Server aging test method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of server aging testing, in particular to a server aging testing method and device, electronic equipment and a storage medium. Then convolution processing is carried out on each aging data queue in a sliding mode through a difference operator, the obtained multiple first convolution queues are segmented one by one, queue segments obtained through segmentation are normalized, and multiple queue segments are obtained; performing dot product calculation on each queue segment, each same-time-segment queue segment and each same-part queue segment to obtain a plurality of dot product results, and adding representative values determined according to the plurality of dot product results into a test analysis matrix; and finally, according to the plurality of row vectors of the test analysis matrix, determining the abnormal probability of the plurality of tested parts of the server. According to the method, the state of the detected component is determined based on the cross point product and probability estimation, so that the problem component can be found early.

Description

Technical Field

[0001] The present invention relates to the technical field of server aging testing, and in particular, to a server aging testing method, device, electronic device and storage medium. Background Art

[0002] Server aging testing, as the name implies, is to simulate the load situation of a server in a long-term working environment and perform continuous, high-intensity, and ultra-large-load operation tests. Server aging testing can expose potential problems that may exist in the server under extreme conditions to ensure the stability and reliability of the server during actual use.

[0003] Currently, the aging testing of servers is to run some special test programs to step by step make the performance indicators of each component of the server reach the maximum value, and at the same time, parameters such as the temperature, voltage, and power of the server are monitored. If problems such as server crashes, restarts, and data errors occur during the testing process, it means that there may be hardware failures in the server, and it is particularly important to detect and repair problems in a timely manner at this time.

[0004] Although the server tests each component step by step, it can quickly determine the cause of the problem when a problem occurs in the server aging test. However, this testing method takes a long time and has low efficiency. For this reason, some technologies propose a method of testing multiple components of the server simultaneously. However, this testing method is not easy to determine the component with problems when some components of the server have problems, and it is even less possible to detect the problem component at an early stage of the test.

[0005] Based on this, it is necessary to develop and design a server aging testing method. Summary of the Invention

[0006] Embodiments of the present invention provide a server aging testing method, device, electronic device and storage medium for solving the problem of low efficiency of server aging testing in the prior art.

[0007] In a first aspect, embodiments of the present invention provide a server aging testing method, including:

[0008] Obtain a plurality of aging data queues, where each aging data queue corresponds to a tested component of the server, and the plurality of aging data queues are obtained based on the same time period;

[0009] Use a difference operator to perform convolutional processing on each aging data queue in a sliding manner, divide each of the obtained plurality of first convolutional queues one by one, and normalize the divided queue segments to obtain a plurality of queue segments;

[0010] For each queue segment, perform a dot product calculation with each queue segment in the same time period and each queue segment of the same component respectively, to obtain a plurality of dot product results, and add the representative values determined according to the plurality of dot product results to the test analysis matrix. Among them, the queue segment in the same time period is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the queue segment of the same component is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed. The plurality of representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods.

[0011] Determine the abnormal probabilities of multiple tested components of the server according to the multiple row vectors of the test analysis matrix.

[0012] In a possible implementation manner, the convolution processing of each aging data queue by using the difference operator slidingly includes:

[0013] For each aging data queue, perform the following steps respectively:

[0014] Obtain the smoothing operator and the difference operator;

[0015] Successively extract a plurality of first intermediate data segments of the same type as the smoothing operator from multiple predetermined positions of the aging data queue;

[0016] Calculate the dot product of the smoothing operator and each first intermediate data segment respectively, and construct a plurality of calculation results into an intermediate queue according to the positions where the first intermediate data segments are taken out;

[0017] Take out a plurality of second intermediate data segments of the same type as the difference operator from the intermediate queue at a predetermined position interval;

[0018] Calculate the dot product of the difference operator and each second intermediate data segment, and construct a plurality of calculation results into a first convolution queue according to the positions where the second intermediate data segments are taken out.

[0019] In a possible implementation manner, the step of performing a dot product calculation with each queue segment in the same time period and each queue segment of the same component respectively for each queue segment, obtaining a plurality of dot product results, and adding the representative values determined according to the plurality of dot product results to the test analysis matrix includes:

[0020] For each queue segment, perform the following steps respectively:

[0021] Calculate the dot product of the queue segment with each queue segment in the plurality of queue segments in the same time period and the plurality of queue segments of the same component respectively, to obtain a plurality of dot product values;

[0022] Remove a plurality of minimum values of a preset quantity and a plurality of maximum values of a preset quantity from the plurality of dot product values;

[0023] The average value of multiple dot product values after calculating and removing multiple minimum values and multiple maximum values is used as a target element and added to the target row vector, where the multiple queue segments corresponding to the multiple elements in the row vector are from the same first convolution queue;

[0024] After the extraction of the target element is completed for multiple queue segments from the same first convolution queue, the row vector corresponding to the first convolution queue is added to the test analysis matrix.

