Computer fault alarm method and system

Abstract

The invention discloses a computer fault alarm method and system, and relates to the technical field of fault alarms, and the method comprises the steps: under the condition that the current to-be-measured temperature values are different, if it is judged that fault hardware with the earliest time point of a fault external display phenomenon and the minimum current to-be-measured temperature value exists, determining that the fault hardware is the fault hardware with the earliest time point of the fault external display phenomenon; if the current to-be-detected temperature values are the same, counting fault correlation degree values among the current fault hardware so as to judge the original fault hardware, and if the current to-be-detected temperature values are the same, judging the original fault hardware according to the number of fault nodes and the weight node ratio in the fault hardware to which the current fault hardware belongs, and sending the original fault hardware to the current to-be-detected temperature values to the current to-be-detected temperature values. And under the condition that part of the current temperature values to be measured are the same and part of the current temperature values to be measured are different, node stability degree values of all nodes in each piece of fault hardware in the current fault hardware are evaluated so as to judge the original fault hardware. According to the computer fault alarm method and system provided by the invention, the efficiency of fault alarm work can be improved.

Description

Technical Field

[0001] This application relates to the technical field of fault alarm, and particularly relates to a computer fault alarm method and system. Background Art

[0002] With the rapid development of the information age, people use computers more and more frequently. When a computer works for a long time, there is a risk of high temperature of some hardware, which may cause failures of the internal hardware of the computer, thus affecting the work efficiency of employees. Therefore, the fault detection and timely alarm of the computer are crucial.

[0003] In the prior art, during the detection and analysis of computer faults, when multiple hardware components inside the computer show different fault symptoms, the first faulty hardware component with the earliest appearance of the fault symptom is determined as the origin faulty hardware component. For the above two judgment situations, there is a large subjective judgment without verification, resulting in a relatively high probability of misjudgment of the final fault judgment result and reducing the efficiency of the entire fault alarm work. Summary of the Invention

[0004] To overcome the deficiencies of the prior art described above, this application provides a computer fault alarm method and system.

[0005] In a first aspect, a computer fault alarm method provided by this application includes:

[0006] Detect the current faulty hardware and the current temperature value to be measured of the current faulty hardware. If all the temperature values in the current temperature values to be measured are different, then determine the faulty hardware with the earliest occurrence time point of the fault manifestation and the largest current temperature value to be measured as the origin faulty hardware, and output the first type of alarm result one;

[0007] If all the temperature values in the current temperature values to be measured are different, and it is determined that there is a faulty hardware with the earliest occurrence time point of the fault manifestation and the smallest current temperature value to be measured, then count the fault correlation degree values between the faulty hardware components in the current faulty hardware, and determine the origin faulty hardware according to the fault correlation degree values, and output the first type of alarm result two;

[0008] If all the temperature values in the current temperature values to be measured are the same, according to the counted number of fault nodes and the weight node ratio in each faulty hardware component to which the current faulty hardware belongs, predict the hardware fault prediction time sequence of the occurrence of the fault phenomenon of the current faulty hardware. If the comparison result of the starting fault time sequence is the same as the comparison between the time sequences of the occurrence of the fault phenomenon of multiple faulty hardware components existing in the current faulty hardware, then determine the faulty hardware to which the starting fault time sequence belongs as the origin faulty hardware, and output the second type of alarm result one;

[0009] If the comparison results of the mid-end fault timings are the same, and the comparison results of the start-end fault timing and the end-end fault timing are opposite, then based on the position characteristics of the weighted fault nodes in the faulty hardware with opposite comparison results statistically obtained, the originating faulty hardware is determined, and the alarm result two of the second case is output;

[0010] When there are both some identical temperature values and some different temperature values in the current temperature value to be measured, the node stability degree values of all nodes in each faulty hardware in the current faulty hardware are evaluated, and based on the node stability degree values, the originating faulty hardware is determined, and the alarm result of the third case is output.

[0011] Preferably, the current faulty hardware with the externally manifested fault phenomenon inside the faulty computer is detected, and the current temperature value to be measured is obtained by detecting the temperature data of the current faulty hardware;

[0012] When there are multiple faulty hardwares in the current faulty hardware, and the multiple faulty hardwares in the current faulty hardware are all externally manifested fault phenomena occurring at different time points, then the situation to be measured is output;

[0013] According to the situation to be measured, when all the temperature values in the current temperature value to be measured are different, if it is judged that the time point when one of the faulty hardwares in the current faulty hardware has the externally manifested fault phenomenon is the earliest, and the current temperature value to which it belongs is the largest, then the origin fault determination condition one is output;

[0014] According to the origin fault determination condition one, one of the faulty hardwares belonging to the origin fault determination condition one is determined as the originating faulty hardware, and an alarm notification is made, and the alarm result one of the first case is output.

[0015] Preferably, when all the temperature values in the current temperature value to be measured are different, if it is judged that the time point when one of the faulty hardwares in the current faulty hardware has the externally manifested fault phenomenon is the earliest, and the current temperature value to which it belongs is the smallest, then the origin fault determination condition two is output;

[0016] The characteristic correlation data is obtained by statistically calculating the lengths of the connection lines between various faulty hardwares in the current faulty hardware and the curvatures of the connection lines between them;

[0017] Based on the characteristic correlation data, the fault transfer correlation degree values between various faulty hardwares in the current faulty hardware are evaluated to obtain the fault correlation degree value;

[0018] If the failure transfer correlation degree value corresponding to a failure hardware in the origin failure determination condition 2 is the maximum value among the failure correlation degree values, then the failure hardware in the origin failure determination condition 2 is determined as the origin failure hardware, and an alarm notification is sent to output the alarm result 2 of the first type of situation.

