the baseline value to obtain the normalized window standard deviation.

[0125] The frequency at which the absolute value of the difference between adjacent preprocessed values ​​is greater than a preset peak detection threshold is used as the peak density.

[0126] The ratio of the difference between the first and last preprocessed values ​​within the current analysis window to the duration of the current analysis window is used as the window change rate.

[0127] Extracting the overall interference, specifically including:

[0128] The preprocessed sequence is cross-correlated based on the target interference frequency to obtain the periodic interference amplitude.

[0129] The periodic interference amplitude is dimensionless based on the window standard deviation of the preprocessed sequence, and the maximum value is taken as the comprehensive interference quantity when there are multiple target interference frequencies.

[0130] Determine the interference level status, specifically including:

[0131] The average values ​​of the comprehensive interference, normalized window standard deviation, and peak density of the continuous analysis window are normalized and weighted summed to obtain the interference level value.

[0132] Hysteresis comparison logic is used to compare the interference level value with the preset first environmental threshold and second environmental threshold to determine whether the current interference level is low, medium or high.

[0133] When the feedback signal of the lamp path is under the background of electromagnetic disturbance exceeding the preset interference threshold, relying solely on the window average value can easily misinterpret periodic fluctuations as real state changes. Therefore, this implementation separates the window standard deviation, peak density and target frequency correlation quantities for processing, and then synthesizes them at the window level; continuous disturbances, discrete peaks and periodic components are judged according to different statistical quantities.

[0134] When calculating window features, the system first obtains the average value of the window from the preprocessed sequence and uses the difference between it and the benchmark value to form the average deviation. The window standard deviation is then normalized when the benchmark value is greater than the preset lower limit to prevent proportional fluctuations exceeding the preset fluctuation threshold at low benchmark levels. If the benchmark value is not higher than the lower limit, the window standard deviation is used directly to avoid the result being distorted due to an excessively small denominator.

[0135] Peak density is calculated by counting the number of times the difference exceeds the threshold and then adjusting it according to the number of sampling intervals; the window change rate is obtained by dividing the difference between the preprocessed values ​​at the beginning and end of the window by the duration of the analysis window, and is used to reflect the directional change within the window.

[0136] Calculate the difference between the last preprocessed value and the first preprocessed value within the current analysis window, then divide by the window duration to obtain the window change rate, i.e.:

[0137]

[0138] in, For window change rate, This is the preprocessed value at the end of the window. This is the first preprocessed value. The duration of the window;

[0139] Since the statistical object of the peak density is the difference of the preprocessed sequence after the amplitude limit is applied, the preset peak detection threshold is preferably set between the upper quantile boundary of the absolute value distribution of the normal lighting difference and the maximum allowable change, and is strictly less than the maximum allowable change, so as to ensure that the peak statistics still have non-zero discrimination after the amplitude limit is applied.

[0140] If an analysis window contains only one valid preprocessed value, resulting in zero sampling intervals or insufficient window duration to support the calculation of the rate of change, then the peak density of the window is set to zero, the window rate of change is set to zero, and the window is marked as a low confidence window to avoid division by zero or amplification of the rate of change by an extremely short duration.

[0141] The preset lower limit is preferably determined based on the lower bound of the baseline value of the stable lighting segment, which is used to distinguish between windows suitable for proportional normalization and windows suitable only for absolute quantity statistics; the analysis window duration is preferably the difference between the timestamps corresponding to the first and last valid sampling points within the window. If the difference is less than one sampling period, it is treated as insufficient to support the calculation of the rate of change.

[0142] Furthermore, the number of sampling intervals is the number of valid preprocessed values ​​minus one; when the number of valid preprocessed values ​​in the window is not less than two, the peak density is calculated by dividing the number of threshold differences by the number of sampling intervals, and the result is limited to between zero and one.

[0143] When there is only one valid preprocessed value in the window, the window mean is taken as the unique valid preprocessed value, the window standard deviation is treated as zero, the peak density and the window rate of change are both treated as zero, and the window is kept in a low confidence mark; when there is zero valid preprocessed value in the window, the window is directly marked as an invalid window and discarded.

[0144] Furthermore, the preferred calculation order for window features is as follows: first, determine the number of valid preprocessed values, then calculate the window average, then calculate the window standard deviation and its normalization result, then count the number of times the threshold difference is exceeded to obtain the peak density, and calculate the window change rate; if the window is marked as a low confidence window, the window average and window standard deviation that can be directly obtained are still retained for the fault scoring link reference, but this window is not used as a new sample to regenerate the current interference level value;

[0145] Furthermore, if there are local sampling omissions within the window but the number of valid preprocessed values ​​still meets the minimum calculation requirements, the missing points will not be reconstructed by interpolation. Instead, the window features will be calculated directly based on the remaining valid points, and a downgraded calculation mark will be added to the window.

[0146] A window with this mark is allowed to enter the fault scoring chain, but it is only allowed to indirectly affect the environmental level switching when its subsequent valid window arrives and is consistent with the aforementioned statistical trend, thereby avoiding additional fluctuations introduced by temporary sampling or forced interpolation.

[0147] The extraction of the comprehensive interference quantity adopts correlation operation matching the target interference frequency; after the preprocessed sequence is correlated with the sine and cosine reference sequences respectively, two orthogonal components are obtained, which are then synthesized into the periodic interference amplitude, and normalized by the sum of the window standard deviation and the minimum constant to obtain the dimensionless periodic interference quantity.

[0148] If there is more than one target interference frequency, the periodic interference quantity corresponding to each frequency is calculated separately, and the maximum value is selected as the comprehensive interference quantity. After the above processing, even if the interference frequency switches between multiple known frequency points, the comprehensive interference quantity can still remain comparable.

[0149] To ensure consistency in the correlation results across different windows, the target interference frequency is preferably a frequency point that can form an integer number or a preset approximate integer number of cycles within an analysis window. When the sampling frequency or window length is adjusted, the reference sequence is reconstructed according to the same sampling benchmark, thereby avoiding abnormal increases in correlation due to window truncation.

[0150] When the window standard deviation is close to zero, the minimum constant remains a fixed positive number and is only used to avoid the normalization denominator being zero. It is not used as an independent decision quantity in subsequent interference level comparisons, thereby avoiding confusion between this constant and the comprehensive interference quantity itself.

[0151] The preferred method for constant normalization is to divide each statistic by its respective reference upper bound, so that the average value of the normalized comprehensive interference, the average value of the normalized window standard deviation, and the average value of the peak density all fall within a comparable range of 0 to 1. The corresponding reference upper bound can be jointly calibrated based on the statistical boundaries of three types of windows: normal lighting, interference enhancement without false alarm codes, and known anomalies, and should be kept fixed under the same lamp type or the same lamp path configuration.