[0025] In a possible implementation manner, the determining of the abnormal probabilities of multiple server components to be tested according to the multiple row vectors of the test analysis matrix includes:

[0026] Extract the row vector from the test analysis matrix and input it into the corresponding component abnormal probability model to obtain the abnormal probability of the test component, where the component abnormal probability model is:

[0027]

[0028] In the formula, P(E|a1a2…a n ) is the probability of the test component being abnormal under the condition of the row vector, a n is the nth element of the row vector, A is the row vector, n is the total number of elements in the row vector, W is the weight coefficient vector, and b is the bias constant;

[0029] If the abnormal probability is greater than the probability threshold, the component corresponding to the row vector is regarded as an abnormal component.

[0030] In a possible implementation manner, the weight coefficient vector and the bias constant of the probability model are determined according to multiple server aging test historical data, including:

[0031] Obtain multiple historical aging test results and multiple historical row vectors, where each historical aging test result corresponds to a historical row vector constructed by convolution, segmentation, and multiple dot products between queue segments based on the historical aging data queue;

[0032] Cluster the multiple historical row vectors to obtain multiple vector classes, where the total number of the multiple vector classes is equal to the sum of the total number of elements of the weight coefficient vector and the number of bias constants;

[0033] Take the class centers of the multiple vector classes as multiple historical class centers;

[0034] Bind the proportion of historical row vectors with abnormal aging test results in the vector class as an intermediate probability value to the corresponding historical class center to form multiple modeling data pairs;

[0035] Solve the weight coefficient vector and bias constant of the abnormal probability model according to the multiple modeling data pairs to obtain the component abnormal probability model.

[0036] In a possible implementation manner, the solving the weight coefficient vector and bias constant of the abnormal probability model according to the multiple modeling data pairs to obtain the component abnormal probability model includes:

[0037] Substitute the intermediate probability values in the multiple modeling data pairs into the first equation respectively to determine a plurality of first intermediate values, where the first equation is:

[0038]

[0039] In the formula, MV is the first intermediate value, and MP is the intermediate probability value;

[0040] Construct a matrix equation according to the plurality of first intermediate values and the historical class centers in the multiple modeling data pairs, where the matrix equation is:

[0041]

[0042] In the formula, MV n+1 is the (n + 1)-th first intermediate value, HA n+1,n is the n elements of the historical class center of the (n + 1)-th modeling data pair, and W n is the n-th element of the weight coefficient vector;

[0043] Solve the weight coefficient vector and bias constant according to the matrix equation to obtain the component abnormal probability model.

[0044] In a possible implementation manner, after determining the abnormal probabilities of multiple components under test of the server according to the multiple row vectors of the test analysis matrix, it includes:

[0045] If there is a component among the multiple components under test whose abnormal probability is greater than the risk threshold, enhance the aging test conditions of the component whose abnormal probability is greater than the risk threshold, and return to the step of obtaining multiple aging data queues.

[0046] In a second aspect, an embodiment of the present invention provides a server aging test device for implementing the server aging test method as described in the first aspect or any possible implementation manner of the first aspect. The server aging test device includes:

[0047] An aging data acquisition module, configured to acquire multiple aging data queues, where each aging data queue corresponds to a component under test of the server, and the multiple aging data queues are acquired based on the same time period;

[0048] An aging data segmentation module, configured to perform convolutional processing on each aging data queue by slidingly using a difference operator, segment each of the obtained multiple first convolutional queues one by one, and normalize the segmented queue segments to obtain multiple queue segments;

[0049] An aging data cross-analysis module, configured to perform dot product calculations on each queue segment with each same-time-period queue segment and each same-component queue segment respectively, obtain multiple dot product results, and add the representative values determined according to the multiple dot product results to a test analysis matrix, where the same-time-period queue segment is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the same-component queue segment is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed, and the multiple representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods;

[0050] And,

[0051] A component anomaly probability analysis module, configured to determine the anomaly probabilities of multiple server components to be tested according to multiple row vectors of the test analysis matrix.

[0052] In a third aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, where a computer program that can run on the processor is stored in the memory, and when the processor executes the computer program, the steps of the method described in the first aspect or any possible implementation manner of the first aspect above are implemented.

[0053] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method described in the first aspect or any possible implementation manner of the first aspect above are implemented.