[0019] Preferably, according to the situation to be measured, when all the temperature values in the current temperature value to be measured are the same, count the number of failure nodes with data processing speed failures among all the nodes performing data processing in each failure hardware to which the current failure hardware belongs, and count the ratio of the weight nodes in the failure nodes to obtain the weight node ratio;

[0020] Obtain the historical failure node characteristics of the current failure hardware for detecting failure node data in the historical period, and establish a failure prediction model 1 according to the historical failure node characteristics;

[0021] Input the number of failure nodes, the weight node ratio, and the current failure hardware into the failure prediction model 1 for testing to obtain the hardware failure prediction time series;

[0022] Compare the hardware failure prediction time series with the time series of the external failure phenomena of each hardware in the situation to be measured to obtain a comparison result 1;

[0023] If the comparison results of the start-end failure time series in the comparison result 1 are the same, then determine the failure hardware to which the start-end failure time series belongs as the origin failure hardware, and send an alarm notification to output the alarm result 1 of the second type of situation.

[0024] Preferably, if only the comparison results of the end failure time series in the comparison result 1 are the same, then input the failure correlation degree value into the failure prediction model 1 for secondary testing to output the secondary prediction information of the failure time series;

[0025] If the comparison results of the middle-end failure time series in the comparison result 1 are the same, and the comparison results of the start-end failure time series and the end failure time series in the comparison result 1 are opposite, then detect the position of the weighted failure nodes with data processing speed failures in the failure hardware with opposite comparison results to obtain the weighted node position characteristics;

[0026] Determine the failure hardware with the opposite comparison result to which the weighted failure node with the most forward position in the weighted node position characteristics belongs as the origin failure hardware, and send an alarm notification to output the alarm result 2 of the second type of situation.

[0027] Preferably, according to the situation to be measured, when there are not only some identical temperature values but also some different temperature values in the current temperature value to be measured, obtain the historical temperature fault degree characteristics of the current faulty hardware for detecting the influence of different temperature values on the hardware fault degree in the historical period, and establish the second fault prediction model according to the historical temperature fault degree characteristics;

[0028] Input the current temperature value to be measured into the second fault prediction model for testing to obtain the current temperature fault influence degree;

[0029] According to the current temperature fault influence degree, evaluate the stability degree value of the processing data speed of all nodes for data processing in each faulty hardware in the current faulty hardware to obtain the node stability degree value;

[0030] Arbitrarily select a faulty hardware from the current faulty hardware as the starting hardware to be measured, and regard the remaining faulty hardware in the current faulty hardware except the starting hardware to be measured as the influencing hardware to be measured;

[0031] According to the node stability degree value, predict the process mode after node switching in the starting hardware to be measured and the influencing hardware to be measured to obtain the to-be-tested node switching characteristics;

[0032] Detect the current node switching characteristics formed between all faulty hardwares in the current faulty hardware, and compare the to-be-tested node switching characteristics with the current node switching characteristics to obtain the second comparison result;

[0033] Determine the starting hardware to be measured with a same comparison result in the second comparison result as the origin faulty hardware, and give an alarm notification to output the alarm result of the third type of situation.

[0034] In a second aspect, a computer fault alarm system includes:

[0035] The determination situation one of the origin faulty hardware is used to detect the current faulty hardware and the current temperature value to be measured of the current faulty hardware. If all the temperature values in the current temperature value to be measured are different, then determine the faulty hardware with the earliest time point of the fault manifestation and the largest current temperature value to be measured as the origin faulty hardware, and output the first type of alarm result one;

[0036] The fault correlation analysis unit is used to, if all the temperature values in the current temperature value to be measured are different, judge that there is a faulty hardware with the earliest time point of the fault manifestation and the smallest current temperature value to be measured, then count the fault correlation degree value between each faulty hardware in the current faulty hardware, and judge the origin faulty hardware according to the fault correlation degree value, and output the first type of alarm result two;

[0037] The fault node feature analysis unit is used to, when all temperature values in the current temperature value to be measured are the same, predict the hardware fault prediction timing of the fault phenomenon occurring in the current faulty hardware according to the number of fault nodes and the weight node ratio within each faulty hardware to which the current faulty hardware belongs. If the comparison result of the starting-end fault timing is the same as that between the timings of the fault phenomena occurring in multiple faulty hardwares existing in the current faulty hardware, it is determined that the faulty hardware to which the starting-end fault timing belongs is the originating faulty hardware, and the alarm result one of the second type is output.

[0038] The second situation for determining the originating faulty hardware is used to judge that if the comparison results of the middle-end fault timings are the same and the comparison results of the starting-end fault timings and the ending-end fault timings are opposite, the originating faulty hardware is determined according to the weight node position characteristics of the weighted fault nodes in the faulty hardware with opposite comparison results, and the alarm result two of the second type is output.

[0039] The third situation for determining the originating faulty hardware is used to, when there are not only some identical temperature values but also some different temperature values in the current temperature value to be measured, evaluate the node stability degree values of all nodes within each faulty hardware in the current faulty hardware, and determine the originating faulty hardware according to the node stability degree values, and output the alarm result of the third type.

