Circuit fault automatic testing method and system based on data drive

Abstract

The present application relates to the technical field of measuring electrical variables, and specifically to a data-driven circuit fault automatic testing method and system, the method comprising: obtaining current data at each sampling moment of a PCBA to be detected area; dividing the current data at all sampling moments into each cycle; constructing a current waveform smoothing index, a breakpoint moment sequence, and a time interval error sequence within each cycle; constructing a period jitter steady-state coefficient of each cycle according to the period current waveform smoothing index of adjacent cycles, combined with the distribution of data in the time interval error sequence and the breakpoint moment sequence of adjacent cycles; constructing a jitter distribution random characteristic coefficient of each cycle according to the distribution characteristics of data in the time interval error sequence of adjacent cycles, combined with the period jitter steady-state coefficient of adjacent cycles; and realizing PCBA circuit fault detection using an abnormality detection algorithm. The present application aims to reduce the problem of misjudgment or missed judgment of PCBA circuit faults.

Description

Technical Field

[0001] The present application relates to the technical field of measuring electrical variables, and in particular to a data-driven automated circuit fault testing method and system. Background Art

[0002] Printed Circuit Board Assembly (PCBA) is a sophisticated process involving multiple links, including patch, soldering and testing stages to ensure that its performance meets the expected functional standards. With the rapid development of electronic product manufacturing technology, the structure of PCBA has become more sophisticated and the production process has become more complex and advanced. For example, in the patch process, any slight component offset, solder paste adhesion or component detachment may cause a circuit short circuit or other functional failure. Therefore, accurate fault diagnosis of PCBA is particularly critical in the production process.

[0003] Current PCBA fault detection technology mainly relies on using probes to make physical contact with components on the board, and then identifying potential anomalies by analyzing the current and voltage data of the target components. If the detected signal value is abnormally high, this is usually regarded as a sign of a PCBA failure. Although this method performs well in quickly discovering problems, due to the differences in abnormal signal thresholds of various components in the PCBA, it shows a certain sensitivity when setting the threshold, which may lead to the problem of misjudgment or omission of PCBA circuit faults. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a data-driven circuit fault automatic testing method and system, and the technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a data-driven automated circuit fault testing method, the method comprising the following steps:

[0006] Obtain the current data of the PCBA to be tested area at each sampling time;

[0007] The current data at all sampling moments are divided into periods using a clustering algorithm;

[0008] According to the distribution of current data in each cycle, a current waveform smoothness index in each cycle is constructed;

[0009] According to the acquisition time of the current data in adjacent cycles, the breakpoint time sequence and the time interval error sequence of each cycle are constructed;

[0010] Constructing a period jitter steady-state coefficient of each period according to the current waveform smoothing index within the period of adjacent periods and the distribution of data within the time interval error sequence and breakpoint time sequence of adjacent periods;

[0011] According to the distribution characteristics of the data in the time interval error sequence of adjacent periods, the jitter distribution random characteristic coefficients of each period are constructed in combination with the period jitter steady-state coefficients of adjacent periods;

[0012] According to the random characteristic coefficients of jitter distribution of all cycles, an anomaly detection algorithm is used to realize PCBA circuit fault detection.

[0013] Furthermore, the clustering algorithm is used to divide the current data at all sampling moments into periods, including:

[0014] The current data at all sampling moments are used as the input of the clustering algorithm, and each cluster category is output. The data in the cluster category corresponding to the minimum value in the sine waveform composed of the current data at all sampling moments is used as the breakpoint value, the sampling moment corresponding to the breakpoint value is recorded as the breakpoint moment, and the time between any two adjacent breakpoint moments is regarded as a period;

[0015] The current data in each cycle are organized into a current time series data sequence of each cycle in ascending time order.

[0016] Furthermore, the method for constructing the current waveform smoothness index within the cycle includes:

[0017] The waveform factor and effective value of each cycle are calculated according to the current data in each cycle, the sum of the waveform factor and the effective value of each cycle is calculated, the maximum value of all data in the current time series data sequence of each cycle is counted, the mean of the ratio of all data in the current time series data sequence of each cycle to the maximum value is calculated, and the ratio of the sum value to the mean is used as the current waveform smoothing index in each cycle.

