A method and system for monitoring and analyzing the working state of an integrated semiconductor transistor

Abstract

The present invention relates to the field of semiconductor device technology, and in particular to a method and system for monitoring and analyzing the working state of an integrated semiconductor transistor. All real-time working parameters on a cloud platform are obtained, and the real-time working parameters are classified and processed to obtain various types of real-time working parameters of a target integrated semiconductor transistor in a preset working time period; various types of real-time working parameters are compared and analyzed with corresponding standard working parameters to obtain the working state of the target integrated semiconductor transistor; if the working state of the target integrated semiconductor transistor is an abnormal state, suspicious components are analyzed according to the abnormal working parameters of the target integrated semiconductor transistor; a fault analysis is performed on the suspicious components to determine whether the suspicious components have failed, and if a fault occurs, the fault type of the faulty component is obtained. The system performance can be optimized, the equipment life can be extended, and the maintenance cost can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor devices, and in particular to a method and system for monitoring and analyzing the working state of an integrated semiconductor transistor. Background Art

[0002] Integrated semiconductor transistors, also known as integrated circuits (ICs), are an important part of microelectronics technology. It integrates multiple electronic components, such as transistors, resistors, capacitors, etc., in a very small size on a silicon substrate and connects them together through precise layout and manufacturing processes. These components work together to realize various logical functions or signal amplification, processing, etc., thus forming the core of modern electronic devices. The emergence of integrated semiconductor transistors has greatly improved the performance density, power efficiency and reliability of electronic devices, and promoted the rapid development of information technology. In the field of electronic equipment, integrated semiconductor transistors are key devices, and the stability and efficiency of their working state are directly related to the performance of the entire system. However, due to the complex microstructure and dynamic behavior inside the transistor, it is a challenge to accurately monitor its working state in real time. Traditional monitoring methods often have problems such as limited accuracy, long response time or high cost. Therefore, the present application proposes a more efficient, accurate and real-time monitoring technology to facilitate early detection of potential faults, optimize system performance, extend equipment life, and reduce maintenance costs to meet the growing complexity and refined management needs of electronic equipment. Summary of the invention

[0003] The present invention overcomes the deficiencies of the prior art and provides a method and system for monitoring and analyzing the working state of an integrated semiconductor transistor.

[0004] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is:

[0005] The first aspect of the present invention discloses a method for monitoring and analyzing the working state of an integrated semiconductor transistor, comprising the following steps:

[0006] Collecting real-time operating parameters of the target integrated semiconductor transistor at several preset time nodes, and sending the collected real-time operating parameters to the cloud platform; wherein the operating parameters include temperature, voltage, current, frequency, resistance and power;

[0007] After the collection is completed, all real-time operating parameters on the cloud platform are obtained, and the real-time operating parameters are classified and processed to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period;

[0008] Compare and analyze various types of real-time operating parameters with corresponding standard operating parameters to obtain the operating state of the target integrated semiconductor transistor;

[0009] If the operating state of the target integrated semiconductor transistor is abnormal, analyzing suspicious components according to the abnormal operating parameters of the target integrated semiconductor transistor;

[0010] Perform fault analysis on suspicious components to determine whether the suspicious components have failed. If a failure has occurred, obtain the fault type of the faulty component, generate a fault report based on the faulty component and the fault type of the faulty component, and send the fault report to a preset platform for display.

[0011] Furthermore, in a preferred embodiment of the present invention, the real-time operating parameters are classified to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, specifically:

[0012] Calculate the covariance between all real-time working parameters on the cloud platform, and generate a covariance matrix according to the calculated covariance; wherein the covariance matrix describes the linear correlation between various real-time working parameters;

[0013] Performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors; the eigenvalues ​​represent the importance of each principal component, and the eigenvectors represent the changes of the working parameters in various directions;

[0014] According to the size of the eigenvalue, the first k principal components are selected; and the high-dimensional real-time working parameters are projected onto the low-dimensional space using the selected first k principal components to obtain the low-dimensional real-time working parameters;

[0015] Obtaining the number of types of the collected real-time working parameters, and initializing a number of classification centers according to the number of types of the collected real-time working parameters; calculating the Manhattan distance between each low-dimensional real-time working parameter and each classification center;

[0016] Determine the membership between each low-dimensional real-time working parameter and each classification center according to the Manhattan distance between each low-dimensional real-time working parameter and each classification center; and sort the membership between each low-dimensional real-time working parameter and each classification center in ascending order to obtain several maximum memberships;

[0017] Allocate each low-dimensional real-time working parameter to a classification center corresponding to the maximum membership degree. After the allocation is completed, combine the low-dimensional real-time working parameters existing in each classification center into a cluster to obtain a plurality of clusters. Perform evaluation and correction processing on each cluster to obtain a corrected cluster.

[0018] The low-dimensional real-time operating parameters in each corrected class group are obtained to obtain the real-time operating parameters of various types of target integrated semiconductor transistors.

[0019] Furthermore, in a preferred embodiment of the present invention, each cluster is evaluated and corrected to obtain a corrected cluster, specifically:

[0020] For each cluster group, the Minkowski distance between each low-dimensional real-time working parameter in the cluster group is calculated, and the average value of the Minkowski distance between each low-dimensional real-time working parameter is calculated to obtain the average Minkowski distance of each cluster group;

[0021] Obtain the position of each average Minkowski distance in the corresponding cluster respectively, and obtain the cluster center of each cluster;

[0022] Calculate the Dunn index of each cluster, and compare the Dunn index of each cluster with a preset index threshold;

[0023] If the Dunn index of a cluster is greater than the preset index threshold, it means that the cluster does not need to be corrected, and the cluster is marked as a normal cluster, and the cluster is directly output as a corrected cluster;

[0024] If the Dunn index of a cluster is not greater than the preset index threshold, it means that the cluster needs to be corrected and marked as an abnormal cluster.