[0152] The average value of the comprehensive interference, the average value of the normalized window standard deviation, and the average value of the peak density are divided by their respective preset reference upper bounds. If the division result is less than zero, it is treated as zero; if it is greater than one, it is treated as one. Once the reference upper bound is determined, it remains unchanged under the same lamp type or the same lamp path configuration and does not change temporarily with a single analysis window, thereby ensuring the comparability between different windows.

[0153] Furthermore, for any target interference frequency First, click the first button in the window. For each valid preprocessed value, a sine reference value and a cosine reference value are constructed corresponding to a timestamp. Then, the sine correlation components are obtained by summing them separately. Cosine-related components The calculation method is as follows:

[0154] Furthermore, before performing the above-mentioned summation, the DC component is estimated using the window average value of the current analysis window, and then... For the One effective preprocessed value is taken to DC, in the aforementioned calculation formula All after removing DC Substitute; the endpoint convention for discrete sampling is: summation is performed with the actual timestamp of each valid preprocessed value as the independent variable, according to... to Sequential accumulation, no endpoint extension, no forced resampling at equal intervals when samples are missing or jittering occurs; the denominator for amplitude normalization is the window standard deviation and the minimum constant. The sum of , The value is a fixed positive number; therefore, the correlation component of the constant input after DC removal is zero, and the correlation result will not be amplified when the window standard deviation is close to zero.

[0155]

[0156]

[0157]

[0158] in, For the first One valid preprocessed value, This is the sampling timestamp corresponding to this value. The number of valid preprocessed values ​​in the current window; by Obtain the periodic interference amplitude corresponding to the target interference frequency; then divide the amplitude by the sum of the window standard deviation and the minimum constant to obtain the periodic interference amount corresponding to the frequency.

[0159] When there are multiple target interference frequencies, they are calculated one by one in a fixed order and the maximum value is taken as the comprehensive interference quantity. When there is only a single target interference frequency, the result corresponding to that frequency is directly used.

[0160] After using the correlation components averaged by the number of valid preprocessed values, the periodic interference amplitude and window standard deviation both maintain the same dimensions as the ADC feedback, thus the periodic interference remains dimensionless and does not increase linearly due to changes in the number of valid samples under the same statistical caliber.

[0161] If there are fewer than two valid preprocessed values ​​in the window, the above-mentioned summation will not be performed, the total interference will be treated as zero and the low confidence mark of the window will be maintained.

[0162] If the number of valid preprocessed values ​​is not less than two but is lower than the lower limit of the preset nominal sample number corresponding to the current lamp type or current lamp path configuration, then the window is also marked as a low confidence window and is not used as a new sample to regenerate the current interference level value, so as to ensure the comparability of the correlation quantities between different windows;

[0163] Furthermore, all related calculations are uniformly performed using the actual timestamps corresponding to the valid preprocessed values. The analysis window duration, reference sequence construction, and target frequency matching are all based on the same time reference.

[0164] When there is jitter in the sampling but the valid sampling criteria are still met, the sample points are not forcibly resampled to the equally spaced sequence, and are directly used for correlation summation and window change rate calculation according to the corresponding timestamp, thereby avoiding time base inconsistency when using the equally spaced assumption in the correlation calculation and the actual time interval in the change rate calculation.

[0165] Furthermore, if there are two or more target frequencies within the same window whose periodic interference amounts are simultaneously within the preset maximum amplitude tolerance range, and the difference between them is less than the preset parallel tolerance, then the first one to appear will still be selected as the source of the comprehensive interference amount according to the preset order, but the remaining frequency points that are close to the maximum value will be marked for subsequent calibration reference, and the interference amounts of multiple frequency points will not be superimposed repeatedly in the current window, so as to avoid the same physical fluctuation being weighted repeatedly.

[0166] Furthermore, the lower limit of the preset nominal sample number is preferably set to 70% to 90% of the nominal sample number of the corresponding analysis window, and is fixed under the same lamp type or the same lamp path configuration; the parallel tolerance is preferably set to 1% to 5% of the current reference upper limit of the comprehensive interference amount, or a fixed value not greater than 0.05 times the reference upper limit of the single target frequency period interference amount, so as to ensure that the judgment of approaching the maximum value has a uniform scale;

[0167] The preset order of multiple target interference frequencies is preferably arranged from low to high frequency, or according to the interference frequency points that are more likely to occur in the history of the corresponding light circuit; once determined, it remains unchanged under the same light type or the same light circuit configuration; thus, the source frequency of the comprehensive interference, the judgment condition of the low confidence window, and the value rules when multiple frequency points are in parallel can all be directly implemented.

[0168] The interference level value is formed by the statistical average of multiple continuous analysis windows; the system performs constant normalization on the average value of the comprehensive interference, the average standard deviation of the normalized window, and the average peak density, and then performs weighted summation.

[0169] Two environmental thresholds are used to construct the hysteresis comparison logic: when the level is below the first environmental threshold, it is judged as low interference level; when it is above the second environmental threshold, it is judged as high interference level; when it is in the middle range, it is judged as medium interference level. A hysteresis interval is set near the first and second environmental thresholds to carry over the previous state. If the level is within the hysteresis interval during the first judgment, it is set to medium interference level. This method avoids the interference level from fluctuating near the boundary, so that the benchmark update and fault correction remain stable.

[0170] Furthermore, the comprehensive interference, normalized window standard deviation, and peak density involved in the average calculation are all taken from the corresponding values ​​of the most recent consecutive valid windows. If a window is marked as a low confidence window, the window is not used as a new sample to regenerate the current interference level value. Only the statistical link corresponding to the previous valid window is retained to avoid the ambiguity of the value objects of each statistic when the window is missing.

[0171] In the weighted summation, each weight is taken as a non-negative value, and preferably the sum of the weights is 1. The weight corresponding to the comprehensive interference is not lower than the weight corresponding to the peak density, so as to highlight the role of periodic interference consistent with the target frequency in environmental judgment.

[0172] The first and second environmental thresholds are both calibrated under the above normalized dimensions and determined according to the principle that the false alarm rate is limited under normal conditions and the false negative rate is limited under strong interference conditions. A hysteresis interval is retained between the two to suppress frequent switching.

[0173] Furthermore, among multiple consecutive analysis windows, the most recent 3 to 8 consecutive valid windows are preferred; when the number of valid windows has not yet reached the preset number, the corresponding average value is calculated based on the valid windows that have been obtained, but the predetermined upper reference limit and weight are not changed due to insufficient samples.

[0174] If a low-confidence window appears in the middle, skip the window and backtrack to an earlier valid window to make up the preset number; if the preset number is still insufficient after backtracking, continue to calculate based on the current valid window and use the previous interference level state to participate in the hysteresis comparison.

