A circuit board fault detection method and system

Abstract

The present application relates to the field of fault detection of circuit boards, and in particular to a fault detection method and system for circuit boards, the method comprising the steps of: calculating the degree of abnormality of electrical data of the circuit board; wherein the electrical data is set to be multi-dimensional data including at least voltage and current; calculating the abnormality score of the electrical data according to the position of the electrical data in an isolation tree; multiplying the degree of abnormality by the abnormality score as a weighting coefficient to obtain a weighted abnormality score, and determining that the detection is abnormal when the weighted abnormality score is greater than a preset abnormality threshold. The present application can effectively isolate abnormal data points, and is suitable for circuit board fault detection containing multi-dimensional electrical data to improve the accuracy of fault detection.

Description

Technical Field

[0001] The present application relates to the field of circuit board fault detection, and in particular to a circuit board fault detection method and system. Background Art

[0002] In modern electronic devices, circuit boards are one of the core components, which involve increasing complexity, especially the application of high-density, high-frequency and multi-layer circuit boards, resulting in increasingly diverse types of faults. Circuit board failures may not only lead to performance degradation of electronic equipment, but may even cause serious system failures. Traditional circuit board fault detection methods mainly rely on manual inspection, visual inspection, and diagnostic methods based on traditional signal testing, which are highly subjective and inefficient. Especially in the detection process of highly complex circuit boards, it is often difficult to meet the requirements of efficiency and accuracy.

[0003] With the rapid development of data processing technology and intelligent technology, the fault detection of circuit boards can effectively detect the abnormality of the working state of the circuit board by collecting various electrical data (such as voltage, current, impedance, etc.) under the working state of the circuit board and monitoring the working state of the circuit board in real time. For example, the method for detecting circuit boards disclosed in the Chinese patent application document with publication number CN1614436A records the state after each potential change, the time of change and the number of state changes of each test point within the waiting time of the power-on test of the qualified circuit board through a microprocessor, and the state after each potential change, the time of change and the number of state changes of each test point within the waiting time of the power-off test; and then records the state after each potential change, the time of change and the number of state changes of each test point within the waiting time of the power-on test on the circuit board to be detected through a microprocessor; and compares the two detection data to determine whether the detected circuit board is qualified.

[0004] Since circuit board fault detection usually involves a large amount of electrical data (such as voltage, current, impedance, etc.), the data required for fault detection is of high dimension, so the isolation forest algorithm can be used for processing. The isolation forest algorithm is a tree-structured anomaly detection method that can be applied to anomaly detection of high-dimensional data. Its basic idea is to construct multiple isolated trees, use the tree structure to divide data points, and then identify abnormal data points in the isolated trees. Among them, the closer the data is to the root node of the isolated tree, the higher the degree of abnormality.

[0005] Since the electrical data of circuit boards has many dimensions, the traditional isolation forest algorithm performs segmentation by randomly selecting thresholds. If there are abnormal points in the electrical data that are significantly different from normal points during the collection process, the data points may not be effectively isolated due to the sparse distribution of certain dimensions, making the abnormal score of the data point insignificant. Summary of the invention

[0006] In order to solve the problem of how to effectively isolate data points, the present application provides a fault detection method and system for a circuit board.

[0007] In a first aspect, the present application provides a fault detection method for a circuit board, which adopts the following technical solution:

[0008] A fault detection method for a circuit board comprises the steps of: calculating the abnormality degree of the circuit board electrical data; wherein the electrical data is set to be multi-dimensional data including at least voltage and current; calculating the abnormality score of the electrical data according to the position of the electrical data in an isolated tree; multiplying the abnormality degree by the abnormality score as a weighting coefficient to obtain a weighted abnormality score, and determining that the detection is abnormal when the weighted abnormality score is greater than a preset abnormality threshold; wherein the calculation formula of the abnormality degree is:

[0009] , where Indicates The first The degree of abnormality of electrical data; Indicates The first The confidence level of the electrical data; Indicates the total number of clusters; Indicates The clusters and The similarity of clusters; Represents the standard normalization function.