[0040] Compared with the prior art, the present invention has the following features and beneficial effects:

[0041] By making a diversity judgment on the temperature environment conditions of multiple hardware components with external failure phenomena inside the computer, such as the cases where the respective temperature values are the same, the respective temperature values are all different, and the situation where the temperature values of some of the faulty hardware are the same while the temperature values of some of the faulty hardware are different, by separately performing statistical analysis on the data characteristics of the differential influencing factors for the current faulty hardware in these three categories of situations to determine the origin faulty hardware, and sending alarm notifications for the origin faulty hardware determined in each situation, in order to improve the timeliness of taking targeted measures for the origin faulty hardware. Through the above judgment of diversity situations, the rigor of the entire data processing process is enhanced to reduce the probability of misjudgment of the final alarm result for the origin faulty hardware. And for the current faulty hardware within the above diversity situations, different influencing factors are separately detected and analyzed. For example, in one case, the relevance of the origin faulty hardware is judged based on the time sequence of the external failure phenomena of each faulty hardware in the current faulty hardware and the magnitude values of the respective temperature and humidity. In one case, the relevance of the origin faulty hardware is judged based on the number of faulty nodes and the weight node ratio for data processing within each faulty hardware in the current faulty hardware. In another case, the relevance of the origin faulty hardware is judged based on the impact of the temperature value to which the current faulty hardware belongs on the node stability value within each faulty hardware, resulting in the node switching process method. Through the above multiple differential data processing methods, multiple verification processes are carried out on the final judgment of the origin faulty hardware and the timely alarm processing result, avoiding the low efficiency of the entire fault detection and alarm work caused by only detecting and analyzing single characteristic information for the origin faulty hardware in traditional technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a block diagram of the steps of a computer fault alarm method mainly embodied in this embodiment.

[0043] Figure 2 is a block diagram of the structure of a computer fault alarm system mainly embodied in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] The present invention will be further described in detail below with reference to the following embodiments.

[0045] Refer to Figure 1 , a computer fault alarm method, which includes the following steps:

[0046] S1. Detect the current faulty hardware and the current temperature value to be measured of the current faulty hardware. If all the temperature values in the current temperature value to be measured are different, then determine the faulty hardware with the earliest time point of the external failure phenomenon and the largest current temperature value to be measured as the origin faulty hardware, and output the first type of alarm result one.

[0047] S2. If, when all the temperature values in the currently measured temperature values are different, it is determined that there is a faulty hardware component with the earliest time point of the appearance of the fault manifestation and the smallest temperature value among the currently measured temperature values, then the fault correlation degree values between the faulty hardware components in the current faulty hardware are statistically calculated. Based on the fault correlation degree values, the originating faulty hardware is determined, and the alarm result two of the first type of situation is output.

[0048] S3. If all the temperature values in the currently measured temperature values are the same, then based on the statistically calculated number of fault nodes and the weight node ratio within each faulty hardware component to which the current faulty hardware belongs, the hardware fault prediction timing of the occurrence of the fault phenomenon of the current faulty hardware is predicted. If the comparison result of the starting - end fault timing is the same as the comparison result between the timings of the occurrence of the fault phenomenon of multiple faulty hardware components existing in the current faulty hardware, then the faulty hardware to which the starting - end fault timing belongs is determined as the originating faulty hardware, and the alarm result one of the second type of situation is output.

[0049] S4. If the comparison result of the middle - end fault timing is the same, and the comparison results of the starting - end fault timing and the ending - end fault timing are opposite, then based on the weight node position characteristics of the weighted fault nodes in the faulty hardware with opposite comparison results statistically calculated, the originating faulty hardware is determined, and the alarm result two of the second type of situation is output.

[0050] S5. When there are not only some identical temperature values but also some different temperature values in the currently measured temperature values, the node stability degree values of all nodes within each faulty hardware component in the current faulty hardware are evaluated. Based on the node stability degree values, the originating faulty hardware is determined, and the alarm result of the third type of situation is output.

[0051] Specifically, by diversely judging the temperature environment conditions of multiple hardware components with external failure phenomena inside the computer, such as the cases where their respective temperature values are the same, where their respective temperature values are all different, and the case where both the situation where the temperature values of some of the faulty hardware components are the same and the situation where the temperature values of some of the faulty hardware components are different exist simultaneously. By separately conducting statistical analysis on the data characteristics of the differential influencing factors for the current faulty hardware in these three categories of situations to determine the originating faulty hardware, and sending alarm notifications for the originating faulty hardware determined in each situation, the timeliness of taking targeted measures for the originating faulty hardware is improved. Through the above-mentioned judgment of diverse situations, the rigor of the entire data processing process is enhanced to reduce the probability of misjudgment of the final judgment and alarm results for the originating faulty hardware. Moreover, for the current faulty hardware within the above-mentioned diverse situations, different influencing factors are separately detected and analyzed. For example, in one situation, the relevance of the originating faulty hardware is judged based on the time sequence of the external failure phenomena of each faulty hardware in the current faulty hardware and the magnitude values of the temperature and humidity to which they belong. In another situation, the relevance of the originating faulty hardware is judged based on the number of faulty nodes and the weight node ratio for data processing within each faulty hardware in the current faulty hardware. In still another situation, the relevance of the originating faulty hardware is judged based on the influence of the temperature value to which the current faulty hardware belongs on the node stability degree value within each faulty hardware, resulting in the node switching process method. Through the above-mentioned multiple differential data processing methods, multiple verification processes are carried out on the final judgment of the originating faulty hardware and the timely alarm processing results, avoiding the low efficiency of the entire fault detection and alarm work caused by only detecting and analyzing single characteristic information for the originating faulty hardware in traditional technologies.