[0018] Furthermore, the method for constructing the breakpoint time sequence and the time interval error sequence includes:

[0019] Obtaining a neighboring data set of each cycle according to the current time series data sequence of adjacent cycles;

[0020] The time interval error of the current signal in each cycle is calculated based on the breakpoint moment, and the breakpoint moments of all cycles in the neighboring data set of each cycle are combined into a breakpoint moment sequence of each cycle in chronological order, and the time interval errors of all cycles in the neighboring data set of each cycle are combined into a time interval error sequence of each cycle in chronological order.

[0021] Furthermore, the method for constructing the nearest neighbor dataset includes:

[0022] The current time series data sequence of each cycle and the current time series data sequence of a preset number of adjacent cycles are used as the neighboring data set of each cycle.

[0023] Furthermore, the method for constructing the periodic jitter steady-state coefficient includes:

[0024] A straight line fitting method is used to obtain a fitting straight line of the breakpoint time sequence of the xth period, and fitting values ​​corresponding to each element in the breakpoint time sequence on the fitting straight line are obtained, and all the fitting values ​​are combined into a control sequence of the breakpoint time sequence of the xth period in chronological order;

[0025] The period jitter steady-state coefficient of the xth period is recorded as , the expression is:

[0026]

[0027] In the formula, is the number of cycles in the neighboring data set of the xth cycle, is the breakpoint time sequence of the xth cycle, is the control sequence of the breakpoint time sequence of the xth period, Indicates the operation of taking the Pearson correlation coefficient. represents the intra-cycle current waveform smoothness index of the jth cycle in the neighboring data set of the xth cycle, and They represent the lengths of the j+1th and jth periods in the neighboring dataset of the xth period, represents the time interval error sequence of the jth period in the neighboring data set of the xth period, To take the variance operation, is a preset modulation coefficient greater than 0, wherein the length of the cycle is the time interval between the start time and the end time of the cycle.

[0028] Furthermore, constructing the random characteristic coefficient of the jitter distribution of each period includes:

[0029] Obtain the deviation value and kurtosis of each period according to the distribution characteristics of the data in the time interval error sequence of each period;

[0030] Calculate the cumulative sum of the deviation values ​​and the kurtosis of each period in the neighbor data set of the x-th period, calculate the ratio of the cumulative sum of each period in the neighbor data set of the x-th period to the period jitter steady-state coefficient, record it as the first ratio, and take the average of the first ratios of all periods in the neighbor data set of the x-th period as the random characteristic coefficient of the jitter distribution of the x-th period.

[0031] Furthermore, the method for constructing the deviation value and kurtosis includes:

[0032] Obtaining the statistical distribution results of each period by using histogram statistics according to the time interval error sequence of each period, taking the statistical distribution results as the input of the goodness of fit test technology, outputting the chi-square statistic value, and taking the difference between each chi-square statistic value and the critical value under the preset confidence level as the deviation value of each period;

[0033] A curve fitting method is used to obtain a fitting curve of the statistical distribution result of each period, and the kurtosis of each fitting curve is calculated as the kurtosis of each period.

[0034] Furthermore, the PCBA circuit fault detection is implemented by using an abnormality detection algorithm according to the random characteristic coefficients of the jitter distribution of all cycles, including:

[0035] The jitter distribution random characteristic coefficients of all cycles are arranged in ascending time order to form a jitter distribution random characteristic coefficient sequence. An anomaly detection algorithm is used to obtain the abnormal probability of each data in the jitter distribution random characteristic coefficient sequence. When data with an abnormal probability greater than a preset jitter distribution random characteristic abnormal threshold appears in the jitter distribution random characteristic coefficient sequence, it indicates that there is a circuit fault in the PCBA. Otherwise, it indicates that there is no circuit fault in the PCBA.

[0036] In a second aspect, an embodiment of the present application also provides a data-driven automatic circuit fault testing system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above methods when executing the computer program.

[0037] This application has at least the following beneficial effects:

[0038] This application first divides the cycle according to the waveform characteristics of the current signal in the PCBA to be detected area, uses the Natural breaks algorithm to divide the cycle, and constructs the current waveform smoothness index within the cycle based on the current time series data sequence of the cycle to determine whether there are abnormal points in the local data;

[0039] Then the jitter characteristics between adjacent cycles are analyzed, and the cycle jitter steady-state coefficient is calculated based on the change in cycle length. The stability of a single cycle is determined through the jitter characteristics of multiple cycles, thereby improving the accuracy of cycle stability analysis.