[0175] The average value of the comprehensive interference, the average value of the normalized window standard deviation, and the average value of the peak density are all calculated as the arithmetic mean of the most recently selected effective window in the statistical link. When there is only one effective window, the value corresponding to that window is taken. When the number of effective windows reaches the preset number, the earliest sample is removed and the latest effective window is added in the sliding window manner.

[0176] Furthermore, the interference level value is calculated using the following formula:

[0177]

[0178] in, This is the result after normalizing the average value of the comprehensive interference. This is the result after normalization of the normalized window standard deviation, mean, and constant number. This is the result after normalizing the average peak density. correspond The weight, correspond The weight, For the corresponding The weights;

[0179] If the calculation result is less than zero, it is treated as zero. If it is greater than the upper limit of each environmental threshold calibration, the original value is retained and the hysteresis comparison is directly entered so that the high interference window can stably fall into the high interference level.

[0180] Furthermore, the environmental level update is preferably performed only once at the end of each analysis window; if the window is not included in new samples, the current environmental level remains unchanged, and the environmental re-judgment will not be triggered again due to the subsequent results of the fault scoring link within the same window, thereby avoiding the reversal of the control order within a single window.

[0181] Furthermore, the weighted selection satisfies , , and keep The first environmental threshold and the second environmental threshold are preferably set to 0.30 to 0.45 and 0.55 to 0.75, respectively, and the difference between the two is not less than 0.10.

[0182] Under the same lamp type or the same lamp path configuration, the weights and environmental thresholds use the same set of calibration values; when switching to different lamp types or different lamp path configurations, another set of pre-stored weights and environmental thresholds is called instead of being recalculated temporarily in a single analysis window; thus, the fusion rules, level boundaries and cross-operating condition switching methods of interference level values ​​remain fixed.

[0183] If the sampling sequence of a certain analysis window has obvious missing parts or local discontinuities, then that window will not participate in the re-switching of the interference level, but will continue to use the interference level status of the previous effective window. The window average and window standard deviation of the current window will only be used as auxiliary references when correcting the fault score in the subsequent process, and the environment level will not be triggered separately based on these. This can avoid the data missing of a single window from directly disturbing the environment judgment link.

[0184] Preferably, the initial fault score is calculated and corrected, specifically including:

[0185] When the ratio of the window average value to the baseline value is less than or equal to the preset open circuit judgment threshold, the initial fault score of the open circuit is calculated based on the average deviation.

[0186] When the illumination time of the vehicle headlight circuit exceeds the preset start-up stabilization waiting time and the ratio is greater than or equal to the preset short circuit judgment threshold, the initial fault score of the short circuit is calculated based on the window change rate.

[0187] The initial fault score for poor contact is calculated based on the weighted sum of the normalized window standard deviation, peak density, and window change rate.

[0188] For exception types ,according to Correct the initial fault score, if Then let ;

[0189] in, Exception type Initial fault score, To correct the fault rating, The interference level is classified as low, medium, or high. Exception type In the interference level state The interference compensation coefficient is greater than zero. The adjusted value is the comprehensive interference amount after being converted according to the upper limit of the scoring range;

[0190] Determine the type of anomaly in the vehicle's headlight circuit, specifically including:

[0191] Within the continuous analysis window, the fault correction score is integrated and accumulated based on a preset attenuation coefficient, and the accumulated value is limited to a valid range.

[0192] An anomaly is confirmed when the cumulative value is greater than or equal to the product of the preset basic confirmation threshold and the preset environmental compensation coefficient.

[0193] Fast short circuit confirmation is performed when the ratio of the window average value to the baseline value is consistently greater than or equal to the preset severe short circuit threshold.

[0194] The recovery score is calculated based on the average deviation, the normalized window standard deviation, and the overall disturbance.

[0195] Once an anomaly is confirmed, if the recovery scores of multiple consecutive analysis windows meet the preset recovery threshold and the cumulative value drops to zero, the anomaly is resolved.

[0196] Once window-level features have formed anomaly candidates, the system delays the output of the final fault conclusion and first corrects the initial fault score based on the anomaly type and environment level.

[0197] The open circuit type is mainly affected by the degree of decrease of the window average value relative to the benchmark value. Therefore, when the ratio is lower than the open circuit judgment threshold, the initial score is formed by the average deviation. The short circuit type requires that the light circuit has passed the start-up stabilization waiting time and the ratio of the window average value to the benchmark value is higher than the short circuit judgment threshold. The score is formed based on the window change rate to avoid the normal surge during the start-up phase being counted as a short circuit. The poor contact type uses the weighted sum of the normalized window standard deviation, peak density, and window change rate as the scoring source to reflect the fluctuation and jitter characteristics.

[0198] In this embodiment, the results used to calculate open circuit, short circuit and poor contact are all referred to as the initial fault score, and the subsequent compensated results are all referred to as the corrected fault score; the cumulative amount in the corresponding continuous window is uniformly referred to as the cumulative value to avoid confusion with the aforementioned scoring concept.

[0199] The open circuit judgment threshold, short circuit judgment threshold, and severe short circuit threshold are preferably uniformly calibrated under the same ratio dimension, and the open circuit judgment threshold is less than 1 and the severe short circuit threshold is greater than the short circuit judgment threshold. The start-up stabilization waiting time preferably covers the normal surge period after the lamp is powered on, and its value is determined according to the duration from the lamp type from lighting to the feedback entering the stable zone.

[0200] Furthermore, the initial fault scores for open circuit, short circuit, and poor contact are preferably mapped to the same scoring range, which is preferably 0 to 100. The open circuit score is implemented as follows: if the ratio of the window average value to the benchmark value is not lower than the open circuit judgment threshold, the initial open circuit fault score is zero.

[0201] If the ratio is lower than the open-circuit determination threshold, the average deviation is divided by the upper limit of the open-circuit deviation reference to obtain the open-circuit scoring coefficient, and this coefficient is limited to a certain range. to Between these intervals, a linear multiplication is used to map the result to the scoring interval, specifically calculated as follows:

[0202]

[0203] in, For initial fault scoring of open circuit, This is the open-circuit scoring coefficient after restrictions. The upper limit of the scoring range Therefore, the ratio is only used to determine the calculation entry point for the open-circuit score, and the value of the initial open-circuit fault score is uniquely determined by the average deviation.

[0204] The short-circuit scoring is implemented as follows: it is only performed when the lighting time exceeds the startup stabilization waiting time and the ratio is higher than the short-circuit judgment threshold; the window change rate component with the same polarity as the short-circuit is divided by the upper bound of the short-circuit change rate reference to obtain the change rate coefficient, and the coefficient is restricted to between 0 and 1, and then mapped to the scoring interval.