[0010] The beneficial effects are: by clustering the electrical data and obtaining the spatial distribution characteristics of the electrical data based on the clustering results, the degree of abnormality of any electrical data point is calculated, and the abnormality score of the electrical data is weighted according to the degree of abnormality of the electrical data point, so as to avoid the problem that the abnormal data and normal data are too close to the root node in the isolated tree, thereby making the abnormal score calculation result more accurate and the early warning result calculation more accurate.

[0011] Optionally, the confidence level of electrical data is calculated as:

[0012] , where Indicates The first The confidence level of the electrical data; Indicates The first The Euclidean distance between each electrical data point and the cluster center; Indicates The total number of electrical data for clusters; Indicates the electrical data ordinal number; Indicates The mean of the Euclidean distances between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function of base .

[0013] The beneficial effects are: Indicates the degree of dispersion of the cluster. Reflects the position of the data point relative to the cluster center. If The smaller the value is and the smaller the discreteness of the cluster is, the closer the data point is to the cluster center and the denser the cluster is. In this case, the confidence level of the data point is higher; otherwise, the confidence level of the data point is lower.

[0014] Optionally, the confidence level of the electrical data is calculated as: , where Indicates The first The confidence level of the electrical data; Indicates The first The Euclidean distance between each electrical data point and the cluster center; Indicated by natural constant The exponential function of base .

[0015] The beneficial effects are: the computational effort is small, and the proximity of data points to the cluster center can be quickly evaluated to quantify the confidence level of the electrical data.

[0016] Optionally, the calculation formula for cluster similarity is:

[0017] , where Indicates The clusters and The similarity of clusters, Indicates The total number of electrical data for clusters; Indicates The total number of electrical data for clusters; Indicates The first Electrical data and The first The Euclidean distance of electrical data; Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function of base .

[0018] The beneficial effects are: the average Euclidean distance between all electrical data in two clusters and the standard deviation of the Euclidean distance between all electrical data in the clusters and the cluster center are comprehensively considered, and the similarity of the two clusters is measured more comprehensively. Both the distance between data points and the distribution of data points relative to their respective cluster centers are considered, which is suitable for better identifying clusters that are similar in spatial distribution and in data distribution characteristics.

[0019] Optionally, the calculation formula for cluster similarity is:

[0020] , where Indicates The clusters and The similarity of clusters, Indicates The total number of electrical data for clusters; Indicates The total number of electrical data for clusters; Indicates The first Electrical data and The first The Euclidean distance of electrical data, Indicated by natural constant The exponential function of base .

[0021] Beneficial effect: Only the average Euclidean distance between all electrical data in two clusters is considered, which simply and directly measures the similarity between clusters. The similarity calculated by this method can quickly evaluate the spatial similarity between clusters, which is suitable for scenarios where the clustering effect needs to be quickly evaluated.

[0022] Optionally, the calculation formula for cluster similarity is: , where Indicates The clusters and The similarity of clusters, Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function of base .

[0023] The beneficial effect is that only the standard deviation difference of the Euclidean distance between all electrical data in the cluster and the cluster center is considered, focusing on measuring the similarity of the data distribution within the cluster. This method is suitable for those clusters with small differences in shape and size, but large differences in internal data distribution characteristics, and can effectively identify the similarity of these clusters.

[0024] Optionally, the clustering algorithm adopts a density peak clustering algorithm.

[0025] In a second aspect, the present application provides a fault detection system for a circuit board, which adopts the following technical solution:

[0026] A fault detection system for a circuit board comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the fault detection method for the circuit board is implemented.

[0027] The beneficial effect is that the above-mentioned circuit board fault detection method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a system based on the memory and the processor for easy use.