[0054] The beneficial effects of the embodiment of the present invention compared with the prior art are:

[0055] An embodiment of the present invention discloses a server aging test method. First, a plurality of aging data queues are obtained, where each aging data queue corresponds to a component to be tested of the server, and the plurality of aging data queues are obtained based on the same time period. Then, each aging data queue is convoluted by using a differential operator in a sliding manner, and the obtained plurality of first convolution queues are segmented one by one, and the queue segments obtained by segmentation are normalized to obtain a plurality of queue segments. Then, for each queue segment, a dot product calculation is respectively performed with each queue segment in the same time period and each queue segment of the same component to obtain a plurality of dot product results, and the representative values determined according to the plurality of dot product results are added to a test analysis matrix. Among them, the queue segment in the same time period is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the queue segment of the same component is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed. The plurality of representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods. Finally, the abnormal probabilities of a plurality of components to be tested of the server are determined according to the plurality of row vectors of the test analysis matrix. The present invention extracts aging data queues based on a plurality of aging components tested in parallel, extracts test data features through differential convolution, queue segmentation, and cross dot product, and finally gives the probability of component abnormality based on the data features. In the embodiment of the present invention, aging tests are performed on multiple components synchronously, and the components with problems in the aging test process are determined through the method of aging data analysis. Therefore, the test efficiency is improved. Moreover, since the method of the present invention determines the state of the component to be tested based on cross dot product and probability estimation, problem components can be discovered early. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0057] Figure 1 is a flowchart of the server aging test method provided by the embodiment of the present invention;

[0058] Figure 2 is a schematic diagram of the process of constructing the test analysis matrix provided by the embodiment of the present invention;

[0059] Figure 3 is a functional block diagram of the server aging test device provided by the embodiment of the present invention;

[0060] Figure 4 is a functional block diagram of the electronic device provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0061] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present invention. However, those skilled in the art should understand that the present invention can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, and methods are omitted to avoid unnecessary details from interfering with the description of the present invention.

[0062] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will illustrate through specific embodiments in conjunction with the accompanying drawings.

[0063] The following provides a detailed description of the embodiments of the present invention. This example is implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following embodiments.

[0064] Figure 1 It is a flowchart of the server aging test method provided for the embodiments of the present invention.

[0065] As Figure 1 shown, it shows the implementation flowchart of the server aging test method provided for the embodiments of the present invention, which is described in detail as follows:

[0066] In step 101, a plurality of aging data queues are obtained, where each aging data queue corresponds to a tested component of the server, and the plurality of aging data queues are obtained based on the same time period.

[0067] In step 102, each aging data queue is convoluted by using a differential operator in a sliding manner, the obtained multiple first convolution queues are segmented one by one, and the segmented queue segments are normalized to obtain a plurality of queue segments.

[0068] In some embodiments, the convoluting each aging data queue by using a differential operator in a sliding manner includes:

[0069] For each aging data queue, the following steps are respectively executed:

[0070] Obtain a smoothing operator and the differential operator;

[0071] Extract a plurality of first intermediate data segments of the same type as the smoothing operator from a plurality of predetermined positions of the aging data queue in sequence;

[0072] Calculate the dot product of the smoothing operator and each first intermediate data segment respectively, and construct a plurality of calculation results into an intermediate queue according to the positions where the first intermediate data segments are taken;

[0073] Take out multiple second intermediate data segments of the same type as the difference operator from the middle queue at a predetermined position interval;

[0074] Calculate the dot product of the difference operator and each second intermediate data segment, and construct multiple calculation results into a first convolution queue according to the extraction position of the second intermediate data segment.

[0075] Exemplarily, the embodiment of the present invention applies loads to multiple components of the server simultaneously, obtains a data queue according to the loads, and further reflects the operating conditions of each component of the server. Due to the certain coupling in the operation of each component of the server, it is not easy to intuitively detect abnormal components from the aging data, especially early abnormalities.

[0076] The aging data queue is obtained based on the monitoring results of each component of the server. In addition to the traditional component temperature, current, and voltage items, it may also include the read / write speed data queue of the component, the working main frequency data queue, etc. Usually, one data queue corresponds to the working condition of one component.

[0077] Since the operation of components affects each other, for example, the change in the read / write speed may be affected by the read / write performance of the memory itself or the operating condition of the processor. Therefore, how to analyze the early abnormalities of the server through aging data and determine the causes of the server abnormalities is a difficult problem faced by the aging test with multiple components loaded simultaneously.