[0025] Furthermore, in a preferred embodiment of the present invention, the following steps are also included:

[0026] Obtaining all abnormal clustering groups, and defining all abnormal clustering groups as to-be-corrected groups; respectively calculating the Minkowski distance between each low-dimensional real-time working parameter in each to-be-corrected group and the group center;

[0027] Compare the calculated Minkowski distances between each low-dimensional real-time working parameter in each group to be corrected and the group center with a preset distance threshold;

[0028] If one or more Minkowski distances in a group to be corrected are greater than a preset distance threshold, it means that the group to be corrected is mixed with other types of low-dimensional real-time working parameters, and the group to be corrected is defined as a type of group to be corrected;

[0029] If there is no situation in the group to be corrected where the Minkowski distance is greater than the preset distance threshold, it means that the group to be corrected lacks sufficient low-dimensional real-time working parameters, and the group to be corrected is defined as the second group to be corrected.

[0030] Furthermore, in a preferred embodiment of the present invention, the following steps are also included:

[0031] If a certain group of classes to be corrected is a class of classes to be corrected, then the low-dimensional real-time working parameters whose Minkowski distance is greater than the preset distance threshold are extracted from the class of classes to be corrected, so as to complete the correction process of the class of classes to be corrected and obtain the corresponding corrected class group; and the low-dimensional real-time working parameters extracted from the class of classes to be corrected are stored in the data storage library;

[0032] After all the first-category groups to be corrected are corrected, the low-dimensional real-time working parameters stored in the data repository are temporarily assigned to the second-category groups to be corrected, and the Minkowski distance between the low-dimensional real-time working parameters temporarily assigned to the second-category groups to be corrected and the group center is calculated;

[0033] If the Minkowski distance between the low-dimensional real-time operating parameter temporarily assigned to the second category of the group to be corrected and the group center is not greater than the preset distance threshold, it means that the low-dimensional real-time operating parameter belongs to the second category of the group to be corrected, then the low-dimensional real-time operating parameter is retained in the second category of the group to be corrected, and the low-dimensional real-time operating parameter retained in the second category of the group to be corrected is deleted from the data storage library;

[0034] If the Minkowski distance between the low-dimensional real-time working parameter temporarily allocated to the second category of the group to be corrected and the group center is greater than the preset distance threshold, it means that the low-dimensional real-time working parameter does not belong to the second category of the group to be corrected, and the low-dimensional real-time working parameter is redeployed back to the data storage library;

[0035] This process is repeated in this way until all low-dimensional real-time working parameters in the data repository are allocated, the correction process of the second category of groups to be corrected is completed, and the corresponding corrected groups are obtained.

[0036] Furthermore, in a preferred embodiment of the present invention, various types of real-time operating parameters are compared and analyzed with corresponding standard operating parameters to obtain the operating state of the target integrated semiconductor transistor, specifically:

[0037] Presetting various types of standard operating parameters of the target integrated semiconductor transistor in a preset working time period; constructing various standard operating parameter curve graphs according to various types of standard operating parameters of the target integrated semiconductor transistor in the preset working time period;

[0038] and obtaining various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, and constructing various real-time operating parameter curve graphs according to various types of real-time operating parameters of the target integrated semiconductor transistor in the preset working time period;

[0039] Calculate the degree of overlap between various real-time working parameter curves and corresponding standard working parameter curves based on the Euclidean distance algorithm;

[0040] If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as a normal working parameter;

[0041] If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is not greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as an abnormal working parameter;

[0042] Performing statistical analysis on the real-time operating parameter state of the target integrated semiconductor transistor, and if there are abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, marking the operating state of the target integrated semiconductor transistor as an abnormal state;

[0043] If there are no abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, the operating state of the target integrated semiconductor transistor is calibrated as a normal state.

[0044] Furthermore, in a preferred embodiment of the present invention, if the operating state of the target integrated semiconductor transistor is abnormal, the suspicious components are analyzed according to the abnormal operating parameters of the target integrated semiconductor transistor, specifically:

[0045] Acquire functional characteristic information of each component in the target integrated semiconductor transistor, and obtain the correlation between each operating parameter and each component based on the functional characteristic information of each component;

[0046] If the correlation between a certain working parameter and a certain component is greater than a preset correlation, the component is defined as a characteristic component having a functional influence on the working parameter; and so on, characteristic components having a functional influence on each working parameter are obtained;

[0047] Constructing a knowledge base, and importing characteristic components having functional impact on each working parameter into the knowledge base;

[0048] Abnormal operating parameters of a target integrated semiconductor transistor are obtained, the abnormal operating parameters are imported into the knowledge base for matching, characteristic components having functional impacts with the abnormal operating parameters are obtained, and the characteristic components having functional impacts with the abnormal operating parameters are defined as suspicious components.

[0049] Furthermore, in a preferred embodiment of the present invention, a fault analysis is performed on the suspected component to determine whether the suspected component has failed. If a failure has occurred, the fault type of the failed component is obtained, specifically:

[0050] Obtaining operation log information of suspicious components, obtaining various fault events of the suspicious components according to the operation log information, obtaining the fault type of each fault event, and performing feature engineering processing on the fault events to obtain fault feature data corresponding to various fault types;

[0051] Acquire real-time characteristic data of each suspicious component, and calculate hash values ​​between the real-time characteristic data and fault characteristic data corresponding to various fault types based on a hash algorithm to obtain multiple hash values;

[0052] Compare the multiple hash values ​​with the preset hash value threshold respectively; if there is at least one hash value greater than the preset hash value, mark the suspicious component as a faulty component, and obtain the fault type of the faulty component;

[0053] If each hash value is not greater than the preset hash value threshold, the suspicious component is continuously monitored;

[0054] The characteristic data include operating current, operating voltage, operating temperature and power consumption.