[0028] The present application has the following technical effects: by clustering electrical data and obtaining the spatial distribution characteristics of the electrical data based on the clustering results, the degree of abnormality of any electrical data point is calculated, and the abnormality score of the electrical data is weighted according to the degree of abnormality of the electrical data point, thereby avoiding the problem of abnormal data and normal data being too close to the root node in an isolated tree due to the small distance between the abnormal data and the normal data, thereby making the abnormal score calculation result more accurate and the early warning result calculation more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easily understood. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-restrictive manner, and the same or corresponding numbers represent the same or corresponding parts.

[0030] Figure 1 It is a method flow chart of a fault detection method for a circuit board according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0032] It should be understood that when the terms "first", "second", etc. are used in the claims, specification and drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0033] The present application embodiment discloses a fault detection method for a circuit board, referring to Figure 1 , including steps S1-S2, which are as follows:

[0034] S1: Calculate the abnormality of the circuit board electrical data; wherein the electrical data is set to be multi-dimensional data including at least voltage and current.

[0035] Collect electrical data of the circuit board; During the operation of the circuit board, electrical data such as voltage, current, and temperature are generated, which can reflect the operating status and performance of each component on the circuit board. Therefore, abnormal detection of the electrical data of the circuit board can reflect the failure of the circuit board. In this application, voltage, current, and temperature are used as examples for description. By arranging voltage sensors, current sensors, and temperature sensors at the nodes of the circuit board, the voltage, current, and temperature data of the circuit board are collected respectively. The preset collection frequency is 5Hz, and the data of all dimensions collected at the same time is regarded as a multi-dimensional electrical data.

[0036] Since there is a big difference in the distribution of abnormal data and normal data in space. Therefore, this application clusters the electrical data of the circuit board and obtains the abnormality scores of all data points based on the clustering results. If there is abnormal data in the electrical data of the circuit board, after clustering, the clustering cluster where the abnormal data is located will be very different from the clustering cluster of normal data. Therefore, this application calculates the similarity between any two clustering clusters to facilitate the subsequent distinction between abnormal clustering clusters and normal clustering clusters.

[0037] In one embodiment, the calculation formula of the similarity of clusters is:

[0038] , where Indicates The clusters and The similarity of clusters, Indicates The total number of electrical data for clusters; Indicates The total number of electrical data for clusters; Indicates The first Electrical data and The first The Euclidean distance of electrical data; It represents the average Euclidean distance between all electrical data in two clusters. The smaller the value is, the higher the similarity between the two clusters is.

[0039] Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; The smaller the value, the closer the distribution of the two clusters is, and the higher the similarity of the two clusters is.

[0040] Indicated by natural constant The purpose of the exponential function with base is to normalize the quantity to a uniform dimension.

[0041] This embodiment comprehensively considers the average Euclidean distance between all electrical data in two clusters and the standard deviation of the Euclidean distance between all electrical data in the clusters and the cluster center, and more comprehensively measures the similarity between the two clusters, considering both the distance between data points and the distribution of data points relative to their respective cluster centers. This embodiment is suitable for better identifying clusters that are similar in spatial distribution and similar in data distribution characteristics.

[0042] In one embodiment, the calculation formula of the similarity of clusters can also be:

[0043] , where Indicates The clusters and The similarity of clusters, Indicates The total number of electrical data for clusters; Indicates The total number of electrical data for clusters; Indicates The first Electrical data and The first The Euclidean distance of electrical data, Indicated by natural constant The exponential function of base .

[0044] This embodiment only considers the average Euclidean distance between all electrical data in two clusters, and simply and directly measures the similarity between clusters. The similarity calculated by this method can quickly evaluate the spatial similarity between clusters, which is suitable for scenarios that require rapid evaluation of clustering effects.

[0045] In one embodiment, the calculation formula of the similarity of clusters is:

[0046] , where Indicates The clusters and The similarity of clusters, Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicates The standard deviation of the Euclidean distance between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function of base .

[0047] This embodiment only considers the standard deviation difference of the Euclidean distance between all electrical data in the cluster and the cluster center, focusing on measuring the similarity of data distribution within the cluster. This method is suitable for those clusters with small differences in shape and size, but large differences in internal data distribution characteristics, and can effectively identify the similarity of these clusters.