[0078] The embodiment of the present invention aims to use the aging data queue, through the convolution of the difference operator, to remove the influence of background factors (such as environmental temperature and humidity) as much as possible, and then extract the features contained in each aging data queue through cross-analysis methods (adjacent time-domain cross and cross between data queues), and construct an analysis matrix. Finally, an abnormal probability estimate of each component is made based on the analysis matrix.

[0079] Based on the above concept, the embodiment of the present invention first uses a smoothing operator and a difference operator to perform smoothing filtering and differencing processing on each aging data queue. In fact, these two processes are a continuous process. First, use the smoothing operator to slide and smooth the aging data queue to obtain a smoothed intermediate queue, and then use the difference operator to slide and perform differencing processing on the intermediate queue to obtain a convolution queue.

[0080] Specifically, in an application scenario, the smoothing operator has the following form:

[0081]

[0082] That is to say, the smoothing operator has a form where the sum of the operators is 1, and the absolute value of the central value is the largest, gradually decreasing towards both sides.

[0083] When the smoothing operator is applied, multiple data of the same type as the smoothing operator are taken from each position in the aging data queue, the dot product between the two is calculated, and the calculation result is added to the intermediate queue. Taking the above as an example, starting from the first position (starting position) of the aging data queue, five consecutive data are taken. These five data are subjected to dot product operation with the smoothing operator, and the result is used as the first data in the intermediate queue. Then, starting from the second position of the aging data queue, five consecutive data are taken, and the above steps of dot product operation and filling data into the intermediate queue are repeated. In this way, a complete intermediate queue is obtained.

[0084] The smoothing of the aging data queue by the smoothing operator can eliminate a part of the data noise, for example, the data fluctuations brought by the test background.

[0085] The intermediate queue can obtain the convolution queue after being processed by the difference operator convolution. The process is similar to the above process, and the intermediate queue is also processed by sliding the difference operator. A first-order difference operator:

[0086] [-1,1]

[0087] Or a second-order difference operator:

[0088] [1,-2,1]

[0089] For the sliding processing of the above intermediate queue, for example, when using the first-order difference operator, two data are taken from a certain position in the intermediate data queue. These two data are multiplied by the first-order difference operator, and the obtained dot product result is added to the convolution queue. Then, one position is offset from this position, and the above dot product operation is repeated again until all the data in the intermediate queue are processed, and the construction of the convolution queue is completed.

[0090] After the construction of the convolution queue is completed, it can be divided into multiple queue segments according to the preset length, and each queue segment is normalized to obtain the final queue segment. We can see that each aging data queue corresponds to multiple queue segments. Since the segmentation length is divided based on the preset length, therefore, for the same time period, there are multiple queue segments from different aging data queues.

[0091] In step 103, for each queue segment, dot product calculations are respectively performed with each queue segment in the same time period and each queue segment of the same component to obtain multiple dot product results, and the representative value determined according to the multiple dot product results is added to the test analysis matrix. Among them, the queue segment in the same time period is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the queue segment of the same component is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed. Multiple representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods.

[0092] In some implementations, for each queue segment, performing dot product calculations with each queue segment of the same period and each queue segment of the same component, respectively, to obtain multiple dot product results, and adding representative values determined according to the multiple dot product results to a test analysis matrix, includes:

[0093] For each queue segment, perform the following steps:

[0094] respectively calculating the dot product of the queue segment with the multiple same-segment queue segments and each queue segment in the multiple same-component queue segments to obtain multiple dot product values;

[0095] removing a preset number of minimum values and removing a preset number of maximum values from the plurality of dot product values;

[0096] Adding the average of multiple dot product values after removing multiple minimum values and multiple maximum values as the target element to the target row vector, wherein the multiple queue segments corresponding to the multiple elements in the row vector are derived from the same first convolution queue;

[0097] After multiple queue segments originating from the same first convolution queue complete the extraction of target elements, the row vector corresponding to the first convolution queue is added to the test analysis matrix.

[0098] For example, Figure 2 As shown, the figure shows a schematic diagram of the test analysis matrix construction process provided by an embodiment of the present invention. The embodiment of the present invention obtains queue segments 202 by dividing the first convolution queue 201, and the queue segments 202 are dot-producted with each other, and representative values 203 are obtained according to the results of the dot product, and a test analysis matrix is constructed by the representative values 203.

[0099] Specifically, the cross-comparison analysis of queue segments can identify queue segments that are significantly different from other queue segments. For example, after the dot product of some queue segments with other queue segments, the values obtained are smaller (after the above process, the dot product results of two arbitrary queue segments are distributed between -1 and 1. The closer the value is to 0, the smaller the correlation between the two queue segments. The closer the value is to -1, the more it indicates that there is a negative correlation between the two queues). This means that these queue segments are significantly different from other queue segments.