[0055] A second aspect of the present invention discloses a working state monitoring and analysis system for an integrated semiconductor transistor, the working state monitoring and analysis system for the integrated semiconductor transistor comprising a memory and a processor, the memory storing a working state monitoring and analysis method program for the integrated semiconductor transistor, and when the working state monitoring and analysis method program for the integrated semiconductor transistor is executed by the processor, any one of the steps of the working state monitoring and analysis method for the integrated semiconductor transistor is implemented.

[0056] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: collecting the real-time operating parameters of the target integrated semiconductor transistor at several preset time nodes, and sending the collected real-time operating parameters to the cloud platform; after the collection is completed, obtaining all the real-time operating parameters on the cloud platform, classifying and processing the real-time operating parameters, so as to classify and obtain various types of real-time operating parameters of the target integrated semiconductor transistor in the preset working time period; comparing and analyzing various types of real-time operating parameters with corresponding standard operating parameters, and analyzing and obtaining the working state of the target integrated semiconductor transistor; if the working state of the target integrated semiconductor transistor is an abnormal state, analyzing the suspicious components according to the abnormal operating parameters of the target integrated semiconductor transistor; performing fault analysis on the suspicious components to determine whether the suspicious components have failed, and if a fault has occurred, obtaining the fault type of the faulty components. The present application proposes a more efficient, accurate and real-time monitoring method, so as to facilitate the early detection of potential faults, optimize system performance, extend equipment life, and reduce maintenance costs, so as to adapt to the increasing complexity and refined management needs of electronic equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0058] Figure 1 The overall method flow chart of this monitoring and analysis method;

[0059] Figure 2 This is a partial method flow chart of this monitoring and analysis method;

[0060] Figure 3 Figure 4 is a system block diagram of the overall system of the monitoring and analysis method. DETAILED DESCRIPTION

[0061] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0063] like Figure 1 As shown, the first aspect of the present invention discloses a method for monitoring and analyzing the working state of an integrated semiconductor transistor, comprising the following steps:

[0064] S102: collecting real-time operating parameters of the target integrated semiconductor transistor at a number of preset time nodes, and sending the collected real-time operating parameters to a cloud platform; wherein the operating parameters include temperature, voltage, current, frequency, resistance and power;

[0065] S104: After the acquisition is completed, all real-time operating parameters on the cloud platform are acquired, and the real-time operating parameters are classified and processed to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period;

[0066] S106: comparing and analyzing various types of real-time operating parameters with corresponding standard operating parameters to obtain the operating state of the target integrated semiconductor transistor;

[0067] S108: If the operating state of the target integrated semiconductor transistor is abnormal, analyzing suspicious components according to the abnormal operating parameters of the target integrated semiconductor transistor;

[0068] S110: Perform fault analysis on the suspected component to determine whether the suspected component has failed. If a failure has occurred, obtain the fault type of the failed component, generate a fault report based on the failed component and the fault type of the failed component, and send the fault report to a preset platform for display.

[0069] Furthermore, in a preferred embodiment of the present invention, the real-time operating parameters are classified to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, specifically:

[0070] Calculate the covariance between all real-time working parameters on the cloud platform, and generate a covariance matrix according to the calculated covariance; wherein the covariance matrix describes the linear correlation between various real-time working parameters;

[0071] Performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors; the eigenvalues ​​represent the importance of each principal component, and the eigenvectors represent the changes of the working parameters in various directions;

[0072] According to the size of the eigenvalue, the first k principal components are selected; and the high-dimensional real-time working parameters are projected onto the low-dimensional space using the selected first k principal components to obtain the low-dimensional real-time working parameters;

[0073] Obtaining the number of types of the collected real-time working parameters, and initializing a number of classification centers according to the number of types of the collected real-time working parameters; calculating the Manhattan distance between each low-dimensional real-time working parameter and each classification center;

[0074] Determine the membership between each low-dimensional real-time working parameter and each classification center according to the Manhattan distance between each low-dimensional real-time working parameter and each classification center; and sort the membership between each low-dimensional real-time working parameter and each classification center in ascending order to obtain several maximum memberships;

[0075] Allocate each low-dimensional real-time working parameter to a classification center corresponding to the maximum membership degree. After the allocation is completed, combine the low-dimensional real-time working parameters existing in each classification center into a cluster to obtain a plurality of clusters. Perform evaluation and correction processing on each cluster to obtain a corrected cluster.

[0076] The low-dimensional real-time operating parameters in each corrected class group are obtained to obtain the real-time operating parameters of various types of target integrated semiconductor transistors.

[0077] It should be noted that during the operation of the target integrated semiconductor transistor, the real-time operating parameters of the target integrated semiconductor transistor can be collected at several preset time nodes based on the Internet of Things, and since the collected real-time operating parameters are high-dimensional data, in order to reduce the amount of calculation and reduce the difficulty of calculation, in the present invention, the principal component analysis method is used to reduce the dimensionality of the high-dimensional real-time operating parameters to obtain low-dimensional real-time operating parameters.