[0040] Furthermore, by analyzing the histogram statistical distribution of the time interval error sequence, the random characteristic coefficient of the jitter distribution is calculated. The jitter characteristics reflect the cause of the jitter of the PCBA, which is convenient for subsequent judgment of whether there is a circuit fault in the area to be detected of the PCBA;

[0041] Finally, based on the random characteristic coefficient of jitter distribution, an abnormal detection algorithm is used to automatically test the circuit faults in the PCBA to be detected area. This application makes up for the sensitivity defects of the existing detection algorithm in setting thresholds to reduce the problem of misjudgment or omission of PCBA circuit faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 A flowchart of the steps of a data-driven circuit fault automatic testing method provided in one embodiment of the present application;

[0044] Figure 2 The figure is a schematic diagram of the process of obtaining the random characteristic coefficient of jitter distribution. DETAILED DESCRIPTION

[0045] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the data-driven circuit fault automatic testing method and system proposed in the present application, its specific implementation method, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0047] The specific scheme of the data-driven automatic circuit fault testing method and system provided by the present application is described in detail below with reference to the accompanying drawings.

[0048] See also Figure 1 , which shows a flowchart of a data-driven circuit fault automatic testing method provided by an embodiment of the present application, the method comprising the following steps:

[0049] Step 1: Obtain the current data of the PCBA to be inspected area at each sampling time.

[0050] During the PCBA production process, manufacturing defects such as component placement deviation, poor solder joints, circuit breakage or short circuits may occur. In addition, the performance of electronic components may be unstable, or the PCBA may be affected by environmental factors such as temperature, humidity, vibration, and electrostatic discharge. These factors will cause the circuit performance to degrade or directly damage the circuit. To ensure circuit safety, the circuit performance is evaluated by performing current detection on the PCBA to promptly detect faults such as welding defects, component damage or assembly errors, thereby improving product quality.

[0051] This embodiment uses a digital oscilloscope and an in-circuit tester (ICT) to capture the current information of a specific area to be tested in the PCBA. The current waveform data obtained is continuous. In order to facilitate subsequent analysis and processing, equal-interval sampling is implemented on the continuous current data. In this embodiment, the sampling time interval is 2 milliseconds and the total sampling time is 5 minutes. It should be noted that the time interval and the total sampling time are artificially set values, which can be set by the implementer. At this point, the current data at each sampling moment can be obtained.

[0052] Step 2: Use a clustering algorithm to divide the current data at all sampling moments into periods.

[0053] Furthermore, in one embodiment of the present application, the division process of each cycle is as follows:

[0054] After obtaining the current signal of the PCBA to be inspected area, the waveform characteristics and jitter characteristics of the current signal are first analyzed. These analyses, as important circuit fault diagnosis methods, can help identify and solve abnormal conditions in the circuit. For example, there is poor contact between the pins, solder joints and electrical connections of the components, and the energized circuit is subject to electromagnetic interference due to production failures. Since the current data at all sampling moments appear as a sine waveform in the time-current coordinate system, although there are abnormal points that are too large or too small in the local data, it has obvious periodic characteristics as a whole. If the current signal between two adjacent minimum points of the sine waveform is taken as a cycle, the current data can be divided into multiple groups with similar properties.

[0055] Therefore, this embodiment adopts the Natural breaks algorithm for classification. The input of the Natural breaks algorithm is the current data at all sampling moments, and the classification level is set to 3. The output of the Natural breaks algorithm is 3 groups of different categories. Then, the current waveform within each cycle and the jitter characteristics between cycles are analyzed. The Naturalbreaks algorithm is a well-known technology, and the specific process will not be repeated. The implementer can also set the value of the classification level according to the actual situation.

[0056] It should be noted that this embodiment only proposes a clustering algorithm for clustering the current data at all sampling moments. There are many existing clustering algorithms. As other implementation methods, implementers may also use other clustering algorithms to cluster the current data at all sampling moments. This application does not impose any special restrictions on this.