[0205] Furthermore, to unify the short-circuit entry and scoring direction of light circuits with different feedback polarities, the polarity coefficients are pre-calibrated. : Pick This indicates that the feedback signal increases as the load current increases. Pick This indicates that the feedback signal decreases as the load current increases; the signed short-circuit deviation rate is defined. The short-circuit entry condition is unified as follows: The value is greater than or equal to the short-circuit detection threshold minus 1, and the short-circuit change rate component is uniformly taken as... With window change rate The product is processed according to the aforementioned limiting method; for Pick For descending light circuits, open circuit determination is required at the same time. If the value is less than zero, it means that the feedback drop direction is consistent with the polarity of the open circuit, so that the feedback of the falling short circuit will not be mistakenly classified as an open circuit.

[0206] The short-circuit rate of change component is uniformly taken as the product of the polarity coefficient s and the window rate of change. If the product is less than zero, it is treated as zero.

[0207] To avoid the inability to confirm a sustained steady-state short circuit due to a zero window rate of change, the initial short-circuit fault score is weighted by the rate of change component and the steady-state deviation component, according to the formula... Calculation, where To limit the short-circuit rate of change coefficient to between 0 and 1, To limit the steady-state deviation coefficient to between 0 and 1, According to the formula calculate, It equals the short-circuit detection threshold minus 1. It equals the severe short circuit threshold minus 1. , Non-negative weights and Therefore, even if the window change rate is zero after the fault enters steady state, the steady state deviation component will continue to generate a non-zero short circuit initial fault score. The continuous steady-state short circuit can still be confirmed by the cumulative path, and the situation where the fault rise occurs before the start-up stabilization waiting time is covered by the severe short circuit rapid confirmation path.

[0208]

[0209]

[0210] Therefore, the lighting time and ratio are used to determine the calculation entry point for the short circuit score. The value of the initial short circuit fault score is uniquely determined by the window change rate that is consistent with the physical direction of the short circuit, and changes that are opposite to the short circuit direction are not included in the short circuit score.

[0211] The contact defect scoring method is as follows: the normalized window standard deviation, peak density and absolute value of window change rate are converted to between 0 and 1 according to their respective reference upper bounds. The part exceeding 1 is treated as 1. The weighted sum is performed according to the preset weight to obtain the contact defect scoring coefficient, and then mapped to the scoring interval.

[0212] The reference upper bounds for open circuit deviation, short circuit change rate, and poor contact scoring mentioned above are all fixed values ​​greater than zero and remain unchanged under the same lamp type or lamp circuit configuration. If the effective calibration of the corresponding reference upper bound has not been completed, the scoring accumulation of the corresponding abnormal type will not be enabled to avoid the denominator being zero or the use of temporary changing upper bounds leading to incomparability of scores in different windows.

[0213] The three types of initial fault scores are preferably calculated in parallel within the same analysis window: first, the ratio of the window average value to the baseline value is calculated, and then it is determined whether the open circuit entry condition and the short circuit entry condition are met in turn. At the same time, the poor contact score is calculated without conditions.

[0214] For anomaly types that do not meet the entry conditions, their corresponding initial fault scores are directly set to zero; thus, within the same window, the three types of anomalies always generate a set of definite scores, which facilitates subsequent compensation and cumulative link calls one by one.

[0215] Furthermore, if the benchmark value used to calculate the ratio is not greater than the lower limit of the ratio calculation, the window will not open the open circuit and short circuit scoring entry, but will only retain the poor contact scoring path, and the window will be marked as a benchmark insufficient window; only after a valid benchmark value is obtained in the subsequent window will the parallel scoring of open circuit and short circuit be restored; this can avoid the accidental triggering of ratio-type anomalies when the benchmark has not been established or the denominator is too small.

[0216] Furthermore, the open circuit determination threshold is preferably set to 0.2 to 0.7, the short circuit determination threshold is preferably set to 1.2 to 2.0, and the severe short circuit threshold is preferably set to 1.8 to 3.0, and the severe short circuit threshold is always greater than the short circuit determination threshold;

[0217] The startup stabilization waiting time is preferably between 100ms and 1000ms, and the fast confirmation time is preferably between 50ms and 300ms, and should be less than or no more than half of the startup stabilization waiting time, so as to ensure that the direct confirmation path for severe short circuits is faster than the general short circuit accumulation path.

[0218] For the same type of lamp or the same lamp circuit configuration, the above threshold and time parameters use the same set of preset values; when switching to different types of lamps or different lamp circuit configurations, the corresponding pre-stored parameter set is called instead of being temporarily changed in a single analysis window;

[0219] Before the initial fault score is accumulated, it must undergo environmental compensation correction. The correction method is to deduct the comprehensive interference amount according to the compensation coefficient corresponding to the anomaly type and interference level. If the result is less than zero, it is truncated to zero. Thus, in an environment where the interference level value is higher than the preset first environmental threshold, the fault confirmation result is triggered based on the environmental interference intensity limitation of the anomaly score that is susceptible to electromagnetic interference. When the interference level value is lower than the preset second environmental threshold, the deduction amount decreases as the compensation coefficient value decreases, and the fault confirmation can still be carried out according to the preset confirmation cycle.

[0220] For the three types of anomalies—open circuit, short circuit, and poor contact—each can maintain its own independent corrected fault score and cumulative value, and update them synchronously within the same analysis window. This ensures that different anomaly types do not overlap with each other, and that the confirmation conditions of each anomaly type can be compared in parallel within the same window.

[0221] Furthermore, anomaly types Only one of the three conditions—open circuit, short circuit, and poor contact—is considered. Take the calculated overall interference value corresponding to the current analysis window. Take the currently determined interference level status. The coefficients are selected from a pre-defined compensation coefficient table based on the combination of anomaly type and interference level, thereby avoiding cross-referencing of symbols when multiple anomaly types are computed in parallel.

[0222] The compensation coefficient table is preferably set according to the sensitivity of different anomaly types to periodic interference and spike interference. Under the same anomaly type, the compensation coefficient corresponding to the high interference level is not less than the corresponding value of the medium interference level, and the corresponding value of the medium interference level is not less than the corresponding value of the low interference level, so as to ensure that the deduction is more sufficient when the environment is more complex. In the initial fault score of poor contact, each weighting item is preferably a quantity with the same dimension or a quantity after dimension unification. Each weight is a non-negative value and the sum of the weights is preferably 1, so as to avoid a single feature dominating the score without constraint.