[0100] In fact, the dot product of the queue segments should follow the adjacent principle. If the data segments are obtained from the same aging data queue, then they should be two adjacent data segments in the time domain, that is, two data segments corresponding to adjacent time periods; while two data segments not based on the same aging data queue should correspond to the same time period. After processing according to the above principles, multiple dot product results can be obtained for each queue segment. Since the dot product results will be affected by abnormal queue segments, in the embodiments of the present invention, the maximum multiple dot product results and the minimum multiple dot product results are removed from the multiple dot product results. For example, each queue segment can obtain 16 dot product results. By removing two maximum values and two minimum values from these 16 dot product results, 12 dot product results are obtained, and then the average value of these 12 dot product results is calculated, and the obtained value is the representative value.

[0101] As we can see, each aging data queue corresponds to multiple queue segments, and each queue segment corresponds to a representative value. The multiple representative values of each aging data queue are arranged in a row vector in chronological order, and multiple row vectors are arranged vertically to obtain a matrix. In fact, this matrix is the differential feature of multiple aging data queues. By analyzing the row vectors of the matrix, the probability of abnormality of the measured component can be estimated.

[0102] In step 104, the abnormality probability of multiple measured components of the server is determined according to the multiple row vectors of the test analysis matrix.

[0103] In some embodiments, the determining the abnormality probability of multiple measured components of the server according to the multiple row vectors of the test analysis matrix includes:

[0104] Extract the row vectors from the test analysis matrix and input them into the corresponding component abnormality probability model to obtain the abnormality probability of the test component, where the component abnormality probability model is:

[0105]

[0106] In the formula, P(E|a1a2…a n ) is the probability of the test component being abnormal under the condition of the row vector, a n is the nth element of the row vector, A is the row vector, n is the total number of elements in the row vector, W is the weight coefficient vector, and b is the bias constant;

[0107] If the abnormality probability is greater than the probability threshold, the component corresponding to the row vector is regarded as an abnormal component.

[0108] In some embodiments, the weight coefficient vector and the bias constant of the probability model are determined according to multiple server aging test historical data, including:

[0109] Obtain a plurality of historical aging test results and a plurality of historical row vectors, where each historical aging test result corresponds to a historical row vector constructed by convolving, segmenting, and performing multiple dot products between queue segments on a historical aging data queue;

[0110] Cluster the plurality of historical row vectors to obtain a plurality of vector classes, where the total number of the plurality of vector classes is equal to the sum of the total number of elements of the weight coefficient vector and the number of bias constants;

[0111] Use the class centers of the plurality of vector classes as a plurality of historical class centers;

[0112] Bind the proportion of historical row vectors with abnormal aging test results in the vector class as an intermediate probability value to the corresponding historical class center to form a plurality of modeling data pairs;

[0113] Solve the weight coefficient vector and the bias constant of the abnormal probability model according to the plurality of modeling data pairs to obtain the component abnormal probability model.

[0114] In some embodiments, the step of solving the weight coefficient vector and the bias constant of the abnormal probability model according to the plurality of modeling data pairs to obtain the component abnormal probability model includes:

[0115] Substitute the intermediate probability values in the plurality of modeling data pairs into the first equation respectively to determine a plurality of first intermediate values, where the first equation is:

[0116]

[0117] where MV is the first intermediate value and MP is the intermediate probability value;

[0118] Construct a matrix equation according to the plurality of first intermediate values and the historical class centers in the plurality of modeling data pairs, where the matrix equation is:

[0119]

[0120] where MV n+1 is the (n + 1)-th first intermediate value, HA n+1,n is the n elements of the historical class center of the (n + 1)-th modeling data pair, and W n is the n-th element of the weight coefficient vector;

[0121] Solve the weight coefficient vector and the bias constant according to the matrix equation to obtain the component abnormal probability model.

[0122] Exemplarily, the abnormal probability of the component under test in the embodiments of the present invention is obtained based on row vector analysis. In fact, it is determined by inputting the row vector into the abnormal probability model (abnormal probability estimation model) of the corresponding component. The abnormal probability model is as follows:

[0123]

[0124] In the formula, P(E|a1a2…a n ) is the probability of the test component being abnormal under the condition of the row vector, a n is the nth element of the row vector, A is the row vector, n is the total number of elements in the row vector, W is the weight coefficient vector, and b is the bias constant.