[0057] According to the sinusoidal waveform characteristics of the current signal and the results of the Natural breaks algorithm classification, the group consisting of the minimum values ​​in the sinusoidal waveform is taken as C1, the group consisting of the maximum values ​​in the current sinusoidal waveform is taken as C2, and the remaining groups are recorded as C3. Combined with the data in C1, all current data can be divided into cycles, the data in C1 is taken as the breakpoint value, and the sampling time corresponding to the breakpoint value is recorded as the breakpoint moment. Thus, the time between any two adjacent breakpoint moments is taken as a cycle. The current data in each cycle is organized into a current time series data sequence of each cycle in ascending time order.

[0058] Step 3: construct a current waveform smoothness index within each cycle according to the distribution of current data within each cycle.

[0059] Further, in one embodiment of the present application, the specific construction process of the current waveform smoothness index within each cycle is:

[0060] First, analyze the current data within the cycle and calculate the The waveform factor and effective value of the cycle are calculated based on the input of the waveform factor and effective value. The current timing data sequence of a cycle is recorded as , and record the effective value of the output as The calculation method of the waveform factor and the effective value is a well-known technology, and the specific process will not be repeated. The current waveform smoothness index within the construction period is:

[0061]

[0062] In the formula, Indicates The current waveform smoothness index within a cycle, Indicates The waveform factor of the cycle, Indicates The effective value of a cycle, Indicates The number of data in a cycle, Indicates The i-th data in a cycle, Indicates The maximum value of all data in the current timing data sequence of a cycle.

[0063] like The larger the value, the more uniform the energy distribution of the AC waveform is, and the smaller the fluctuation in the current timing data sequence waveform is. The larger the value is, the greater the power passing through the wire per unit time is, and the stronger the energy transmission capacity of the circuit in the PCBA to be tested area is. The smaller the value, the fewer local abnormal peaks there are in the current time series data sequence waveform, the higher the smoothness of the waveform, and the larger the current waveform smoothness index calculated in the final cycle. Electromagnetic interference can cause data errors, thereby increasing local abnormal peaks and destroying the smoothness of the current waveform. The larger it is, the less electromagnetic interference there is from the electronic components in the PCBA inspection area.

[0064] However, the electromagnetic interference in PCBA is instantaneous and suitable for detection and analysis through short-term data within a cycle. However, the damage of electronic components or poor solder joints in the area to be detected in the PCBA requires analysis of current data within multiple consecutive cycles to determine.

[0065] Step 4: construct the breakpoint time sequence and time interval error sequence of each cycle according to the collection time of the current data in adjacent cycles.

[0066] Furthermore, in one embodiment of the present application, the specific construction process of the breakpoint time sequence and the time interval error sequence of each cycle is as follows:

[0067] The length of each cycle is calculated based on the breakpoint time. The time interval between the start and end of each cycle is taken as the length of each cycle. The time interval error of the current signal in each cycle can be calculated based on the breakpoint time. The current time series data sequence of the xth cycle is used as the neighboring data set of the xth cycle. The breakpoint moments of all cycles in the neighboring data set of the xth cycle are combined into the breakpoint moment sequence of the xth cycle in chronological order. The time interval errors of all cycles in the neighboring data set of the xth cycle are combined into the time interval error sequence of the xth cycle in chronological order. Among them, The implementer can set the value of is 10.

[0068] Step 5: construct the period jitter steady-state coefficient of each cycle according to the current waveform smoothing index within the adjacent cycles, combined with the time interval error sequence and the distribution of data within the breakpoint time sequence of the adjacent cycles.

[0069] Further, in one embodiment of the present application, the specific construction process of the period jitter steady-state coefficient of each period is:

[0070] The least square method is used to obtain the fitting straight line of the breakpoint time sequence of the xth period, and the fitting values ​​corresponding to each element in the breakpoint time sequence on the fitting straight line are obtained. All the fitting values ​​are combined into a control sequence of the breakpoint time sequence of the xth period in chronological order, which is recorded as The calculation of the time interval error and the least square method are well-known technologies, and the specific process will not be described in detail. The implementer may also use other straight line fitting methods, and this embodiment does not impose any special restrictions on this.