[0223] Furthermore, to avoid excessive deduction due to environmental interference in actual contact defects, the deduction amount is... Simultaneously satisfying: First, Not greater than , The preferred value is 0.5 to 0.8, ensuring that the fault correction score remains at least [value missing]. The first is the proportional evidence of the fault; the second is that when the interference level does not escalate within multiple consecutive analysis windows and the corrected fault score remains greater than zero, the corresponding cumulative value continues to accumulate without being interrupted by single-window deduction, thus preserving the continuous evidence of the fault. The scores, deductions, cumulative values, and final outputs under the three operating conditions of normal interference, real fault, and fault superposition interference are listed in the verification example.

[0224]

[0225] First, read the compensation coefficient for each anomaly type under the current interference level, and then multiply the compensation coefficient by the calculated comprehensive interference amount of the current window to obtain the deduction amount;

[0226] The difference between the initial fault score and the deduction amount is used as the corrected fault score. Since the deduction amount is always less than the corresponding initial fault score due to the upper limit of deduction, further, if the corrected fault score after deduction is less than the preset zeroing threshold, the corrected fault score for this type of anomaly is zero. The preset zeroing threshold is preferably equal to the attenuation coefficient, so that the residual score below the single window attenuation amount no longer participates in the accumulation. The compensation coefficient table for the three types of anomalies can adopt a unified storage structure. The index item should at least include the anomaly type identifier and the interference level identifier so that the deduction amount can be obtained in parallel within the same window.

[0227] Furthermore, the correction steps are preferably performed in the order of open circuit, short circuit, and poor contact, respectively, reading the compensation coefficient, calculating the deduction amount, generating the correction fault score, and writing back the corresponding three sets of correction results at once before the end of the same window; if any of the abnormal types is directly confirmed as a short circuit fault corresponding to a severe short circuit in the current window, only the cumulative links of the short circuit type are frozen, while the open circuit and poor contact links continue to be updated according to their respective scores.

[0228] Furthermore, in the compensation coefficient table, the compensation coefficients for open circuit, short circuit, and poor contact are preferably 0.05 to 0.30 at low interference levels, 0.20 to 0.60 at medium interference levels, and 0.40 to 1.00 at high interference levels; among them, the compensation coefficient corresponding to poor contact is preferably not less than the compensation coefficient corresponding to open circuit, so as to reflect that spike and jitter characteristics are more sensitive to interference;

[0229] Furthermore, the actual range of the comprehensive interference quantity after dimensionless processing is 0 to... Preferred The total interference quantity used for determining the interference level and the total interference quantity used for scoring deduction shall be of the same dimensionless caliber, not exceeding 1; in the scoring revision process... The total interference, converted to the scoring interval, is equal to the dimensionless total interference plus the upper limit of the scoring interval. The product of, that is This ensures that the deduction amount directly matches the rating scale from 0 to 100; Taking a dimensionless comprehensive interference value of 0.5 as an example, the deduction amounts for open-circuit types at low, medium, and high interference levels are 2.5 to 15, 10 to 30, and 20 to 50, respectively. Not greater than The constraints can both suppress false alarms in the interference environment and prevent the corrected fault score of the real fault from being suppressed to zero in the long term.

[0230]

[0231] In the initial fault score for poor contact, the normalized window standard deviation, peak density, and absolute value of window change rate are preferably weighted at 0.3 to 0.5, 0.3 to 0.5, and 0.1 to 0.3, respectively, and the sum of their weights is 1. Thus, the magnitude of the compensation in the score correction link and the multi-index combination rules for poor contact score are both clearly defined.

[0232] During the anomaly type confirmation phase, the system adds the cumulative value from the previous window to the corrected fault score in the current window and subtracts the preset attenuation coefficient, while limiting the cumulative value to between zero and the upper limit of the score; the corresponding anomaly type is confirmed only when the cumulative value reaches the product of the basic confirmation threshold and the environmental compensation coefficient.

[0233] For severe short circuits, if the ratio of the window average value to the baseline value remains higher than the severe short circuit threshold and persists until the rapid confirmation time, the short circuit fault is directly confirmed by bypassing the general cumulative path. This path is used to handle anomalies that have a significant impact on the power supply branch and does not rely on the general cumulative threshold to be gradually raised.

[0234] The rapid confirmation time can be obtained by splicing multiple consecutive analysis windows. As long as each window involved in the splicing meets the severe short circuit threshold condition, the corresponding duration is continuously accumulated. Once any window no longer meets the condition, the continuous accumulation is cleared and restarted, thus ensuring that the direct confirmation is based on continuous anomalies rather than isolated spikes.

[0235] If the same analysis window simultaneously meets the general short-circuit cumulative confirmation condition and the severe short-circuit direct confirmation condition, the short-circuit fault confirmation result corresponding to the severe short-circuit will be output first, and the cumulative value of this short-circuit type will be clamped to the integration upper limit or directly set to the confirmation state hold value to avoid the same fault being repeatedly triggered for confirmation in subsequent windows.

[0236] The attenuation coefficient, integration upper limit, basic confirmation threshold, and environmental compensation coefficient are preferably calibrated in a coordinated manner. The attenuation coefficient is used to suppress the isolated accumulation of occasional windows, the integration upper limit is used to limit the cumulative saturation when the anomaly is maintained for a long time, and the environmental compensation coefficient increases monotonically with the increase of the interference level, so that the confirmation threshold required in the high interference environment is higher than that in the low interference environment.

[0237] Furthermore, the cumulative update is performed in a fixed order: first, the cumulative value of the previous window is read, then the fault correction score of the current window is added, and then the attenuation coefficient is subtracted to truncate the upper and lower limits; if the result after subtraction is less than zero, the current cumulative value is zero; if it is greater than the upper limit of the score, the current cumulative value is the upper limit of the score.

[0238] The product of the basic confirmation threshold and the environmental compensation coefficient is calculated only once within a single analysis window and serves as the unique confirmation threshold for the corresponding anomaly type within that window, thereby avoiding inconsistencies in the threshold objects when making repeated comparisons within the same window.

[0239] Furthermore, for each anomaly type, it is preferable to maintain three states: unconfirmed state, confirmed state, and recovery observation state. When the cumulative value reaches the corresponding confirmation threshold for the first time, it switches from the unconfirmed state to the confirmed state. In the confirmed state, if subsequent windows no longer meet the confirmation threshold but the recovery conditions have not yet been met, it enters the recovery observation state.

[0240] Furthermore, the state transitions for each anomaly type are as follows: In the unconfirmed state, the cumulative value is incremented and decayed; in the confirmed state, the fault output is maintained, the cumulative value continues to decay and is truncated according to the maximum score; in the recovery observation state, the cumulative value continues to decay, and after the recovery score continuously meets the standard and the cumulative value drops to zero, the anomaly is removed and the state returns to the unconfirmed state; the critical short circuit fast confirmation entry directly enters the confirmed state, and the score accumulation and threshold comparison are frozen, while decay deduction and recovery count are maintained. When the ratio of the window average value to the baseline value falls below the critical short circuit threshold and remains below the preset time, the freeze is lifted and the state returns to the recovery observation state.

