A hardware device damage detection method based on big data

Abstract

The present application relates to the technical field of data processing, and particularly relates to a hardware device damage detection method based on big data, which comprises the following steps: obtaining the possibility of burst situation anomaly of each target segment according to the abnormality degree of the target segment on the target switch packet forwarding rate curve and the packet forwarding rate data difference between the target segment and the normal curve segment; screening out non-burst abnormal segments from the target segment according to the possibility of burst situation anomaly of each target segment; recording the target switch with non-burst abnormal segments on the curve as non-burst abnormal switches; obtaining the performance attenuation degree of each non-burst abnormal switch in the test group according to the packet forwarding rate data of each non-burst abnormal segment on the packet forwarding rate curve of the non-burst abnormal switch in the control group and the test group and the possibility of burst situation anomaly; and screening out fault segments from the non-burst abnormal segments according to the performance attenuation degree of each non-burst abnormal switch in the test group. The present application improves the hardware device damage detection accuracy.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a hardware device damage detection method based on big data. Background Technology

[0002] With the continuous development of science and technology in the era of big data, network security issues have gradually gained attention. Data centers are crucial hubs for network data exchange and information security. Physical damage or virus intrusion can lead to data transmission interruptions, information loss, and even network paralysis. Therefore, it is necessary to strengthen network security protection and perform timely maintenance on equipment within the data center to ensure normal network operation. Taking common network equipment such as switches as an example, data transmission capacity is a key indicator of switch performance. However, in actual use, due to sudden increases in network load and performance degradation caused by long-term use, the data transmission capacity of switches may decrease (e.g., reduced data processing volume per unit time, longer processing time, etc.), which can seriously lead to network paralysis. Therefore, timely and accurate detection and maintenance of switch faults are essential for the normal operation of the network. Existing technology analyzes whether the data transmission capacity of a switch is up to standard by setting a fixed threshold, thereby detecting the quality of the switch. However, in actual work, a decrease in the data transmission capacity of a switch may be a sudden phenomenon caused by network congestion. In this case, using a single threshold is not accurate enough and may lead to inaccurate detection of hardware damage. Summary of the Invention

[0003] This invention provides a hardware device damage detection method based on big data to solve the problem that existing technologies analyze the data transmission capability of switches by setting a fixed threshold to detect the quality of switches. However, in actual work, the decline in the data transmission capability of switches may be a sudden phenomenon caused by network congestion. In this case, using a single threshold for judgment is not accurate enough and may lead to inaccurate hardware device damage detection.

[0004] The hardware device damage detection method based on big data of the present invention adopts the following technical solution:

[0005] One embodiment of the present invention provides a hardware device damage detection method based on big data, the method comprising the following steps:

[0006] Obtain the packet forwarding rate curve for each switch, where the horizontal axis of the packet forwarding rate curve represents time and the vertical axis represents the packet forwarding rate data; the switches belong to the control group and the test group respectively.

[0007] The packet forwarding rate curve of each switch is divided into several curve segments, and the approximate APCA data value of each curve segment is obtained. Based on the approximate APCA data value of each curve segment on the packet forwarding rate curve of each switch and the packet forwarding rate data, the fluctuation index of each curve segment is obtained.

[0008] Based on the duration of each curve segment, packet forwarding rate data, and fluctuation indicators on the packet forwarding rate curve of each switch, the degree of abnormality of each curve segment is obtained; based on the degree of abnormality of each curve segment, several abnormal curve segments and normal curve segments are selected; switches with abnormal curve segments on the packet forwarding rate curve are recorded as target switches.

[0009] The target segment is formed by taking adjacent abnormal curve segments on the packet forwarding rate curve of the target switch, and the average abnormality of all abnormal curve segments in the target segment is taken as the abnormality of the target segment.

[0010] Based on the degree of anomaly of the target segment on the packet forwarding rate curve of the target switch and the difference in packet forwarding rate data between the target segment and the normal curve segment, the probability of each target segment being an anomaly due to a sudden situation is obtained; based on the probability of each target segment being an anomaly due to a sudden situation, non-sudden anomaly segments are selected from the target segments; target switches with non-sudden anomaly segments on the packet forwarding rate curve are denoted as non-sudden anomaly switches.

[0011] Based on the packet forwarding rate data of each non-burst abnormal segment on the packet forwarding rate curve of the non-burst abnormal switch in the control group and the test group, as well as the probability of the abnormality being a burst situation, the performance degradation degree of each non-burst abnormal switch in the test group is obtained.