[0052] Specifically, step S1 includes the following sub-steps:

[0053] Detect the current faulty hardware with external failure phenomena inside the faulty computer, and conduct temperature data detection on the current faulty hardware to obtain the current temperature value to be measured.

[0054] When there are multiple faulty hardware components in the current faulty hardware, and the external failure phenomena of the multiple faulty hardware components in the current faulty hardware occur at different time points, then output the situation to be measured.

[0055] According to the situation to be measured, in the case where all the temperature values in the current temperature value to be measured are different, if it is judged that the time point when one of the faulty hardware components in the current faulty hardware has an external failure phenomenon is the earliest, and the current temperature value to which it belongs is the largest, then output the first origin fault determination condition.

[0056] According to the first origin fault determination condition, determine the one faulty hardware component belonging to the first origin fault determination condition as the originating faulty hardware, and send an alarm notification, and output the first type of situation alarm result one.

[0057] Specifically, the externally manifested fault phenomena here refer to faults that can be directly observed, such as data processing lag, memory quality problems, data reading error problems, etc. For the situation to be measured (for example, the current faulty hardware includes A, B, C, D, and E, and if the time points of the fault phenomena of these five faulty hardware are all different), when all the temperature values in the current measured temperature values are different (such as the temperature values of A, B, C, D, and E are all different, but the temperatures are all relatively high, some reaching the critical value of tolerable high temperature, and some within the range of tolerable high temperature, etc.), the first origin fault determination condition (for example, the temperature value of A is the highest (compared with the temperature values of B, C, D, and E), and it is also the first to show externally manifested fault phenomena), the alarm result of the first type of situation one (then determine that A is the origin faulty hardware, and then it is very likely that the other faulty hardware is induced by the temperature transfer of A to show externally manifested fault phenomena, then upload this judgment result to the backend, and take targeted countermeasures for the fault data of A).

[0058] Specifically, step S2 includes the following sub-steps:

[0059] When all the temperature values in the current measured temperature values are different, if it is determined that the time point of the externally manifested fault phenomenon of one of the faulty hardware in the current faulty hardware is the earliest and the current measured temperature value it belongs to is the smallest, then output the second origin fault determination condition.

[0060] Statistically obtain the characteristic correlation data for the lengths and curvatures of the connection lines between various faulty hardware in the current faulty hardware.

[0061] According to the characteristic correlation data, evaluate the degree of fault transfer correlation between various faulty hardware in the current faulty hardware to obtain the degree of fault correlation value.

[0062] If the degree of fault transfer correlation value corresponding to one of the faulty hardware belonging to the second origin fault determination condition is the maximum value among the degrees of fault correlation values, then determine the one of the faulty hardware belonging to the second origin fault determination condition as the origin faulty hardware, and issue an alarm notification, and output the alarm result of the first type of situation two.

[0063] Specifically, for the origin fault determination condition 2 (if the temperature value of B is the lowest and it is also the first to show external fault phenomena, then further judgment is required at this time: whether it is a self-fault condition induced by its own temperature influence or an external fault hardware fault transfer induced by other factors), characteristic correlation data (such as the route connection length between hardware affects the temperature transfer efficiency. The longer the route, the lower the temperature transfer efficiency, and the degree of curvature of the route connecting between hardware also affects the temperature transfer efficiency. The longer the route curvature, the lower the temperature transfer efficiency, and both show a positive ratio relationship), fault correlation degree value (according to the statistically obtained characteristic correlation data, match the corresponding positive ratio relationship ratio from the correlation comparison table statistically obtained from the historical detection data detected in the historical period, so as to evaluate the fault correlation degree value between various fault hardwares in the current fault hardware according to the characteristic correlation data and the positive ratio relationship ratio), the alarm result 2 of the first type of situation (such as the comprehensive correlation degree value between B and A, C, D, E in the fault correlation degree value is Z1, the comprehensive correlation degree value between A and B, C, D, E is Z2, and so on, C - Z3, D - Z4, E - Z5. If Z1 is the maximum value, then determine the corresponding B as the origin fault hardware, and then it is very likely that the other fault hardwares are external fault phenomena induced by the temperature transfer of B. Then upload this judgment result to the backend and take targeted countermeasures for the fault data of B).

[0064] Specifically, step S3 includes the following sub-steps:

[0065] According to the situation to be measured, when all the temperature values in the current temperature value to be measured are the same, count the number of fault nodes with data processing speed faults among all the nodes that perform data processing in each fault hardware to which the current fault hardware belongs, and count the ratio of the weight nodes in the fault nodes to obtain the weight node ratio.

[0066] Obtain the historical fault node characteristics of the current fault hardware for detecting fault node data in the historical period, and establish a fault prediction model 1 according to the historical fault node characteristics.

[0067] Input the number of fault nodes, the weight node ratio, and the current fault hardware into the fault prediction model 1 for testing to obtain the hardware fault prediction time series.

[0068] Compare the hardware fault prediction time series with the time series of the external fault phenomena of each hardware in the situation to be measured to obtain a comparison result 1.

[0069] If there is a same comparison result of the starting-end fault time series in the comparison result 1, then determine the fault hardware to which the starting-end fault time series belongs as the origin fault hardware, and give an alarm notification, and output the alarm result 1 of the second type of situation.