[0078] Moreover, since the real-time working parameters collected in the cloud platform are messy, the data types of the real-time working parameters are unknown at this time, such as which are real-time temperature parameters and which are real-time voltage parameters. At this time, these working parameters need to be classified and processed. Specifically, a number of classification centers are initialized according to the number of types of real-time working parameters collected, and the Manhattan distance between each low-dimensional real-time working parameter and each classification center is calculated, so as to determine the corresponding membership degree. According to the membership degree, each low-dimensional real-time working parameter is assigned to the corresponding classification center, and a number of class groups are obtained. Each class group is evaluated and corrected to obtain a corrected class group; the low-dimensional real-time working parameters in each corrected class group are obtained to obtain various types of real-time working parameters of the target integrated semiconductor transistor, so as to quickly classify which are real-time temperature parameters and which are real-time voltage parameters.

[0079] like Figure 2 As shown, further, in a preferred embodiment of the present invention, each class group is evaluated and corrected to obtain a corrected class group, specifically:

[0080] S202: For each cluster group, calculate the Minkowski distance between each low-dimensional real-time working parameter in the cluster group, and calculate the average value of the Minkowski distance between each low-dimensional real-time working parameter to obtain the average Minkowski distance of each cluster group;

[0081] S204: respectively obtaining the position of each average Minkowski distance in the corresponding cluster, and obtaining the cluster center of each cluster;

[0082] S206: Calculate the Dunn index of each class group, and compare the Dunn index of each class group with a preset index threshold;

[0083] S208: If the Dunn index of a cluster is greater than a preset index threshold, it means that the cluster does not need to be corrected, and the cluster is marked as a normal cluster, and the cluster is directly output as a corrected cluster;

[0084] S210: If the Dunn index of a certain cluster is not greater than a preset index threshold, it indicates that the cluster needs to be corrected and marked as an abnormal cluster.

[0085] It should be noted that when classifying the real-time operating parameters of integrated semiconductor transistors, although the above classification algorithm can automatically divide different class groups according to the similarity of parameter data, due to problems such as the algorithm's sensitivity to data, some data may be misclassified, such as mistaking a certain real-time voltage data for real-time current data and classifying the real-time voltage data into the current data set. Therefore, it is necessary to evaluate and correct each class group to evaluate whether certain parameters are incorrectly assigned to a certain class group, so as to adjust and correct the classification results and further improve the reliability of the classification results.

[0086] The Dunn Index is a statistic used to evaluate clustering quality, especially when evaluating classification or clustering algorithms based on the distance or similarity between samples. The larger the Dunn Index, the better the clustering effect. On the contrary, if the Dunn Index is small, it means that the clustering effect may not be good and the sample points within the cluster may be too scattered. If the Dunn Index of a cluster is greater than the preset index threshold, it means that the clustering effect of the cluster is good (better compactness), which means that there is no need to correct the cluster, and the cluster is marked as a normal cluster, and the cluster is directly output as a corrected cluster; if the Dunn Index of a cluster is not greater than the preset index threshold, it means that the clustering effect of the cluster is poor (poor compactness), which means that the cluster needs to be corrected and marked as an abnormal cluster.

[0087] Furthermore, in a preferred embodiment of the present invention, the following steps are also included:

[0088] Obtaining all abnormal clustering groups, and defining all abnormal clustering groups as to-be-corrected groups; respectively calculating the Minkowski distance between each low-dimensional real-time working parameter in each to-be-corrected group and the group center;

[0089] Compare the calculated Minkowski distances between each low-dimensional real-time working parameter in each group to be corrected and the group center with a preset distance threshold;

[0090] If one or more Minkowski distances in a group to be corrected are greater than a preset distance threshold, it means that the group to be corrected is mixed with other types of low-dimensional real-time working parameters, and the group to be corrected is defined as a type of group to be corrected;

[0091] If there is no situation in the group to be corrected where the Minkowski distance is greater than the preset distance threshold, it means that the group to be corrected lacks sufficient low-dimensional real-time working parameters, and the group to be corrected is defined as the second group to be corrected.

[0092] It should be noted that after analyzing and obtaining the cluster groups of each abnormal cluster, all the cluster groups of the abnormal cluster are defined as the cluster groups to be corrected. If a certain cluster group to be corrected is defined as a type of cluster group to be corrected, it means that the cluster group to be corrected is mixed with other types of low-dimensional real-time working parameters, such as the current parameter cluster group is mixed with voltage parameters. If a certain cluster group to be corrected is defined as a type of cluster group to be corrected, it means that there are not enough low-dimensional real-time working parameters in the cluster group to be corrected, such as there are not enough current parameters in the current parameter cluster group.

[0093] Furthermore, in a preferred embodiment of the present invention, the following steps are also included:

[0094] If a certain group of classes to be corrected is a class of classes to be corrected, then the low-dimensional real-time working parameters whose Minkowski distance is greater than the preset distance threshold are extracted from the class of classes to be corrected, so as to complete the correction process of the class of classes to be corrected and obtain the corresponding corrected class group; and the low-dimensional real-time working parameters extracted from the class of classes to be corrected are stored in the data storage library;

[0095] After all the first-category groups to be corrected are corrected, the low-dimensional real-time working parameters stored in the data repository are temporarily assigned to the second-category groups to be corrected, and the Minkowski distance between the low-dimensional real-time working parameters temporarily assigned to the second-category groups to be corrected and the group center is calculated;

[0096] If the Minkowski distance between the low-dimensional real-time operating parameter temporarily assigned to the second category of the group to be corrected and the group center is not greater than the preset distance threshold, it means that the low-dimensional real-time operating parameter belongs to the second category of the group to be corrected, then the low-dimensional real-time operating parameter is retained in the second category of the group to be corrected, and the low-dimensional real-time operating parameter retained in the second category of the group to be corrected is deleted from the data storage library;

[0097] If the Minkowski distance between the low-dimensional real-time working parameter temporarily allocated to the second category of the group to be corrected and the group center is greater than the preset distance threshold, it means that the low-dimensional real-time working parameter does not belong to the second category of the group to be corrected, and the low-dimensional real-time working parameter is redeployed back to the data storage library;

[0098] This process is repeated in this way until all low-dimensional real-time working parameters in the data repository are allocated, the correction process of the second category of groups to be corrected is completed, and the corresponding corrected groups are obtained.