[0071] The period jitter steady-state coefficient is constructed from this:

[0072]

[0073] In the formula, Indicates The period jitter steady-state coefficient of a cycle, is the number of cycles in the neighboring data set of the xth cycle, is the breakpoint time sequence of the xth cycle, is the control sequence of the breakpoint time sequence of the xth period, Indicates the operation of taking the Pearson correlation coefficient. represents the intra-cycle current waveform smoothness index of the jth cycle in the neighboring data set of the xth cycle, and They represent the lengths of the j+1th and jth periods in the neighboring dataset of the xth period, represents the time interval error sequence of the jth period in the neighboring data set of the xth period, To take the variance operation, The adjustment coefficient is preset to be greater than 0, and its function is to avoid the denominator being 0. In this embodiment, the adjustment coefficient is set to 1. It should be noted that the value of the adjustment coefficient is artificially set and can be set by the implementer. Among them, the calculation of the Pearson correlation coefficient is a well-known technology, and the specific process will not be repeated.

[0074] like The larger the value, the higher the linear correlation of the current signal breakpoint moment, and the more average the data of each cycle divided by the C1 group. The smaller the value, the smaller the length difference between adjacent cycles of the circuit in the PCBA to be tested area, indicating that the circuit has a stronger stability. The time interval error refers to the deviation between the current signal that changes with time and its ideal position. The smaller it is, the lower the deviation of the current signal from its ideal position in the jth cycle is, and the final period jitter steady-state coefficient is The larger the value, the more stable the circuit in the PCBA to be tested area.

[0075] Step 6: construct the jitter distribution random characteristic coefficient of each cycle according to the distribution characteristics of the data in the time interval error sequence of adjacent cycles and the period jitter steady-state coefficients of adjacent cycles.

[0076] Furthermore, in one embodiment of the present application, the specific construction process of the jitter distribution random characteristic coefficient is:

[0077] The histogram statistics of the time interval error sequence give different distributions, which are caused by different reasons. For example, the jitter caused by the internal thermal noise of the device, the random vibration of the crystal, etc. is random and is recorded as random jitter. Therefore, the distribution obtained by the histogram statistics presents a Gaussian distribution. The reflection, crosstalk, and power supply interference of the circuit signal in the PCBA will cause deterministic jitter, which is recorded as intrinsic jitter. These jitters are repeatable and predictable, and the histogram statistics of the time interval error sequence present a uniform distribution.

[0078] Taking the xth period as an example, the time interval error sequence of the xth period is statistically analyzed by histogram, and the statistical distribution result is recorded as The jitter in the current timing data sequence is usually composed of random jitter and intrinsic jitter, resulting in both Gaussian distribution components and mean distribution components in the histogram statistical distribution. Analyze the distribution of the time interval error sequence histogram statistics to determine Whether it is a uniform distribution, this embodiment uses the Chi-Square goodness of fit test technology, and the input of the Chi-Square goodness of fit test technology is the statistical distribution result , set the significance level to 0.05, and set the null hypothesis to It is consistent with the uniform distribution, and the output is the chi-square statistic value. It should be noted that the value of the significance level is set manually and can be set by the implementer. The difference between the chi-square statistic of the xth cycle and the critical value at the significance level of 0.05 is taken as the deviation value of the xth cycle. The least squares method is used to obtain the fitting curve of the statistical distribution results, which is recorded as the statistical distribution curve , calculate the kurtosis of the statistical distribution curve as the kurtosis of the xth period, denoted as The implementer may also use other curve fitting methods, which are not particularly limited in this embodiment. The Chi-Square goodness of fit test technology and the kurtosis calculation method are well-known technologies, and the specific process will not be repeated here.

[0079] The random characteristic coefficients of the jitter distribution are constructed as follows:

[0080]

[0081] In the formula, Indicates The random characteristic coefficient of the jitter distribution of the cycle, represents the number of cycles in the neighboring data set of the xth cycle, Indicates The nearest neighbor data set of the cycle The period jitter steady-state coefficient of a cycle, Indicates The nearest neighbor data set of the cycle The kurtosis of the cycle, Indicates The deviation value of the cycle. Recorded as the first ratio.