[0070] Specifically, when all the temperature values in the currently measured temperature values are the same (if the temperature values of A, B, C, D, and E are all the same and are relatively high, such as reaching the critical value of tolerable high temperature or being within the range of tolerable high temperature, etc.), the nodes here refer to the steps of data processing, data storage, data communication, etc. in the hardware memory. The weighted node ratio (refers to the nodes with fault tolerance as weighted nodes, and the weighted node ratio is the ratio of the number of weighted fault nodes in the number of fault nodes to the number of fault nodes), the first fault prediction model (the historical fault node characteristics here refer to the number of fault nodes in the faulty hardware detected in the historical period and the ratio of weighted fault nodes, which are also characteristic data, as well as the corresponding historical faulty hardware and the timing characteristics of the origin and induced faulty hardware in the corresponding historical faulty hardware. According to the characteristic data detected in different historical periods, a comparison and matching table can be listed to match the number of fault nodes, weighted node ratio, and current faulty hardware detected in the current detection period with the listed comparison and matching table for feature information matching, so as to match the timing of the origin and induced faulty hardware to which the same feature information belongs, that is, obtain the hardware fault prediction timing), the first comparison result (if the timing of the external fault phenomena of each hardware in the measured situation is C - A - E - D - B, and the hardware fault prediction timing includes various situations such as C - D - B - E - A, A - B - D - C - B, etc., the two are compared), the first alarm result of the second type of situation (such as one of the prediction timings C - D - B - E - A in the hardware fault prediction timing is the same as the starting - end timing C of the timing of the external fault phenomena of each hardware in the measured situation C - A - E - D - B, then it can be judged that C is the origin faulty hardware, that is, the verification is passed. Then, it is very likely that the other faulty hardware is induced by the external fault phenomenon under the influence of the temperature transfer of C. Then, this judgment result is uploaded to the backend, and targeted countermeasures are taken for the fault data of C).

[0071] Specifically, step S4 includes the following sub - steps:

[0072] If there is only the same comparison result of the end - point fault timing in the first comparison result, the fault correlation degree value is input into the first fault prediction model for secondary testing, and the secondary prediction information of the fault timing is output.

[0073] If there is the same comparison result of the middle - point fault timing in the first comparison result, and the comparison results of the starting - end fault timing and the end - point fault timing in the first comparison result are opposite, then the position of the weighted fault nodes with data - processing speed faults in the faulty hardware with opposite comparison results is detected to obtain the weighted node position characteristics.

[0074] The faulty hardware with the opposite comparison result belonging to the weighted faulty node with the most forward position in the weighted node position feature is determined as the originating faulty hardware, and an alarm notification is made, and the second alarm result 2 of the second category is output.

[0075] Specifically, such as the secondary prediction information of the fault timing (such as a predicted timing AEDCB in the hardware fault prediction timing, which is the same as the B end timing in the timing CAEDB of the external fault phenomenon of each hardware in the test situation, it is impossible to determine which type of hardware is the source of the fault, so a secondary verification judgment is required, and the fault correlation degree value with correlation influence is compared with the historical fault correlation feature data of the same category contained in the comparable matching table, so as to extract several matching fault timings from the previously matched source and inducing fault hardware fault timings, for example, there are five types of fault timings matched previously, and there are only three types of fault timings matched twice), if the comparison result 1 has the same comparison result of the middle-end fault timing, and the comparison result 1 has the opposite comparison results of the starting-end fault timing and the end-end fault timing (such as Among them, a prediction timing AEDCB in the hardware fault prediction timing is the same as the EDC mid-end fault timing in the timing BEDCA of the external fault phenomena of each hardware in the test situation, while the starting segment timing and the end timing are opposite), weight node position characteristics (such as detecting the positions of the weighted fault nodes in the A and B hardware, if the weighted fault node in A is located close to the starting fault node, and the weighted fault node in B is located far away from the starting fault node), the second type of situation alarm result two (the closer the position of the weighted fault node is, the greater the probability of hardware failure, that is, the earlier the failure occurs, then A is judged to be the origin fault hardware, and then other faulty hardware is most likely to be induced by the temperature transfer of A, resulting in the external fault phenomenon, then this judgment result is uploaded to the back end, and targeted countermeasures are taken for the fault data of A).

[0076] The specific step S5 includes the following sub-steps:

[0077] According to the test situation, when the current temperature values ​​to be tested contain not only some identical temperature values ​​but also some different temperature values, the historical temperature fault degree characteristics of the current faulty hardware in the historical period that affect the detection of hardware fault degrees are obtained, and according to the historical temperature fault degree characteristics, fault prediction model 2 is established.

[0078] The current temperature value to be measured is input into the fault prediction model 2 for testing to obtain the impact degree of the current temperature fault.

[0079] Evaluate the stability degree value of the data processing speed of all nodes in each faulty hardware in the current faulty hardware according to the impact degree of the current temperature fault to obtain the node stability degree value.

[0080] Arbitrarily select a faulty hardware from the current faulty hardware as the starting hardware to be tested, and regard the remaining faulty hardware in the current faulty hardware except the starting hardware to be tested as the influencing hardware to be tested.

[0081] According to the node stability degree value, predict the process mode after node switching in the starting hardware to be tested and the influencing hardware to be tested to obtain the node switching characteristics to be tested.