[0241] If the recovery score continuously meets the target and the cumulative value returns to zero during the recovery observation state, the anomaly is resolved and the state returns to unconfirmed. If the fault score rises again during the recovery observation period and the cumulative value reaches the confirmation threshold again, the state returns to confirmed and remains unchanged.

[0242] Furthermore, the environmental compensation coefficient is calibrated and called separately from the aforementioned compensation coefficient table. The environmental compensation coefficient is only used to convert the basic confirmation threshold into the confirmation threshold of the current window and does not participate in the initial fault score correction; the aforementioned compensation coefficient table... Used only for generating deductions from aggregated interference;

[0243] Therefore, the scoring correction link and the confirmation threshold link correspond to independent parameters, avoiding the overlap of control meaning caused by the same environmental compensation coefficient bearing both the deduction amount and the threshold amplification amount at the same time.

[0244] Furthermore, if multiple anomaly types in the same window reach their respective confirmation thresholds simultaneously, the confirmation results are preferably output in the order of direct confirmation of severe short circuit, general short circuit, open circuit, and poor contact. Among them, direct confirmation of severe short circuit takes priority over all cumulative confirmations, general short circuit takes priority over open circuit and poor contact, and when open circuit and poor contact reach their thresholds at the same time, they can maintain their respective confirmation states in parallel.

[0245] After adopting the above priority order, anomalies with higher power supply risk can occupy the control response path first, while anomalies with lower risk retain their independent state and will not be overwritten by concurrent confirmations within a single window.

[0246] Furthermore, the attenuation coefficient is preferably 1 to 10, the integration upper limit is preferably 100 to 300, and the basic confirmation threshold is preferably 40 to 120; the environmental compensation coefficients corresponding to low interference level, medium interference level and high interference level are preferably 1.0 to 1.2, 1.2 to 1.6 and 1.6 to 2.2 respectively, and satisfy the condition that they do not decrease monotonically as the interference level increases;

[0247] Furthermore, the above parameters must satisfy the accessibility constraint of the confirmation threshold: for each anomaly type and its applicable disturbance level, the upper limit of integration. Not less than the basic confirmation threshold Environmental compensation coefficient The product of, that is Furthermore, the attenuation coefficient is less than the typical amplitude of the fault score correction under fault conditions; when the parameter combination within a given interval does not satisfy this constraint, the upper limit of integration or the environmental compensation coefficient is recalibrated according to the reachability constraint; for example, taking... , The time confirmation threshold is 264, and the maximum points are capped. The value should be no less than 264. The original preferred value range of 100 to 300 is only applicable to parameter combinations where the confirmation threshold does not exceed the upper limit of the integral.

[0248]

[0249] For the same anomaly type, if the cumulative value reaches the integration limit but the recovery condition is not triggered, the acknowledgment state is maintained and the value continues to be recorded according to the integration limit, without further overflow. Thus, the accumulation, decay, saturation, and acknowledgment thresholds in the accumulation chain all have quantization boundaries that can be directly invoked.

[0250] After confirmation, the system continues to observe the recovery score; in this embodiment, the recovery score is used to characterize the degree of recovery of the light route from an abnormal state to a stable state, and its value is positively correlated with the recovery state; correspondingly, the smaller the average deviation, the smaller the normalized window standard deviation, and the lower the comprehensive interference, the higher the recovery score; only when the recovery score is higher than the recovery threshold for a consecutive preset number of analysis windows, and the corresponding cumulative value drops to zero, is the current abnormal type removed.

[0251] If the corrected fault score rises again in any window during the recovery observation period, the recovery count will restart to avoid clearing the fault state directly after a short period of stabilization; if multiple anomaly types enter the confirmed state one after another, each anomaly type will independently perform the release judgment according to its own recovery score, recovery count and cumulative value, and will not be released synchronously due to the recovery or confirmation of another anomaly type.

[0252] In this embodiment, the average deviation, normalized window standard deviation and comprehensive interference are first converted into components that are positively correlated with recovery, and then weighted and summed. The recovery threshold is calibrated in a uniform dimension accordingly. Each weighted component is preferably limited to a fixed upper bound, and if any component exceeds the corresponding abnormal upper bound, it is truncated according to the upper bound to avoid a situation where a single abnormality deviates significantly but the overall score still exceeds the preset upper limit.

[0253] Furthermore, the preferred conversion method is to first convert the average deviation, normalized window standard deviation, and comprehensive interference to between 0 and 1 according to their respective upper bounds of the recovery reference, and then subtract the conversion results from 1 to obtain the recovered positive correlation component; if the conversion result exceeds 1, the corresponding recovered positive correlation component is set to zero.

[0254] Each recovery reference upper bound is a fixed value greater than zero and remains unchanged under the same lamp type or the same lamp circuit configuration; a weighted sum is calculated according to preset recovery weights, and then mapped to the same scoring interval as the aforementioned initial fault score by multiplying by the upper limit of the score. The calculation formula is as follows:

[0255]

[0256] in, To restore the rating, The positive correlation component is the recovered mean deviation. The recovered positive correlation component is the normalized window standard deviation. To recover the positive correlation component of the overall interference, for The recovery weight, for The recovery weight, for The recovery weight, This is the upper limit of the scoring range, then... The input, transformation direction, comparison unit, and output range of the recovery score are all determined, making it easy to implement in parallel with the anomaly confirmation link.

[0257] Furthermore, for any confirmed anomaly type, the calculation of the recovery score and the update of the recovery count are performed after the correction fault score and the cumulative value update corresponding to that anomaly type; if the cumulative value of the current window has not yet dropped to zero, then even if the recovery score is higher than the recovery threshold, only the recovery score of that window is recorded and the recovery count is not increased.

[0258] Furthermore, the number of consecutive preset analysis windows is preferably 2 to 5, the recovery threshold is preferably 60 to 90, and the recovery weight preferably satisfies the following conditions: average deviation weight is 0.4 to 0.6, normalized window standard deviation weight is 0.2 to 0.35, comprehensive interference weight is 0.15 to 0.3, and the sum of the weights is 1.

[0259] The above-mentioned upper limit of recovery reference, recovery threshold and recovery weight remain fixed under the same lamp type or the same lamp path configuration, and are stored independently and called separately from the aforementioned confirmation link parameters, so as to avoid the recovery criteria and confirmation criteria from overwriting each other;

[0260] If a certain analysis window cannot form a complete corrected fault score due to insufficient sampled data, then that window will not participate in the accumulation, only the accumulated value of the previous window will be retained, and it will wait for the next valid window to re-enter the judgment; this can ensure that the fault confirmation link is not interrupted when local data is missing.