[0082] like The larger the value, the more the distribution of the time interval error sequence conforms to the characteristics of the Gaussian distribution. Combined with the above analysis, it shows that there are more random jitter components in the PCBA circuit. The larger the value, the less uniform the distribution of the time interval error sequence is, indicating that the inherent jitter component in the PCBA circuit is smaller. The smaller it is, the more unstable the circuit in the PCBA area to be tested is. The larger the jitter distribution random characteristic coefficient calculated at the end is, the worse the quality stability of the circuit in the PCBA area to be tested is, and the higher the possibility of failure of the electronic components is. The schematic diagram of the process of obtaining the jitter distribution random characteristic coefficient is as follows: Figure 2 shown.

[0083] Step 7: PCBA circuit fault detection is implemented using an anomaly detection algorithm based on the random characteristic coefficients of the jitter distribution of all cycles.

[0084] Further, in one embodiment of the present application, the specific process of implementing PCBA circuit fault detection is:

[0085] Calculate the jitter distribution random characteristic coefficient of each cycle within the total sampling time, and form a jitter distribution random characteristic coefficient sequence in ascending time order , the SOS (Stochastic Outlier Selection Algorithm) algorithm is used for anomaly detection. The input of the algorithm is the random characteristic coefficient sequence of the jitter distribution. The output of the algorithm is a random characteristic coefficient sequence of jitter distribution The abnormal probability of each data in . The SOS algorithm is a well-known technology, and the specific process will not be repeated here.

[0086] It should be noted that this embodiment only proposes an anomaly detection algorithm for obtaining the anomaly probability of each data in the jitter distribution random feature coefficient sequence. There are many existing anomaly detection algorithms. As other implementation methods, implementers may also use other anomaly detection algorithms to obtain the anomaly probability of each data in the jitter distribution random feature coefficient sequence. This application does not impose any special restrictions on this.

[0087] This embodiment collects the current signal of the PCBA to be detected area, analyzes the periodic jitter characteristics of the current signal to construct the jitter distribution random characteristic coefficient, and diagnoses the circuit fault based on the jitter distribution random characteristic coefficient combined with the SOS algorithm. If the circuit in the PCBA to be detected area is good, the smaller the output abnormal probability value is, if there is a fault in the circuit in the PCBA to be detected area, the larger the output abnormal probability value is. The specific test method is: setting the jitter distribution random characteristic abnormal threshold , The implementer can set the value of If the data with abnormal probability greater than the abnormal threshold of the random characteristic of jitter distribution appears in the random characteristic coefficient sequence of jitter distribution obtained by collecting data during circuit operation, it indicates that there is a circuit fault in the PCBA, and a circuit fault warning is issued. Otherwise, it indicates that there is no circuit fault in the PCBA.

[0088] Based on the same inventive concept as the above method, an embodiment of the present application also provides a data-driven automatic circuit fault testing system, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned data-driven automatic circuit fault testing methods are implemented.

[0089] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0090] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0091] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A data-driven circuit fault automated testing method, characterized in that: The method comprises the following steps: Obtain the current data of the PCBA to be tested area at each sampling time; The current data at all sampling moments are divided into periods using a clustering algorithm, and the current data within the period are organized into a current time series data sequence of each period in ascending order of time; According to the distribution of current data in each cycle, a current waveform smoothness index in each cycle is constructed; Obtain the neighboring data set of each cycle according to the current time series data sequence of adjacent cycles; construct the breakpoint time sequence and time interval error sequence of each cycle according to the acquisition time of the current data in the adjacent cycles; Constructing a period jitter steady-state coefficient of each period according to the current waveform smoothing index within the period of adjacent periods and the distribution of data within the time interval error sequence and breakpoint time sequence of adjacent periods; According to the distribution characteristics of the data in the time interval error sequence of adjacent periods, the jitter distribution random characteristic coefficients of each period are constructed in combination with the period jitter steady-state coefficients of adjacent periods; PCBA circuit fault detection is achieved by using anomaly detection algorithms based on the random characteristic coefficients of jitter distribution in all cycles; The method for constructing the current waveform smoothness index within the cycle includes: Calculate the waveform factor and effective value of each cycle according to the current data in each cycle, calculate the sum of the waveform factor and the effective value of each cycle, count the maximum value of all data in the current time series data sequence of each cycle, calculate the average of the ratios of all data in the current time series data sequence of each cycle to the maximum value, and use the ratio of the sum to the average as the current waveform smoothing index in each cycle; The method for constructing the periodic jitter steady-state coefficient comprises: A straight line fitting method is used to obtain a fitting straight line of the breakpoint time sequence of the xth period, and fitting values ​​corresponding to each element in the breakpoint time sequence on the fitting straight line are obtained, and all the fitting values ​​are combined into a control sequence of the breakpoint time sequence of the xth period in chronological order; The step of constructing the random characteristic coefficient of the jitter distribution of each period includes: Obtain the deviation value and kurtosis of each period according to the distribution characteristics of the data in the time interval error sequence of each period; Calculate the cumulative sum of the deviation values ​​and the kurtosis of each period in the neighbor data set of the x-th period, calculate the ratio of the cumulative sum of each period in the neighbor data set of the x-th period to the period jitter steady-state coefficient, record it as the first ratio, and take the average of the first ratios of all periods in the neighbor data set of the x-th period as the random characteristic coefficient of the jitter distribution of the x-th period.

2. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The method for constructing the period jitter steady-state coefficient also includes recording the period jitter steady-state coefficient of the xth period as , whose expression is: In the formula, is the number of cycles in the neighboring data set of the xth cycle, is the breakpoint time sequence of the xth cycle, is the control sequence of the breakpoint time sequence of the xth period, Indicates the operation of taking the Pearson correlation coefficient. represents the intra-cycle current waveform smoothness index of the jth cycle in the neighboring data set of the xth cycle, and They represent the lengths of the j+1th and jth periods in the neighboring dataset of the xth period, represents the time interval error sequence of the jth period in the neighboring data set of the xth period, To take the variance operation, is a preset modulation coefficient greater than 0, wherein the length of the cycle is the time interval between the start time and the end time of the cycle.

3. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The clustering algorithm is used to divide the current data at all sampling moments into periods, including: The current data at all sampling moments are used as the input of the clustering algorithm, and each cluster category is output. The data within the cluster category corresponding to the minimum value in the sine waveform composed of the current data at all sampling moments is used as the breakpoint value, the sampling moment corresponding to the breakpoint value is recorded as the breakpoint moment, and the time between any two adjacent breakpoint moments is taken as a period.

4. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The method for constructing the breakpoint time sequence and the time interval error sequence comprises: The time interval error of the current signal in each cycle is calculated based on the breakpoint moment, and the breakpoint moments of all cycles in the neighboring data set of each cycle are combined into a breakpoint moment sequence of each cycle in chronological order, and the time interval errors of all cycles in the neighboring data set of each cycle are combined into a time interval error sequence of each cycle in chronological order.

5. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The method for constructing the nearest neighbor dataset comprises: The current time series data sequence of each cycle and the current time series data sequence of a preset number of adjacent cycles are used as the neighboring data set of each cycle.

6. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The method for constructing the deviation value and the kurtosis includes: Obtaining the statistical distribution results of each period by using histogram statistics according to the time interval error sequence of each period, taking the statistical distribution results as the input of the goodness of fit test technology, outputting the chi-square statistic value, and taking the difference between each chi-square statistic value and the critical value under the preset confidence level as the deviation value of each period; A curve fitting method is used to obtain a fitting curve of the statistical distribution result of each period, and the kurtosis of each fitting curve is calculated as the kurtosis of each period.

7. The data-driven automated circuit fault testing method according to claim 1, characterized in that: The PCBA circuit fault detection is implemented by using an abnormality detection algorithm according to the random characteristic coefficients of the jitter distribution of all cycles, including: The jitter distribution random characteristic coefficients of all cycles are arranged in ascending time order to form a jitter distribution random characteristic coefficient sequence. An anomaly detection algorithm is used to obtain the abnormal probability of each data in the jitter distribution random characteristic coefficient sequence. When data with an abnormal probability greater than a preset jitter distribution random characteristic abnormal threshold appears in the jitter distribution random characteristic coefficient sequence, it indicates that there is a circuit fault in the PCBA. Otherwise, it indicates that there is no circuit fault in the PCBA.

8. A data-driven circuit fault automatic test system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images