[0099] It should be noted that if the Minkowski distance between some low-dimensional real-time working parameters in a class of groups to be corrected and the group center is greater than the preset distance threshold, it means that these low-dimensional real-time working parameters do not belong to this class of groups. At this time, the low-dimensional real-time working parameters whose Minkowski distance is greater than the preset distance threshold are extracted from the class of groups to be corrected, and the extracted low-dimensional real-time working parameters are temporarily stored in the data storage library, thereby completing the correction process of the class of groups to be corrected.

[0100] When all the first-category groups to be corrected have been corrected, the low-dimensional real-time working parameters stored in the data repository are temporarily allocated to the second-category groups to be corrected. If the Minkowski distance between the low-dimensional real-time working parameters temporarily allocated to the second-category groups to be corrected and the group center is not greater than the preset distance threshold, it means that the low-dimensional real-time working parameters belong to the second-category groups to be corrected, and the low-dimensional real-time working parameters are retained in the second-category groups to be corrected; if the Minkowski distance between the low-dimensional real-time working parameters temporarily allocated to the second-category groups to be corrected and the group center is greater than the preset distance threshold, it means that the low-dimensional real-time working parameters do not belong to the second-category groups to be corrected, and the low-dimensional real-time working parameters are reallocated back to the data repository, and the above steps are repeated until all the low-dimensional real-time working parameters in the data repository are allocated, the correction process of the second-category groups to be corrected is completed, and the corresponding corrected groups are obtained.

[0101] Furthermore, in a preferred embodiment of the present invention, various types of real-time operating parameters are compared and analyzed with corresponding standard operating parameters to obtain the operating state of the target integrated semiconductor transistor, specifically:

[0102] Presetting various types of standard operating parameters of the target integrated semiconductor transistor in a preset working time period; constructing various standard operating parameter curve graphs according to various types of standard operating parameters of the target integrated semiconductor transistor in the preset working time period;

[0103] and obtaining various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, and constructing various real-time operating parameter curve graphs according to various types of real-time operating parameters of the target integrated semiconductor transistor in the preset working time period;

[0104] Calculate the degree of overlap between various real-time working parameter curves and corresponding standard working parameter curves based on the Euclidean distance algorithm;

[0105] If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as a normal working parameter;

[0106] If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is not greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as an abnormal working parameter;

[0107] Performing statistical analysis on the real-time operating parameter state of the target integrated semiconductor transistor, and if there are abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, marking the operating state of the target integrated semiconductor transistor as an abnormal state;

[0108] If there are no abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, the operating state of the target integrated semiconductor transistor is calibrated as a normal state.

[0109] It should be noted that various types of standard operating parameters of the target integrated semiconductor transistor in the preset working time period are preset by relevant technical personnel, that is, various normal working parameters corresponding to the target integrated semiconductor transistor when it works normally in the preset working time period. Then, a real-time working parameter curve chart and a standard working parameter curve chart are drawn according to the standard working parameters and the real-time working parameters to visualize the parameter data and improve the efficiency of subsequent analysis. If the degree of overlap between a real-time working parameter curve chart and the corresponding standard working parameter curve chart is not greater than the preset overlap, the corresponding real-time working parameter is calibrated as an abnormal working parameter. If there are abnormal working parameters in the real-time working parameters of the target integrated semiconductor transistor, the working state of the target integrated semiconductor transistor is calibrated as an abnormal state. Through this method, the working state of the target integrated semiconductor transistor can be quickly evaluated and analyzed in combination with the real-time working parameters and the standard working parameters, and an analysis and evaluation result with high reliability can be obtained.

[0110] Furthermore, in a preferred embodiment of the present invention, if the operating state of the target integrated semiconductor transistor is abnormal, the suspicious components are analyzed according to the abnormal operating parameters of the target integrated semiconductor transistor, specifically:

[0111] Acquire functional characteristic information of each component in the target integrated semiconductor transistor, and obtain the correlation between each operating parameter and each component based on the functional characteristic information of each component;

[0112] If the correlation between a certain working parameter and a certain component is greater than a preset correlation, the component is defined as a characteristic component having a functional influence on the working parameter; and so on, characteristic components having a functional influence on each working parameter are obtained;

[0113] Constructing a knowledge base, and importing characteristic components having functional impact on each working parameter into the knowledge base;

[0114] Abnormal operating parameters of a target integrated semiconductor transistor are obtained, the abnormal operating parameters are imported into the knowledge base for matching, characteristic components having functional impacts with the abnormal operating parameters are obtained, and the characteristic components having functional impacts with the abnormal operating parameters are defined as suspicious components.

[0115] It should be noted that the functional characteristic information includes data on the specifications of components (such as resistance, capacity, operating voltage, current, etc.), electrical characteristics (such as frequency response, temperature coefficient, conduction and cutoff characteristics, etc.), physical size, packaging form, service life, etc. In integrated semiconductor transistors, operating parameters such as current gain, input resistance and output resistance are closely related to the normal operation of components. When these parameters deviate from their normal range, it often reflects that the transistor may be faulty. For example, if the current gain value is too low, it may indicate that the emitter or collector inside the transistor is in poor contact or the material is aging; a decrease in input resistance may indicate that the input signal is over-amplified, which may be due to incorrect base bias or transistor damage; abnormal output resistance may reflect problems with the load capacity of the transistor or noise problems. Components such as transistors and bases are characteristic components that have functional impacts on the corresponding operating parameters. Therefore, by monitoring and analyzing these real-time operating parameters of integrated semiconductor transistors, potential component failure states can be associated and diagnosed so that timely repairs or replacements can be made to ensure the stable operation of the circuit system.