[0048] The clustering algorithm adopts the density peak clustering algorithm. The density peak clustering algorithm is a prior art and will not be described in detail here. Other clustering algorithms applicable to the present application may also be adopted.

[0049] At this point, the similarity calculation between any two clusters is completed.

[0050] Due to the differences in data distribution in different clusters, for clusters with denser distribution, the degree of abnormality of all data points in the cluster can be reflected by the degree of abnormality of the cluster, while for clusters with more discrete distribution, the degree of abnormality of the cluster cannot well reflect the degree of abnormality of the cluster. Therefore, this application needs to calculate the confidence level of any electrical data in any cluster.

[0051] In one embodiment, the calculation formula for the confidence level of electrical data is:

[0052] , where Indicates The first The confidence level of the electrical data; Indicates The first The Euclidean distance between each electrical data point and the cluster center; Indicates The total number of electrical data for clusters; Indicates the electrical data ordinal number; Indicates The mean of the Euclidean distances between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function of base .

[0053] in, It indicates the degree of discreteness of the cluster. The larger the value, the more discrete the electrical data distribution in the cluster. Conversely, the smaller the value, the denser the electrical data distribution in the cluster.

[0054] Quantified the The first The distance between a data point and the center of the cluster. This distance can reflect the position of the data point relative to the cluster center. The smaller the distance, the closer the data point is to the cluster center and may be more similar to other data points in the cluster center; the larger the distance, the farther the data point is from the cluster center and may be different from other data points in terms of characteristics.

[0055] like The smaller the value, and The smaller the value, the closer the data point is to the cluster center and the denser the cluster is. In this case, the confidence level of the data point is higher. Otherwise, the confidence level of the data point is lower.

[0056] In one embodiment, the confidence level of electrical data is calculated as follows: , where Indicates The first The confidence level of the electrical data; Indicates The first The Euclidean distance between each electrical data point and the cluster center; Indicated by natural constant An exponential function with a base of . Compared with the previous embodiment, the amount of calculation is small, and the proximity of the data point to the cluster center can be quickly evaluated to quantify the confidence level of the electrical data.

[0057] The abnormality of the data point is calculated based on the similarity between any cluster and other clusters and the confidence of any data point in the cluster; during the operation of the circuit board, abnormal data may appear occasionally due to interference from the external environment or component failure, so the number of abnormal data is small relative to normal data. Therefore, after clustering the electrical data of the circuit board, the cluster with lower similarity to other clusters has higher abnormality.

[0058] After obtaining the similarity of the clusters and the confidence of the electrical data, the degree of abnormality is calculated. Specifically, the calculation formula of the degree of abnormality is: , where Indicates The first The degree of abnormality of electrical data; Indicates The first The confidence level of the electrical data; Indicates the total number of clusters; Indicates The clusters and The similarity of clusters; Represents a standard normalization function, which is used to limit the abnormality value range between 0 and 1. If the confidence level of the electrical data is higher and the similarity between the cluster to which it belongs and other clusters is lower, the abnormality of the electrical data point is higher.

[0059] S2: Calculate the anomaly score of the electrical data according to the position of the electrical data in the isolated tree; multiply the anomaly score by the anomaly degree as a weighting coefficient to obtain a weighted anomaly score. When the weighted anomaly score is greater than a preset anomaly threshold, the detection is determined to be abnormal.

[0060] Construct an isolation tree based on the isolation forest algorithm. The traditional isolation forest algorithm constructs an isolation tree by randomly setting a segmentation threshold at each layer of the isolation tree, and calculates the abnormality score of the data point based on the distance of the data point from the root node in the isolation tree, so that the closer the data point is to the root node, the higher the abnormality score. Since the traditional isolation forest algorithm only considers the distance between the data point and the root node in the process of calculating the abnormality score of the data point, and during the operation of the circuit board, there may be abnormal data points due to component failure or external environmental interference. These data points are quite different from normal data. When using the isolation forest algorithm to segment them, the difference between the abnormal data points and the normal data in the isolation tree is not large, so the calculation results of the abnormality score will be closer, making the calculation results of the abnormality score of the abnormal data and the normal data too close.