[0261] For recovery judgment, if a low confidence window appears in the continuous observation window, the window will not be included in the statistics of the continuous preset number, but the recovery count that has been formed will not be cleared, and will only continue to accumulate in the next valid window, so as to avoid the recovery link being erroneously reset due to short-term sample loss.

[0262] Compared with the fixed threshold scheme and the scheme without interference correction, in the above example, this scheme has no false alarms under normal interference conditions and no missed alarms under real fault conditions. The confirmation time is determined by the cumulative window number and the analysis window length. By combining baseline update, interference deduction and cumulative confirmation, the false alarm rate, missed alarm rate and confirmation time can all be reproduced by the aforementioned characteristic parameters and threshold calibration.

[0263] Calculations are performed window-by-window based on three operating conditions: normal interference, actual fault, and fault-plus-interference. All calculations are labeled as examples, and the timing sequence is determined as follows: Figure 5 As shown: Condition 1, normal lighting with periodic interference, reference value. The stability is 1000, the interference level is medium, the comprehensive interference quantity is converted to 30, the initial fault scores for open circuit, short circuit and poor contact are all less than 5, the residual corrected fault score after deduction is less than the zeroing threshold and is taken as zero, the cumulative value remains zero, and there are no false alarms or fault outputs.

[0264] Condition 2: Stable faults above the general threshold and below the severe threshold. Taking poor contact as an example, the normalized window standard deviation, peak density, and absolute value of the window change rate, after being converted according to their respective reference upper limits, are 0.6, 0.5, and 0.5, respectively, with corresponding weights of 0.5, 0.4, and 0.1. The initial fault score for poor contact is 55. Under the medium interference level, the compensation coefficient is 0.3, the converted comprehensive interference amount is 50, and the deduction amount is 15, which meets the upper limit constraint of deduction. The corrected fault score is 40, and the attenuation coefficient is 5. The accumulated values ​​are 35, 70, and 105 respectively, following the rule of first superimposing the corrected fault score and then deducting the attenuation coefficient. The confirmation threshold is the product of the basic confirmation threshold of 40 and the environmental compensation coefficient of 1.2, which is 48. The accumulated value of the second window reaches the confirmation threshold. The confirmation time is about 2 analysis windows. There are no missed reports, and when the window change rate returns to zero under steady state, the normalized window standard deviation and peak density terms still ensure that the score is non-zero.

[0265] Condition 3: Severe short circuit. The signed short circuit deviation rate continuously reaches the severe short circuit threshold and remains there until the rapid confirmation time. The short circuit fault is directly confirmed by bypassing the cumulative path. The fault resolution phase is cleared after the recovery score continuously meets the standard and the cumulative value decays to zero.

[0266] Please see Figure 2 A vehicle headlight circuit anomaly detection system based on ADC feedback includes:

[0267] The sampling circuit is electrically connected to the vehicle lamp path and is used to acquire the ADC sampling sequence of the vehicle lamp path when it is lit.

[0268] The acquisition module is used to read the ADC sampling sequence and cache it according to the analysis window;

[0269] The preprocessing module is used to perform median filtering and amplitude limiting on the ADC sampling sequence to obtain a preprocessed sequence. The baseline value is updated based on the preprocessed sequence, the interference level value of the previous analysis window, and the average absolute value of the difference between adjacent preprocessed values ​​in the current analysis window.

[0270] The feature extraction module is used to extract feature parameters including window mean, average deviation, normalized window standard deviation, peak density and window change rate based on the preprocessed sequence and benchmark value, and extract the comprehensive interference amount based on the target interference frequency.

[0271] The evaluation module is used to determine the interference level status based on the comprehensive interference amount, normalized window standard deviation, and peak density.

[0272] The scoring module calculates the initial fault score for open circuit, short circuit and poor contact anomaly types respectively, and corrects it based on the compensation coefficient corresponding to the comprehensive interference amount and interference level status to obtain the corrected fault score.

[0273] The judgment module is used to accumulate the fault correction score within a continuous analysis window. When the accumulated value is greater than or equal to the preset confirmation threshold corresponding to the interference level, the abnormality type of the vehicle light circuit is determined.

[0274] This embodiment provides a vehicle lamp path anomaly determination system based on ADC feedback to execute the aforementioned determination method. The system performs analog-to-digital conversion on the vehicle lamp path in the lit state through a sampling circuit electrically connected to the vehicle lamp path at a preset sampling frequency to obtain the ADC sampling sequence. The acquisition module reads the ADC sampling sequence and caches it according to the analysis window in a ring buffer mode that advances unidirectionally with the sampling clock.

[0275] The preprocessing module takes the median of multiple adjacent sampled values ​​of the ADC sampling sequence and limits the difference between it and the previous preprocessed value to the maximum allowable range of change, thereby completing median filtering and amplitude limiting to obtain the preprocessed sequence. When the normal range condition is met, the benchmark update coefficient is determined by the average absolute value of the difference between the preprocessed sequence, the interference level value of the previous analysis window, and the difference between adjacent preprocessed values ​​in the current window to update the benchmark value.

[0276] The feature extraction module calculates the average value of the preprocessed values ​​in the current analysis window as the window average, the absolute value of the difference between the average value and the benchmark value as the average deviation, the normalized window standard deviation, the peak density of the difference exceeding the threshold, and the ratio of the difference between the first and last preprocessed values ​​to the duration as the window change rate, based on the preprocessed sequence and the benchmark value. At the same time, it performs cross-correlation calculation on the preprocessed sequence based on the target interference frequency to extract the dimensionless comprehensive interference.

[0277] Furthermore, the evaluation module normalizes the average value of the comprehensive interference amount, normalized window standard deviation and peak density of the continuous analysis window by a constant and calculates the interference level value by weighted summation. It then uses hysteresis comparison logic to compare it with the preset first environmental threshold and second environmental threshold to determine whether the current interference level is low, medium or high.

[0278] Based on this, the scoring module calculates the initial fault score for open circuit, short circuit and poor contact anomaly types according to the above feature parameters, and deducts and corrects the score based on the product of the comprehensive interference amount and the compensation coefficient corresponding to the current interference level.

[0279] The judgment module integrates and accumulates the scores of various corrective faults within a continuous analysis window, combining the preset attenuation coefficient and limiting it to a valid range. When the accumulated value is greater than or equal to the confirmation threshold formed by the product of the basic confirmation threshold and the environmental compensation coefficient, the corresponding abnormality type of the vehicle lamp circuit is determined. At the same time, when the ratio of the window average value to the baseline value continuously reaches the severe short circuit threshold, a rapid confirmation is performed, and the abnormality is removed when the recovery score meets the standard for multiple consecutive frames and the accumulated value drops to zero.