[0082] Detect the current node switching characteristics formed among all faulty hardwares in the current faulty hardware, and compare the node switching characteristics to be tested with the current node switching characteristics to obtain the comparison result two.

[0083] Determine the starting hardware to be tested belonging to the same comparison result in the comparison result two as the origin faulty hardware, and give an alarm notification, and output the alarm result of the third type of situation.

[0084] Specifically, when there are not only some identical temperature values but also some different temperature values in the current temperature value to be measured (for example, the temperature values of A and C are the same, the temperature values of B, D, and E are the same, and the temperature values between A, C and B, D, E are different), for the second fault prediction model (such as according to the fault degree of the faulty hardware induced by temperature detected in the historical period, and listing it in a comparison reference table, which is the same as the explanation of the first fault prediction model, except that the second working prediction model focuses on the influence of temperature on the fault degree of the hardware), the current temperature fault influence degree (by comparing the current temperature value to be measured with the characteristic data in the comparison reference table to match the corresponding fault influence degree value of the current faulty hardware affected by temperature, that is, the current temperature fault influence degree), the node stability degree value (such as according to the historical correlation influence relationship between all nodes in the faulty hardware detected in the historical period and the fault influence degree value, which causes the change of the node stability degree value, and then according to the comparison of the characteristic data between the historical correlation influence relationship in the historical detection data and the current temperature fault influence degree, correspondingly evaluating the data processing speed stability degree value of all nodes in each faulty hardware in the current faulty hardware for data processing, that is, the node stability degree value), the starting hardware to be measured (if it is A or B or C or D or E), the affected hardware to be measured (if the starting hardware to be measured is A, the affected hardware to be measured is B, C, D, E, and so on), the node switching feature of the hardware to be measured (if the nodes in A include: node 1, node 2, node 3, node 4, node 5, and if the node stability degree values of their respective nodes are 90%, 85%, 75%, 90%, 90% respectively, if when the stability degree value is lower than 85%, it indicates that data processing is very likely to be abnormal, then node 3 is the switching node. If the original node processing flow is node 1 - node 2 - node 3 - node 4 and then data transmission processing is carried out on the nodes in another hardware, and finally returns to node 5 to continue processing or does not return, and the node switching flow is node 1 - node 2 - node 3 and then data transmission processing is carried out on the nodes in another hardware. If it is connected to node 2 in B, then to node 3 in E, then to node 1 in C, then to node 4 in D, and if it finally returns to node 5 in A, then the flow mode after the switching node is the node switching feature of the hardware to be measured. Here, only the case where the starting hardware to be measured is A and the affected hardware to be measured is B, C, D, E is taken as an example for analysis), the current node switching feature (if the time sequence of the external fault phenomena of each hardware in the measurement situation is B - E - D - C - A, and the current node switching feature belongs to the node switching flow mode of node 1 - node 2 - node 3 in B and then connected to node 4 in E, then to node 3 in D, then to node 2 in C, then to node 2 in A), the alarm result of the third type of situation (if there is a matching identical result in the comparison between the node switching feature of the hardware to be measured and the current node switching feature, it is determined that B is the origin faulty hardware, that is, the verification is passed,Subsequently, it is highly likely that other faulty hardware is induced by the temperature transfer of A, resulting in an obvious fault phenomenon. Then, upload this judgment result to the backend and take targeted countermeasures for the fault data of A).

[0085] A computer fault alarm system, by applying a computer fault alarm method as described above, includes the determination of the origin faulty hardware in case one, the fault correlation analysis unit, the fault node feature analysis unit, the determination of the origin faulty hardware in case two, and the determination of the origin faulty hardware in case three. Refer to Figure 2 , detect the current faulty hardware and the current temperature value to be measured of the current faulty hardware through the determination of the origin faulty hardware in case one. If all the temperature values in the current temperature value to be measured are different, then determine the faulty hardware with the earliest time point of the fault appearance and the largest current temperature value to be measured as the origin faulty hardware, and output the first type of alarm result one; through the fault correlation analysis unit, if all the temperature values in the current temperature value to be measured are different, and it is determined that there is a faulty hardware with the earliest time point of the fault appearance and the smallest current temperature value to be measured, then count the fault correlation degree values between the faulty hardware in the current faulty hardware, and determine the origin faulty hardware according to the fault correlation degree values, and output the first type of alarm result two; through the fault node feature analysis unit, if all the temperature values in the current temperature value to be measured are the same, according to the counted number of fault nodes and the weight node ratio within each faulty hardware to which the current faulty hardware belongs, predict the hardware fault prediction time sequence of the fault phenomenon of the current faulty hardware. If the comparison result of the starting-end fault time sequence is the same as the comparison result between the time sequences of multiple faulty hardware with fault phenomena in the current faulty hardware, then determine the faulty hardware to which the starting-end fault time sequence belongs as the origin faulty hardware, and output the second type of alarm result one; through the determination of the origin faulty hardware in case two, if it is determined that the comparison result of the middle-end fault time sequence is the same, and the comparison results of the starting-end fault time sequence and the ending-end fault time sequence are opposite, then determine the origin faulty hardware according to the weight node position feature of the weighted fault nodes in the faulty hardware with the opposite comparison result, and output the second type of alarm result two; through the determination of the origin faulty hardware in case three, in the case where some of the temperature values in the current temperature value to be measured are the same and some are different, evaluate the node stability degree values of all nodes in each faulty hardware in the current faulty hardware, and determine the origin faulty hardware according to the node stability degree values, and output the third type of alarm result.