[0061] For any electrical data, the anomaly score of the electrical data is calculated according to the position of the electrical data in the isolated tree;

[0062] The calculation method of abnormal score is: set any electrical data as a sample , calculate the sample Anomaly score , , where Representation sample The average path length among all isolated trees; represents the total number of samples (electrical data) in the isolated tree; is a constant, representing the expected value of the path length; The calculation formula is as follows: ; ; In the formula yes The harmonic number is the number from arrive The sum of the reciprocals of . Among them, calculating the abnormal score of the electrical data point according to the position of the electrical data in the isolation tree is the existing technology of the isolation forest algorithm, which will not be described in detail here;

[0063] The product of the abnormality score of the electrical data point and its abnormality degree is used as the weighted abnormality score of the electrical data point. Specifically, the abnormality degree of any electrical data point is used as a weighting coefficient, and the product of the weighting coefficient and the abnormality score of the electrical data point is used as the weighted abnormality score.

[0064] Exemplarily, the preset abnormal threshold is 0.8. The setting of the abnormal threshold can be adjusted according to the actual application scenario, which will not be elaborated here.

[0065] In one embodiment, if the abnormality score of the collected electrical data is greater than a preset abnormality threshold, the detection abnormality is determined; if the abnormality scores of 10 consecutive (which can be adjusted according to the actual application scenario) electrical data collected are all greater than the preset abnormality threshold, it is determined that the circuit board is faulty.

[0066] An embodiment of the present application also discloses a fault detection system for a circuit board, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the fault detection method for the circuit board according to the present application is implemented.

[0067] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0068] In the present application, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high bandwidth memory HBM (High Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0069] Although this specification has shown and described a plurality of embodiments of the present application, it is obvious to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, modifications and alternatives without departing from the thought and spirit of the present application. It should be understood that in the process of practicing the present application, various alternatives to the embodiments of the present application described herein may be adopted.

[0070] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A fault detection method for a circuit board, characterized in that: Includes steps: Clustering the electrical data of the circuit board to obtain multiple clusters; Calculate the degree of abnormality of the electrical data of the circuit board; wherein the electrical data is set to be multi-dimensional data including at least voltage and current; calculate the abnormality score of the electrical data according to the position of the electrical data in the isolated tree; multiply the abnormality degree by the abnormality score as a weighting coefficient to obtain a weighted abnormality score, and when the weighted abnormality score is greater than a preset abnormality threshold, determine that the detection is abnormal; The calculation formula of abnormality degree is: , where Indicates The first The degree of abnormality of electrical data; Indicates The first The confidence level of the electrical data; Represents the total number of clusters; Indicates The clusters and The similarity of the clusters; represents the standard normalization function; The confidence level of electrical data is calculated as: , where Indicates The first The confidence level of the electrical data; Indicates The first The Euclidean distance between each electrical data point and the cluster center; Indicates The total number of electrical data for clusters; Indicates the electrical data ordinal number; Indicates The mean of the Euclidean distances between all electrical data in a cluster and the cluster center; Indicated by natural constant The exponential function with base ; The calculation formula for the similarity of clusters is: , where Indicates The clusters and The similarity of clusters, Indicates The total number of electrical data for clusters; Indicates The total number of electrical data for clusters; Indicates The first Electrical data and The first The Euclidean distance of electrical data, Expressed as a natural constant The exponential function of base .

2. The fault detection method for a circuit board according to claim 1, characterized in that: The clustering algorithm uses the density peak clustering algorithm.

3. A fault detection system for a circuit board, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the fault detection method for a circuit board according to claim 1 or 2 is implemented.

Images