[0280] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for determining abnormalities in vehicle headlight circuits based on ADC feedback, characterized in that, include: The sampling circuit, which is electrically connected to the vehicle headlight circuit, acquires the ADC sampling sequence of the headlight circuit when it is lit, and caches it according to the analysis window. The ADC sampling sequence is filtered and limited to obtain a preprocessed sequence, and the baseline value is updated based on the average absolute value of the difference between the interference level value of the previous analysis window and the difference between adjacent preprocessed values ​​in the current analysis window. Based on the preprocessed sequence and the benchmark value, feature parameters including window average, average deviation, normalized window standard deviation, peak density and window change rate are extracted, and the comprehensive interference amount is extracted based on the preset target interference frequency. The interference level is determined based on the comprehensive interference amount, the normalized window standard deviation, and the peak density. For open circuit, short circuit and poor contact anomaly types respectively, an initial fault score is calculated based on the characteristic parameters, and the initial fault score is corrected by using the comprehensive interference amount and the compensation coefficient corresponding to the interference level state to obtain a corrected fault score. The fault correction score is accumulated within a continuous analysis window. When the accumulated value reaches the preset confirmation threshold corresponding to the interference level state, the abnormality type corresponding to the vehicle light circuit is determined.

2. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, The filtering and limiting processing of the ADC sampling sequence specifically includes: The median value is taken as the median sample value from multiple adjacent ADC sample values. The absolute value of the difference between the median sample value and the previous preprocessed value is limited to a preset maximum allowable range of change, and this value is used as the current preprocessed value.

3. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Updating the baseline value specifically includes: When the vehicle headlight circuit does not currently output an abnormal type, and the ratio of the current analysis window's average value to the previous analysis window's baseline value is within a preset normal range, the current baseline value is updated according to the following formula: in, The sampling point number, This is the current baseline value. The previous baseline value, This is the current preprocessed value. As the baseline update coefficient, Based on the updated coefficients, This is the interference level value from the previous analysis window. It is the average of the absolute values ​​of the differences between adjacent preprocessed values ​​within the current analysis window. correspond The weighting coefficients, correspond Weighting coefficients; like Then let ; like Then let ;in, To update the lower bound of the coefficients, To update the upper limit of the coefficients, and .

4. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Extracting the feature parameters specifically includes: Calculate the average value of the preprocessed values ​​within the current analysis window as the window average; The absolute value of the difference between the average value of the window and the baseline value of the current analysis window is calculated as the average deviation; Calculate the window standard deviation of the preprocessed sequence. When the benchmark value is greater than a preset lower limit, divide the window standard deviation by the benchmark value to obtain the normalized window standard deviation. The frequency at which the absolute value of the difference between adjacent preprocessed values ​​is greater than a preset peak detection threshold is taken as the peak density; The ratio of the difference between the first and last preprocessed values ​​within the current analysis window to the duration of the current analysis window is calculated as the window change rate.

5. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Extracting the comprehensive interference quantity specifically includes: The preprocessed sequence is cross-correlated based on the target interference frequency to obtain the periodic interference amplitude. The periodic interference amplitude is dimensionless based on the window standard deviation of the preprocessed sequence, and the maximum value is taken as the comprehensive interference quantity when there are multiple target interference frequencies.

6. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Determine the interference level status, specifically including: The average values ​​of the comprehensive interference, normalized window standard deviation, and peak density of the continuous analysis window are normalized and weighted summed to obtain the interference level value. Using hysteresis comparison logic, the interference level value is compared with a preset first environmental threshold and a second environmental threshold to determine whether the current interference level is low, medium or high.

7. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Calculating and correcting the initial fault score specifically includes: When the ratio of the average value of the window to the benchmark value is less than or equal to the preset open circuit determination threshold, the initial fault score of the open circuit is calculated based on the average deviation. When the illumination time of the vehicle light circuit exceeds the preset start-up stabilization waiting time, and the ratio is greater than or equal to the preset short circuit judgment threshold, the initial fault score of the short circuit is calculated based on the window change rate. The initial fault score for poor contact is calculated based on the weighted sum of the normalized window standard deviation, the peak density, and the window change rate. For exception types ,according to Correct the initial fault score if Then let ; in, Exception type Initial fault score, To correct the fault rating, The interference level is classified as low, medium, or high. Exception type In the interference level state The interference compensation coefficient is greater than zero. The value is the converted value of the comprehensive interference quantity after being converted according to the upper limit of the scoring range.

8. The method for determining vehicle headlight circuit anomalies based on ADC feedback according to claim 1, characterized in that, Determining the type of anomaly in the vehicle headlight circuit specifically includes: Within the continuous analysis window, the corrected fault score is integrated and accumulated in combination with a preset attenuation coefficient, and the accumulated value is limited to a valid range. An anomaly is confirmed when the cumulative value is greater than or equal to the product of a preset basic confirmation threshold and a preset environmental compensation coefficient. When the ratio of the window average value to the baseline value is continuously greater than or equal to the preset severe short circuit threshold, fast short circuit confirmation is performed. The recovery score is calculated based on the average deviation, the normalized window standard deviation, and the comprehensive interference. Once an anomaly is confirmed, if the recovery scores of multiple consecutive analysis windows meet the preset recovery threshold and the cumulative value drops to zero, the anomaly is resolved.

9. A vehicle headlight circuit anomaly determination system based on ADC feedback, used to execute the method as described in any one of claims 1-8, characterized in that, include: The sampling circuit is electrically connected to the vehicle lamp path and is used to acquire the ADC sampling sequence of the vehicle lamp path when it is lit. The acquisition module is used to read the ADC sampling sequence and cache it according to the analysis window; The preprocessing module is used to perform median filtering and amplitude limiting on the ADC sampling sequence to obtain a preprocessed sequence, and to update the reference value based on the preprocessed sequence, the interference level value of the previous analysis window, and the average absolute value of the difference between adjacent preprocessed values ​​in the current analysis window. The feature extraction module is used to extract feature parameters including window average, average deviation, normalized window standard deviation, peak density and window change rate based on the preprocessed sequence and the benchmark value, and to extract the comprehensive interference amount based on the target interference frequency. The evaluation module is used to determine the interference level status based on the comprehensive interference amount, the normalized window standard deviation, and the peak density; The scoring module calculates initial fault scores for open circuit, short circuit, and poor contact anomaly types, and corrects them based on the comprehensive interference amount and the compensation coefficient corresponding to the interference level state to obtain a corrected fault score. The determination module is used to accumulate the corrected fault score within a continuous analysis window. When the accumulated value is greater than or equal to the preset confirmation threshold corresponding to the interference level state, the abnormality type corresponding to the vehicle light circuit is determined.