[0116] Furthermore, in a preferred embodiment of the present invention, a fault analysis is performed on the suspected component to determine whether the suspected component has failed. If a failure has occurred, the fault type of the failed component is obtained, specifically:

[0117] Obtaining operation log information of suspicious components, obtaining various fault events of the suspicious components according to the operation log information, obtaining the fault type of each fault event, and performing feature engineering processing on the fault events to obtain fault feature data corresponding to various fault types;

[0118] Acquire real-time characteristic data of each suspicious component, and calculate hash values ​​between the real-time characteristic data and fault characteristic data corresponding to various fault types based on a hash algorithm to obtain multiple hash values;

[0119] Compare the multiple hash values ​​with the preset hash value threshold respectively; if there is at least one hash value greater than the preset hash value, mark the suspicious component as a faulty component, and obtain the fault type of the faulty component;

[0120] If each hash value is not greater than the preset hash value threshold, the suspicious component is continuously monitored;

[0121] The characteristic data include operating current, operating voltage, operating temperature and power consumption.

[0122] It should be noted that the operation log information of components refers to the record and tracking data of the working status of components during the use of electronic equipment. This includes the changes in key parameters such as the operating temperature, voltage, current, frequency, power consumption, as well as the operating time of components, the number of failures, maintenance records, etc.

[0123] The real-time characteristic data of each suspicious component is obtained, and the hash value between the real-time characteristic data and the fault characteristic data corresponding to various fault types is calculated based on the hash algorithm to obtain multiple hash values; wherein, the real-time characteristic data and the fault characteristic data are hashed respectively to generate respective hash codes, and then, by comparing the two hash values, if their hash values ​​are very close, it means that the two sets of characteristic data are similar to a certain extent, because the hash function will map data with similar content to similar hash values. If there is at least one hash value greater than the preset hash value, the suspicious component is marked as a faulty component, and the fault type of the faulty component is obtained, and a fault report is generated according to the faulty component and the fault type of the faulty component, and the fault report is sent to the preset platform for display.

[0124] In summary, by real-time monitoring of key parameters, an alarm can be issued when a transistor has a slight abnormality, and potential faults can be discovered and prevented in advance, thereby optimizing the circuit in time and reducing system instability or crashes caused by transistor failures; and through precise fault location, problematic transistors can be replaced or repaired in a targeted manner, avoiding unnecessary overall replacement and saving costs to adapt to the growing complexity of electronic equipment and the need for refined management.

[0125] In addition, the method further comprises the following steps:

[0126] During the operation of the target integrated semiconductor transistor, obtaining actual electromagnetic intensity values ​​of the target integrated semiconductor transistor at each preset position node, and constructing an actual electromagnetic intensity distribution map according to the actual electromagnetic intensity values;

[0127] Based on the independent component analysis algorithm, the actual electromagnetic intensity distribution map is subjected to eigendecomposition processing to obtain an eigenvector matrix and a diagonal matrix; wherein the eigenvector matrix represents the rotation or projection of the original actual electromagnetic intensity value, and the diagonal matrix reflects the importance of each principal component or the purity of the independent component;

[0128] Select any extreme eigenvector of the eigenvector matrix as a construction reference point, and construct a plane coordinate system according to the construction reference point;

[0129] Re-importing the eigenvector matrix and the diagonal matrix into the plane coordinate system, obtaining a coordinate number set corresponding to each feature point, and obtaining a feature point coordinate number set;

[0130] Based on the LOF algorithm and in combination with the feature point coordinate number set, the LOF value corresponding to each feature point is calculated, and the feature points whose LOF value is greater than a preset value are screened out to obtain a screened feature point coordinate number set;

[0131] Reconstructing an actual electromagnetic intensity distribution map according to the filtered feature point coordinate number set to obtain a corrected actual electromagnetic intensity distribution map;

[0132] The corrected actual electromagnetic intensity distribution diagram is compared with the standard electromagnetic intensity distribution diagram; if the similarity between the corrected actual electromagnetic intensity distribution diagram and the standard electromagnetic intensity distribution diagram is not greater than a preset similarity, it indicates that the working state of the target integrated semiconductor transistor is abnormal.

[0133] Among them, the LOF (Local Outlier Factor) algorithm is a density-based unsupervised learning algorithm used for outlier detection. It is mainly used to find those points in the data set whose density in the neighborhood is significantly lower than that of other data points, that is, abnormal points or outliers.

[0134] It should be noted that integrated semiconductor transistors generate electromagnetic fields when working, and this electromagnetic field is affected by the internal current, voltage and working state of the transistor. By monitoring and analyzing the magnetic field strength distribution map, we can obtain information about the electromagnetic behavior of the transistor when it is running. The change in magnetic field strength can reflect the dynamic behavior inside the transistor, such as current flow, power consumption, temperature change, etc., which are important indicators of the working state of the transistor. When the transistor is in normal working state, the magnetic field strength distribution it generates should show a specific pattern. However, once the transistor fails, such as short circuit, overheating or performance degradation, these patterns will change, causing the magnetic field strength distribution map to deviate from the normal pattern. By comparing and analyzing these changes, we can identify possible anomalies, thereby diagnosing the working state of the transistor in time, preventing or repairing potential problems, and ensuring the stability and performance of the entire circuit system. Therefore, the magnetic field strength distribution map is an effective non-invasive monitoring tool that can be used for real-time monitoring and fault analysis of the working state of integrated semiconductor transistors.