[0125] There are two undetermined parameters in this model, namely the weight coefficient vector and the bias constant. These two undetermined parameters are finally determined based on statistical data, specifically historical aging test results and multiple historical row vectors. It should be noted that the acquisition process of the historical row vectors is the same as that of the row vectors of the above matrix (test analysis matrix), and will not be elaborated here.

[0126] When determining the two undetermined parameters, first cluster multiple historical row vectors. For example, use the DBSCAN clustering method to cluster these multiple historical row vectors into multiple classes. For example, cluster them into C1 - C m classes. For each class, we then count the proportion (intermediate probability value) of those with abnormal historical aging test results. For example, it is MP1 - MP m . Thus, we obtain multiple pairs, the class centers CC1 - CC m corresponding to the proportion pairs MP1 - MP m (intermediate probability value). Based on these pairs, solve the two undetermined parameters.

[0127] In fact, after obtaining these pairs, first transform the proportions in the pairs, and then based on the transformation results, construct a matrix equation to determine the solutions of the two undetermined parameters. When transforming, apply the first equation:

[0128]

[0129] In the formula, MV is the first intermediate value, and MP is the intermediate probability value.

[0130] Then, based on the transformed values, construct a matrix equation:

[0131]

[0132] In the formula, MV n+1 is the (n + 1)th first intermediate value, HA n+1,n is the n elements of the historical class center of the (n + 1)th modeling data pair, Wn is the n-th element of the weight coefficient vector.

[0133] According to the above equation, we can determine two parameters and finally complete the construction of the component anomaly probability model.

[0134] In addition, in some embodiments, after determining the anomaly probabilities of multiple server components to be tested according to multiple row vectors of the test analysis matrix, it includes:

[0135] If there is a component among the multiple components to be tested whose anomaly probability is greater than the risk threshold, enhance the aging test conditions of the component whose anomaly probability is greater than the risk threshold, and return to the step of obtaining multiple aging data queues.

[0136] Exemplarily, after the above probability estimation, if we detect that some components may be abnormal (based on probability estimation and estimation results), then we can increase the aging test conditions of the component, such as increasing the main frequency of the test and increasing the test voltage. After increasing the aging test conditions, return to the original step of the aging data queue. Obviously, doing so can detect abnormal components earlier on the basis of parallel testing of multiple components.

[0137] In the embodiment of the server aging test method of the present invention, it first obtains multiple aging data queues, where each aging data queue corresponds to a component to be tested of the server, and the multiple aging data queues are obtained based on the same time period; then uses the difference operator to perform convolutional processing on each aging data queue in a sliding manner, divides each of the obtained multiple first convolutional queues one by one, and normalizes the divided queue segments to obtain multiple queue segments; then for each queue segment, perform dot product calculations with each queue segment in the same time period and each queue segment of the same component respectively to obtain multiple dot product results, and add the representative values determined according to the multiple dot product results to the test analysis matrix, where the queue segment in the same time period is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the queue segment of the same component is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed, and multiple representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods; finally, determine the anomaly probabilities of multiple server components to be tested according to multiple row vectors of the test analysis matrix. The present invention extracts aging data queues based on multiple parallelly tested aging components, extracts test data features through the methods of differential convolution, queue segmentation, and cross dot product, and finally gives the probability of component anomalies based on the data features. In the embodiment of the present invention, multiple components are aged and tested synchronously, and the components with problems during the aging test are determined through the method of aging data analysis. Therefore, the test efficiency is improved, and since the method of the present invention determines the state of the component to be tested based on cross dot product and probability estimation, problem components can be discovered early.

[0138] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not imply the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0139] The following are the device embodiments of the present invention. For the details not described in detail herein, reference may be made to the corresponding method embodiments above.

[0140] Figure 3 is a functional block diagram of a server aging test device provided by an embodiment of the present invention. Referring to Figure 3 , the server aging test device includes: an aging data acquisition module 301, an aging data segmentation module 302, an aging data cross-analysis module 303, and a component anomaly probability analysis module 304, where:

[0141] The aging data acquisition module 301 is configured to acquire a plurality of aging data queues, where each aging data queue corresponds to a DUT (Device Under Test) component of the server, and the plurality of aging data queues are acquired based on the same time period;

[0142] The aging data segmentation module 302 is configured to perform convolutional processing on each aging data queue by sliding the difference operator, divide each of the obtained plurality of first convolutional queues one by one, and normalize the divided queue segments to obtain a plurality of queue segments;

[0143] The aging data cross-analysis module 303 is configured to perform dot product calculations on each queue segment with each same-time-period queue segment and each same-component queue segment respectively, obtain a plurality of dot product results, and add the representative values determined according to the plurality of dot product results to a test analysis matrix, where the same-time-period queue segment is a queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the same-component queue segment is a queue segment with the same source queue as the queue segment for which the dot product calculation is performed. The plurality of representative values corresponding to the same test component are added to the same row of the matrix in the order of time period;

[0144] The component anomaly probability analysis module 304 is configured to determine the anomaly probabilities of a plurality of DUT components of the server according to the plurality of row vectors of the test analysis matrix.