[0086] The above are all the preferred embodiments of this application, and the protection scope of this application is not limited accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of this application should be covered within the protection scope of this application.

Claims

1. A computer fault alarm method, characterized in that: The following steps are involved: Detect the current faulty hardware and the current temperature value to be measured of the current faulty hardware. If all the temperature values ​​in the current temperature values ​​to be measured are different, the faulty hardware with the earliest fault manifestation time and the largest current temperature value to be measured is determined as the origin faulty hardware, and the first alarm result of the first category is output; If all the temperature values ​​in the current temperature values ​​to be measured are different, it is determined that there is a faulty hardware with the earliest time point of fault manifestation and the smallest current temperature value to be measured, then the fault correlation degree values ​​between the faulty hardware in the current faulty hardware are counted, and the origin faulty hardware is determined according to the fault correlation degree values, and the alarm result 2 of the first category is output; If all the temperature values ​​in the current temperature values ​​to be measured are the same, the hardware fault prediction timing of the fault phenomenon of the current faulty hardware is predicted based on the statistics of the number of faulty nodes and the weighted node ratio in each faulty hardware to which the current faulty hardware belongs. If the comparison result of the starting end fault timing between the timing comparisons of the fault phenomena of multiple faulty hardware with the current faulty hardware is the same, then the faulty hardware to which the starting end fault timing belongs is determined to be the origin faulty hardware, and the second type of alarm result one is output; If the comparison results of the middle-end fault sequence are the same, and the comparison results of the start-end fault sequence and the end-end fault sequence are opposite, then the source fault hardware is determined based on the statistical weight node position characteristics of the weighted fault nodes in the fault hardware with opposite comparison results, and the second alarm result 2 of the second category is output; When some of the current temperature values ​​to be measured are the same while others are different, the node stability values ​​of all nodes in each faulty hardware in the current faulty hardware are evaluated, and the source faulty hardware is determined based on the node stability values, and the alarm result of the third category of situation is output.

2. A computer fault alarm method according to claim 1, characterized in that: The steps of detecting the current faulty hardware and the current temperature value to be measured of the current faulty hardware, and if all the temperature values ​​in the current temperature values ​​to be measured are different, the faulty hardware with the earliest fault manifestation time point and the largest current temperature value to be measured is determined as the origin faulty hardware, and outputting the first alarm result of the first category is specifically as follows: Detecting the current faulty hardware that has a faulty external phenomenon inside the faulty computer, and performing temperature data detection on the current faulty hardware to obtain a current temperature value to be measured; When there are multiple faulty hardware in the current faulty hardware, and the multiple faulty hardware in the current faulty hardware are fault manifestation phenomena that occur at different time points, then the condition to be tested is output; According to the test situation, when all the temperature values ​​in the current temperature values ​​to be measured are different, if it is determined that the time point at which the fault manifestation phenomenon of one of the faulty hardware in the current faulty hardware occurs is the earliest and the current temperature value to be measured is the largest, then the origin fault determination condition 1 is output; According to the origin fault determination condition one, a type of faulty hardware belonging to the origin fault determination condition one is determined as the origin faulty hardware, and an alarm notification is made, and an alarm result one of the first category of situations is output.

3. A computer fault alarm method according to claim 2, characterized in that: If all the temperature values ​​in the current temperature values ​​to be measured are different, it is determined that there is a faulty hardware with the earliest time point of fault manifestation and the smallest current temperature value to be measured, then the fault correlation degree values ​​between the faulty hardware in the current faulty hardware are counted, and according to the fault correlation degree values, the source faulty hardware is determined, and the steps of outputting the second alarm result of the first category of situation are specifically as follows: In the case where all temperature values ​​in the current temperature values ​​to be measured are different, if it is determined that the time point at which the fault manifestation phenomenon of one of the faulty hardware in the current faulty hardware occurs is the earliest and the current temperature value to be measured is the smallest, then the second origin fault determination condition is output; The length and curvature of the connection lines between various faulty hardware in the current faulty hardware are counted to obtain characteristic correlation data; According to the characteristic association data, the fault transfer association degree values ​​between various faulty hardware in the current faulty hardware are evaluated to obtain a fault association degree value; If the fault transfer correlation degree value corresponding to a type of faulty hardware in the second origin fault judgment condition is the maximum value among the fault correlation degree values, then the type of faulty hardware in the second origin fault judgment condition is judged as the origin faulty hardware, and an alarm notification is made, and the second alarm result for the first category of situations is output.

4. A computer fault alarm method according to claim 3, characterized in that: If all the temperature values ​​in the current temperature values ​​to be measured are the same, the hardware fault prediction timing of the fault phenomenon of the current faulty hardware is predicted based on the statistics of the number of faulty nodes and the weighted node ratio in each faulty hardware to which the current faulty hardware belongs. If the comparison result of the starting end fault timing between the timing comparisons of the fault phenomena of multiple faulty hardware with the current faulty hardware is the same, then the faulty hardware to which the starting end fault timing belongs is determined to be the origin faulty hardware, and the steps of outputting the alarm result one of the second category are specifically as follows: According to the test situation, when all temperature values ​​in the current temperature values ​​to be tested are the same, the number of faulty nodes with data processing speed failures in all nodes performing data processing in each faulty hardware to which the current faulty hardware belongs is counted, and the proportion of weighted nodes in the faulty nodes is counted to obtain a weighted node ratio; Acquire historical fault node features of the current faulty hardware during fault node data detection in historical periods, and establish a first fault prediction model based on the historical fault node features; The number of faulty nodes, the weighted node ratio and the current faulty hardware are input into the first fault prediction model for testing to obtain a hardware fault prediction sequence; Comparing the hardware fault prediction timing with the timing of each hardware external fault phenomenon in the test situation to obtain a comparison result 1; If the comparison result 1 is the same as the comparison result of the starting end fault sequence, the faulty hardware to which the starting end fault sequence belongs is determined to be the originating faulty hardware, and an alarm notification is made, outputting the second category alarm result 1.