[0135] It should be noted that due to the accuracy of the acquisition equipment, the acquisition environment, and the quality of signal transmission, some of the actual electromagnetic intensity values ​​collected will be distorted and drifted. Distorted and drifted data will reduce the accuracy of the magnetic field intensity distribution map, distort the true situation of the magnetic field intensity distribution, and make it impossible to accurately reflect the working state of the transistor. Therefore, it is necessary to correct the actual magnetic field intensity distribution map to improve the accuracy, reliability, stability and interpretability of the actual magnetic field intensity distribution map.

[0136] In summary, this method can monitor and analyze the working state of integrated semiconductor transistors based on the magnetic field intensity distribution map, and can obtain electromagnetic information during transistor operation in real time. The non-invasive monitoring method has high sensitivity and can capture small abnormal changes, thereby improving the stability and safety of the system. At the same time, the magnetic field intensity distribution map can provide intuitive information about the working state of the transistor, which helps to optimize circuit design and improve device performance.

[0137] like Figure 3 As shown, the second aspect of the present invention discloses a working state monitoring and analysis system for an integrated semiconductor transistor, the working state monitoring and analysis system for the integrated semiconductor transistor comprising a memory 15 and a processor 16, the memory 15 stores a working state monitoring and analysis method program for the integrated semiconductor transistor, when the working state monitoring and analysis method program for the integrated semiconductor transistor is executed by the processor 16, any one of the steps of the working state monitoring and analysis method for the integrated semiconductor transistor is implemented.

[0138] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0139] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0140] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0141] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0142] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0143] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for monitoring and analyzing the working state of an integrated semiconductor transistor, characterized in that: The following steps are involved: Collecting real-time operating parameters of the target integrated semiconductor transistor at several preset time nodes, and sending the collected real-time operating parameters to the cloud platform; wherein the operating parameters include temperature, voltage, current, frequency, resistance and power; After the collection is completed, all real-time operating parameters on the cloud platform are obtained, and the real-time operating parameters are classified and processed to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period; Compare and analyze various types of real-time operating parameters with corresponding standard operating parameters to obtain the operating state of the target integrated semiconductor transistor; If the operating state of the target integrated semiconductor transistor is abnormal, analyzing suspicious components according to the abnormal operating parameters of the target integrated semiconductor transistor; Perform fault analysis on the suspected components to determine whether the suspected components have failed. If a failure has occurred, obtain the fault type of the failed component, generate a fault report based on the failed component and the fault type of the failed component, and send the fault report to a preset platform for display; The real-time operating parameters are classified to obtain various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, specifically: Calculate the covariance between all real-time working parameters on the cloud platform, and generate a covariance matrix according to the calculated covariance; wherein the covariance matrix describes the linear correlation between various real-time working parameters; Performing eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors; the eigenvalues ​​represent the importance of each principal component, and the eigenvectors represent the changes of the working parameters in various directions; According to the size of the eigenvalue, the first k principal components are selected; and the high-dimensional real-time working parameters are projected onto the low-dimensional space using the selected first k principal components to obtain the low-dimensional real-time working parameters; Obtaining the number of types of the collected real-time working parameters, and initializing a number of classification centers according to the number of types of the collected real-time working parameters; calculating the Manhattan distance between each low-dimensional real-time working parameter and each classification center; Determine the membership between each low-dimensional real-time working parameter and each classification center according to the Manhattan distance between each low-dimensional real-time working parameter and each classification center; and sort the membership between each low-dimensional real-time working parameter and each classification center in ascending order to obtain several maximum memberships; Allocate each low-dimensional real-time working parameter to a classification center corresponding to the maximum membership degree. After the allocation is completed, combine the low-dimensional real-time working parameters existing in each classification center into a cluster to obtain a plurality of clusters. Perform evaluation and correction processing on each cluster to obtain a corrected cluster. Obtaining low-dimensional real-time operating parameters in each corrected class group to obtain real-time operating parameters of various types of target integrated semiconductor transistors; Among them, various types of real-time operating parameters are compared and analyzed with corresponding standard operating parameters, and the operating state of the target integrated semiconductor transistor is obtained by analysis, specifically: Presetting various types of standard operating parameters of the target integrated semiconductor transistor in a preset working time period; constructing various standard operating parameter curve graphs according to various types of standard operating parameters of the target integrated semiconductor transistor in the preset working time period; and obtaining various types of real-time operating parameters of the target integrated semiconductor transistor in a preset working time period, and constructing various real-time operating parameter curve graphs according to various types of real-time operating parameters of the target integrated semiconductor transistor in the preset working time period; Calculate the degree of overlap between various real-time working parameter curves and corresponding standard working parameter curves based on the Euclidean distance algorithm; If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as a normal working parameter; If the degree of overlap between a certain real-time working parameter curve and the corresponding standard working parameter curve is not greater than a preset degree of overlap, the corresponding real-time working parameter is calibrated as an abnormal working parameter; Performing statistical analysis on the real-time operating parameter state of the target integrated semiconductor transistor, and if there are abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, marking the operating state of the target integrated semiconductor transistor as an abnormal state; If there are no abnormal operating parameters in the real-time operating parameters of the target integrated semiconductor transistor, calibrating the operating state of the target integrated semiconductor transistor as a normal state; If the working state of the target integrated semiconductor transistor is abnormal, the suspicious components are analyzed according to the abnormal working parameters of the target integrated semiconductor transistor, specifically: Acquire functional characteristic information of each component in the target integrated semiconductor transistor, and obtain the correlation between each operating parameter and each component based on the functional characteristic information of each component; If the correlation between a certain working parameter and a certain component is greater than a preset correlation, the component is defined as a characteristic component having a functional influence on the working parameter; and so on, characteristic components having a functional influence on each working parameter are obtained; Constructing a knowledge base, and importing characteristic components having functional impact on each working parameter into the knowledge base; Acquire abnormal operating parameters of the target integrated semiconductor transistor, import the abnormal operating parameters into the knowledge base for matching, obtain characteristic components having functional impacts with the abnormal operating parameters, and define the characteristic components having functional impacts with the abnormal operating parameters as suspicious components; Among them, the fault analysis is performed on the suspicious components to determine whether the suspicious components have failed. If a failure has occurred, the fault type of the failed component is obtained, specifically: Obtaining operation log information of suspicious components, obtaining various fault events of the suspicious components according to the operation log information, obtaining the fault type of each fault event, and performing feature engineering processing on the fault events to obtain fault feature data corresponding to various fault types; Acquire real-time characteristic data of each suspicious component, and calculate hash values ​​between the real-time characteristic data and fault characteristic data corresponding to various fault types based on a hash algorithm to obtain multiple hash values; Compare the multiple hash values ​​with the preset hash value threshold respectively; if there is at least one hash value greater than the preset hash value, mark the suspicious component as a faulty component, and obtain the fault type of the faulty component; If each hash value is not greater than the preset hash value threshold, the suspicious component is continuously monitored; The characteristic data include operating current, operating voltage, operating temperature and power consumption.