[0145] Figure 4 is a functional block diagram of an electronic device provided by an embodiment of the present invention. As Figure 4 shown, the electronic device 4 of this embodiment includes: a processor 400 and a memory 401, and a computer program 402 that can run on the processor 400 is stored in the memory 401. When the processor 400 executes the computer program 402, the steps in the above various server aging test methods and embodiments are implemented, such as Figure 1 the steps 101 to 104 shown.

[0146] Exemplarily, the computer program 402 can be divided into one or more modules / units, which are stored in the memory 401 and executed by the processor 400 to implement the present invention.

[0147] The electronic device 4 can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The electronic device 4 may include, but is not limited to, a processor 400 and a memory 401. Those skilled in the art can understand that Figure 4 merely examples of the electronic device 4, which do not constitute a limitation on the electronic device 4, may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the electronic device 4 may further include input / output devices, network access devices, a bus, etc.

[0148] The so-called processor 400 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0149] The memory 401 may be an internal storage unit of the electronic device 4, such as a hard disk or a memory of the electronic device 4. The memory 401 may also be an external storage device of the electronic device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the electronic device 4. Further, the memory 401 may also include both the internal storage unit and the external storage device of the electronic device 4. The memory 401 is used to store the computer program 402 and other programs and data required by the electronic device 4. The memory 401 may also be used to temporarily store data that has been output or will be output.

[0150] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of the functional units and modules are only for the convenience of mutual distinction and do not limit the protection scope of the present application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0151] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0152] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in the form of hardware or software depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0153] In the embodiments provided by the present invention, it should be understood that the disclosed device / electronic device and method can be implemented in other ways. For example, the device / electronic device embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical, mechanical or other form.

[0154] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0155] In addition, each functional unit in various embodiments of the present invention may be integrated into one processing unit, may exist physically alone for each unit, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0156] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, all or part of the processes in the above-described embodiments of the present invention may also be completed by a computer program instructing relevant hardware. The computer program may be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method and apparatus embodiments may be implemented. Among them, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0157] The above-described embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention and should all be included in the protection scope of the present invention.

Claims

1. A server aging test method, characterized in that Including: Obtain multiple aging data queues, where each aging data queue corresponds to a component under test of the server, and the multiple aging data queues are obtained based on the same time period; Use a differential operator to perform convolutional processing on each aging data queue in a sliding manner, divide each of the obtained multiple first convolutional queues one by one, and normalize the divided queue segments to obtain multiple queue segments; For each queue segment, perform dot product calculations with each queue segment in the same time period and each queue segment of the same component respectively, obtain multiple dot product results, and add the representative values determined according to the multiple dot product results to the test analysis matrix, where the queue segments in the same time period are the queue segments in the same time period as the queue segment for which the dot product calculation is performed, and the queue segments of the same component are the queue segments with the same source queue as the queue segment for which the dot product calculation is performed. The multiple representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods; Determine the anomaly probabilities of multiple components under test of the server according to the multiple row vectors of the test analysis matrix.

2. The server aging test method according to claim 1, wherein The step of using a differential operator to perform convolutional processing on each aging data queue in a sliding manner includes: For each aging data queue, perform the following steps respectively: Obtain a smoothing operator and the differential operator; Successively extract multiple first intermediate data segments of the same type as the smoothing operator from multiple predetermined positions of the aging data queue; Calculate the dot product of the smoothing operator and each first intermediate data segment respectively, and construct multiple calculation results into an intermediate queue according to the extraction positions of the first intermediate data segments; Extract multiple second intermediate data segments of the same type as the differential operator from the intermediate queue at a predetermined position interval; Calculate the dot product of the differential operator and each second intermediate data segment, and construct multiple calculation results into a first convolutional queue according to the extraction positions of the second intermediate data segments.