5. A computer fault alarm method according to claim 4, characterized in that: If the comparison results of the middle-end fault sequence are the same, and the comparison results of the start-end fault sequence and the end-end fault sequence are opposite, then the origin fault hardware is determined according to the statistical weight node position characteristics of the weight fault node in the fault hardware with opposite comparison results, and the steps of outputting the second alarm result of the second category are specifically as follows: If only the comparison result of the terminal fault sequence is the same in the comparison result 1, the fault correlation degree value is input into the fault prediction model 1 for secondary testing, and the secondary prediction information of the fault sequence is output; If the comparison results of the middle-end fault sequence in the comparison result 1 are the same, and the comparison results of the start-end fault sequence and the end-end fault sequence in the comparison result 1 are opposite, then the position of the weighted fault node with the data processing speed fault in the faulty hardware with opposite comparison results is detected to obtain the weighted node position feature; A faulty hardware having an opposite comparison result and belonging to a weighted faulty node with the most forward position in the weighted node position feature is determined as the originating faulty hardware, and an alarm notification is made, and the second alarm result of the second category is outputted.

6. A computer fault alarm method according to claim 5, characterized in that: In the case that some of the temperature values ​​to be measured are the same as others, the steps of evaluating the node stability values ​​of all nodes in each faulty hardware in the current faulty hardware, determining the origin of the faulty hardware according to the node stability values, and outputting the alarm result of the third category are as follows: According to the test situation, when some of the temperature values ​​to be tested are the same and some are different, the historical temperature fault degree characteristics of the current faulty hardware in the historical period that different temperature values ​​affect the hardware fault degree detection are obtained, and the fault prediction model 2 is established according to the historical temperature fault degree characteristics; The current temperature value to be measured is input into the fault prediction model 2 for testing to obtain the impact degree of the current temperature fault; According to the impact degree of the current temperature fault, the stability degree values ​​of the processing data speeds of all nodes performing data processing in each faulty hardware in the current faulty hardware are evaluated to obtain a node stability value; Any faulty hardware is selected from the current faulty hardware as the starting hardware to be tested, and the remaining faulty hardware in the current faulty hardware except the starting hardware to be tested is used as the affecting hardware to be tested; According to the node stability value, predicting the process mode after the starting hardware to be tested and the node switching in the hardware to be tested to obtain the node switching characteristics to be tested; Detecting the current node switching feature formed between all faulty hardware in the current faulty hardware, and comparing the node switching feature to be tested with the current node switching feature to obtain a second comparison result; The initial hardware to be tested to which there is an identical comparison result in the second comparison result belongs is determined as the origin fault hardware, and an alarm notification is issued, and an alarm result of the third category of situations is output.

7. A computer fault alarm system, characterized in that: The system is used to implement a computer fault alarm method according to any one of claims 1 to 6, comprising: The first case of the origin fault hardware determination is used to detect the current fault hardware and the current temperature value to be measured of the current fault hardware. If all the temperature values ​​in the current temperature values ​​to be measured are different, the fault hardware with the earliest fault manifestation time point and the largest current temperature value to be measured is determined as the origin fault hardware, and the first alarm result of the first case is output; The fault correlation analysis unit is used to determine that there is a faulty hardware with the earliest fault manifestation time point and the smallest current temperature value to be measured, if all temperature values ​​in the current temperature values ​​to be measured are different, then to count the fault correlation degree values ​​between the faulty hardware in the current faulty hardware, to determine the origin of the faulty hardware according to the fault correlation degree values, and to output the second alarm result of the first category of situations; The fault node characteristic analysis unit is used to predict the hardware fault prediction timing of the fault phenomenon of the current faulty hardware according to the number of faulty nodes and the weighted node ratio in each faulty hardware to which the current faulty hardware belongs, if all the temperature values ​​in the current temperature values ​​to be measured are the same. If the comparison result of the timing of the fault phenomenon of multiple faulty hardwares with the current faulty hardware is the same as the comparison result of the starting end fault timing, the faulty hardware to which the starting end fault timing belongs is determined to be the origin faulty hardware, and the alarm result 1 of the second category is output; The second case of the origin fault hardware determination is used to determine if the comparison results of the middle fault sequence are the same, and the comparison results of the start fault sequence and the end fault sequence are opposite, then the origin fault hardware is determined based on the statistical weight node position characteristics of the weight fault nodes in the fault hardware with opposite comparison results, and the second alarm result of the second case is output; The third situation of determining the origin of the faulty hardware is used to evaluate the node stability values ​​of all nodes in the faulty hardware in the current faulty hardware when the current temperature values ​​to be measured contain not only some identical temperature values ​​but also some different temperature values. According to the node stability values, the origin of the faulty hardware is determined, and the alarm result of the third situation is output.

Images