2. The method for monitoring and analyzing the working state of an integrated semiconductor transistor according to claim 1, characterized in that: Each cluster is evaluated and corrected to obtain the corrected clusters, specifically: For each cluster group, the Minkowski distance between each low-dimensional real-time working parameter in the cluster group is calculated, and the average value of the Minkowski distance between each low-dimensional real-time working parameter is calculated to obtain the average Minkowski distance of each cluster group; Obtain the position of each average Minkowski distance in the corresponding cluster respectively, and obtain the cluster center of each cluster; Calculate the Dunn index of each cluster, and compare the Dunn index of each cluster with a preset index threshold; If the Dunn index of a cluster is greater than the preset index threshold, it means that the cluster does not need to be corrected, and the cluster is marked as a normal cluster, and the cluster is directly output as a corrected cluster; If the Dunn index of a cluster is not greater than the preset index threshold, it means that the cluster needs to be corrected and marked as an abnormal cluster.

3. The method for monitoring and analyzing the working state of an integrated semiconductor transistor according to claim 2, characterized in that: The following steps are also included: Obtaining all abnormal clustering groups, and defining all abnormal clustering groups as to-be-corrected groups; respectively calculating the Minkowski distance between each low-dimensional real-time working parameter in each to-be-corrected group and the group center; Compare the calculated Minkowski distances between each low-dimensional real-time working parameter in each group to be corrected and the group center with a preset distance threshold; If one or more Minkowski distances in a group to be corrected are greater than a preset distance threshold, it means that the group to be corrected is mixed with other types of low-dimensional real-time working parameters, and the group to be corrected is defined as a type of group to be corrected; If there is no situation in the group to be corrected where the Minkowski distance is greater than the preset distance threshold, it means that the group to be corrected lacks sufficient low-dimensional real-time working parameters, and the group to be corrected is defined as the second group to be corrected.

4. The method for monitoring and analyzing the working state of an integrated semiconductor transistor according to claim 3, characterized in that: The following steps are also included: If a certain group to be corrected is a group to be corrected, then the low-dimensional real-time working parameters whose Minkowski distance is greater than the preset distance threshold are extracted from the group to be corrected, so as to complete the correction process of the group to be corrected and obtain the corresponding corrected group; and storing the low-dimensional real-time working parameters extracted from the group of classes to be corrected in a data storage library; After all the first-category groups to be corrected are corrected, the low-dimensional real-time working parameters stored in the data repository are temporarily assigned to the second-category groups to be corrected, and the Minkowski distance between the low-dimensional real-time working parameters temporarily assigned to the second-category groups to be corrected and the group center is calculated; If the Minkowski distance between the low-dimensional real-time operating parameter temporarily assigned to the second category of the group to be corrected and the group center is not greater than the preset distance threshold, it means that the low-dimensional real-time operating parameter belongs to the second category of the group to be corrected, then the low-dimensional real-time operating parameter is retained in the second category of the group to be corrected, and the low-dimensional real-time operating parameter retained in the second category of the group to be corrected is deleted from the data storage library; If the Minkowski distance between the low-dimensional real-time working parameter temporarily allocated to the second category of the group to be corrected and the group center is greater than the preset distance threshold, it means that the low-dimensional real-time working parameter does not belong to the second category of the group to be corrected, and the low-dimensional real-time working parameter is redeployed back to the data storage library; This process is repeated in this way until all low-dimensional real-time working parameters in the data repository are allocated, the correction process of the second category of groups to be corrected is completed, and the corresponding corrected groups are obtained.

5. A working state monitoring and analysis system for an integrated semiconductor transistor, characterized in that: The working state monitoring and analysis system of the integrated semiconductor transistor includes a memory and a processor, wherein the working state monitoring and analysis method program of the integrated semiconductor transistor is stored in the memory, and when the working state monitoring and analysis method program of the integrated semiconductor transistor is executed by the processor, the working state monitoring and analysis method steps of the integrated semiconductor transistor as described in any one of claims 1 to 4 are implemented.

Images