3. The server aging test method according to claim 1, characterized in that The step of, for each queue segment, performing dot product calculations with each queue segment in the same time period and each queue segment of the same component respectively, obtaining multiple dot product results, and adding the representative values determined according to the multiple dot product results to the test analysis matrix includes: For each queue segment, perform the following steps respectively: Calculate the dot product of the queue segment with each queue segment among the multiple queue segments in the same time period and the multiple queue segments of the same component respectively, to obtain multiple dot product values; Remove a preset number of multiple minimum values and a preset number of multiple maximum values from the multiple dot product values; Take the average value of the multiple dot product values after removing the multiple minimum values and the multiple maximum values as the target element and add it to the target row vector, where the multiple queue segments corresponding to the multiple elements in the row vector come from the same first convolutional queue; After the extraction of the target elements for all queue segments originating from the same first convolutional queue is completed, add the row vector corresponding to the first convolutional queue to the test analysis matrix.

4. The server aging test method according to any one of claims 1-3, characterized in that, The step of determining the anomaly probabilities of multiple components under test of the server according to the multiple row vectors of the test analysis matrix includes: Extract the row vectors from the test analysis matrix and input them into the corresponding component anomaly probability model to obtain the anomaly probability of the test component, where the component anomaly probability model is: Wherein, P(E|a1a2…a n ) is the probability of the abnormal test component under the condition of the row vector, a n is the nth element of the row vector, A is the row vector, n is the total number of elements in the row vector, W is the weight coefficient vector, and b is the bias constant; If the abnormal probability is greater than the probability threshold, the component corresponding to the row vector is regarded as an abnormal component.

5. The server aging test method according to claim 4, wherein The weight coefficient vector and the bias constant of the probability model are determined according to multiple historical data of server aging tests, including: Obtain multiple historical aging test results and multiple historical row vectors, where each historical aging test result corresponds to a historical row vector constructed by convolution, segmentation, and multiple dot products between queue segments based on a historical aging data queue; Cluster the multiple historical row vectors to obtain multiple vector classes, where the total number of the multiple vector classes is equal to the sum of the total number of elements of the weight coefficient vector and the number of bias constants; Take the class centers of the multiple vector classes as multiple historical class centers; Bind the proportion of historical row vectors with abnormal aging test results in the vector class as an intermediate probability value to the corresponding historical class center to form multiple modeling data pairs; Solve the weight coefficient vector and the bias constant of the abnormal probability model according to the multiple modeling data pairs to obtain the component abnormal probability model.

6. The server aging test method according to claim 5, wherein The solving of the weight coefficient vector and the bias constant of the abnormal probability model according to the multiple modeling data pairs to obtain the component abnormal probability model includes: Substitute the intermediate probability values in the multiple modeling data pairs into the first equation respectively to determine multiple first intermediate values, where the first equation is: In the formula, MV is the first intermediate value, and MP is the intermediate probability value; Construct a matrix equation according to the multiple first intermediate values and the historical class centers in the multiple modeling data pairs, where the matrix equation is: where, MV n+1 is the (n + 1)-th first intermediate value, HA n+1,n is the n elements of the historical class center of the (n + 1)-th modeling data pair, and W n is the n-th element of the weight coefficient vector; Solve the weight coefficient vector and the bias constant according to the matrix equation to obtain the component abnormal probability model.

7. The server aging test method according to claim 1, characterized in that After determining the abnormal probabilities of multiple components to be tested of the server according to the multiple row vectors of the test analysis matrix, it includes: If there are components with abnormal probabilities greater than the risk threshold among the multiple components to be tested, enhance the aging test conditions of the components with abnormal probabilities greater than the risk threshold and return to the step of obtaining multiple aging data queues.

8. A server aging test device, characterized in that For implementing the server aging test method according to any one of claims 1-7, the server aging test device includes: An aging data acquisition module for acquiring multiple aging data queues, where each aging data queue corresponds to a component to be tested of the server, and the multiple aging data queues are acquired based on the same time period; An aging data segmentation module for performing convolutional processing on each aging data queue by sliding the difference operator, dividing each of the obtained multiple first convolutional queues one by one, and normalizing the queue segments obtained by the division to obtain multiple queue segments; An aging data cross-analysis module is used to perform dot product calculations for each queue segment with each same-time-period queue segment and each same-component queue segment respectively, obtain multiple dot product results, and add the representative values determined according to the multiple dot product results to a test analysis matrix, where the same-time-period queue segment is the queue segment in the same time period as the queue segment for which the dot product calculation is performed, and the same-component queue segment is the queue segment with the same source queue as the queue segment for which the dot product calculation is performed. The multiple representative values corresponding to the same test component are added to the same row of the matrix in the order of time periods. And, A component anomaly probability analysis module is used to determine the anomaly probabilities of multiple server under-test components according to the multiple row vectors of the test analysis matrix.

9. An electronic device, comprising a memory and a processor, wherein a computer program capable of running on the processor is stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 7 above.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1 to 7 above.

Images