[0012] Based on the performance degradation of each non-burst-abnormal switch in the test group, faulty segments are selected from the non-burst-abnormal segments.

[0013] Preferably, the step of obtaining the fluctuation index for each curve segment based on the approximate APCA data value and packet forwarding rate data of each curve segment on the packet forwarding rate curve of each switch includes the following specific formula:

[0014]

[0015] In the formula, Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch Approximate APCA data values ​​for each curve segment; Indicates the first The mean of the approximate APCA data values ​​of all curve segments on the packet forwarding rate curve of a switch; Indicates the first The packet forwarding rate curve of the first switch The number of all data points within each curve segment; Indicates the first The packet forwarding rate curve of the first switch Within the curve segment, the th Packet forwarding rate data for each data point; It is an absolute value function.

[0016] Preferably, the specific steps for determining the degree of anomaly of each curve segment based on the duration of each curve segment, packet forwarding rate data, and fluctuation indicators on the packet forwarding rate curve of each switch are as follows:

[0017] Select the first The packet forwarding rate curve of the first switch The curve segment whose center time is closest to the center time of each of the other switch packet forwarding rate curves is denoted as the i-th curve segment. The packet forwarding rate curve of the first switch The corresponding comparison segment for each curve segment;

[0018] No. The packet forwarding rate curve of the first switch The specific formula for calculating the degree of anomaly of each curve segment is as follows:

[0019]

[0020] In the formula, Indicates the first The packet forwarding rate curve of the first switch The degree of abnormality of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The duration of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average duration of all comparison segments corresponding to each curve segment; Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average of the volatility indicators of all comparison segments corresponding to each curve segment. Indicates the first The packet forwarding rate curve of the first switch The maximum value of packet forwarding rate data in each curve segment; Indicates the first The packet forwarding rate curve of the first switch The corresponding curve segment is the th The maximum value of packet forwarding rate data in each comparison segment; Indicates the first The packet forwarding rate curve of the first switch The number of comparison segments corresponding to each curve segment; It is an absolute value function; This is the normalization function.

[0021] Preferably, the specific steps for selecting several abnormal curve segments and normal curve segments based on the degree of abnormality of each curve segment are as follows:

[0022] On each switch packet forwarding rate curve, the anomaly level exceeding a preset first threshold will be considered. The curve segment with an abnormality level less than or equal to a preset first threshold is designated as the abnormal curve segment. The curve segment that is not a curve segment is called the normal curve segment.

[0023] Preferably, the specific steps for determining the probability of each target segment being an anomaly due to a sudden occurrence based on the degree of anomaly of the target segment on the target switch's packet forwarding rate curve and the difference in packet forwarding rate data between the target segment and the normal curve segment include the following:

[0024] Based on the degree of anomaly of each target segment on the packet forwarding rate curve of the target switch and the difference in packet forwarding rate data between the target segment and the normal curve segment, the degree to which each target segment conforms to the characteristics of a sudden situation is obtained.

[0025] No. The target switch packet forwarding rate curve on the first The method for calculating the degree to which an abnormal curve segment conforms to the characteristics of a sudden event is as follows:

[0026]

[0027] In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each abnormal curve segment conforms to the characteristics of a sudden situation; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment;

[0028] In the On the packet forwarding rate curves of each target switch, abnormal curve segments with a positive degree of conformity to the characteristics of a sudden event are designated as primary abnormal curve segments. Min-max normalization is used to normalize the degree of conformity to the characteristics of a sudden event for all primary abnormal curve segments. Normalized values ​​of conformity to the characteristics of a sudden event that are greater than a preset first threshold are considered normalized segments. The main anomaly curve segment is denoted as the valid anomaly segment;

[0029] Based on the degree to which each target segment on the packet forwarding rate curve of the target switch conforms to the characteristics of a sudden situation, the effective abnormal segments, and the abnormal curve segments, the probability that each target segment on the packet forwarding rate curve of the target switch is a sudden situation abnormality can be obtained.

[0030] Preferably, the specific steps for determining the degree to which each target segment conforms to the characteristics of a sudden event based on the degree of anomaly of each target segment on the target switch's packet forwarding rate curve and the difference in packet forwarding rate data between the target segment and the normal curve segment include the following:

[0031] Select the first The target switch packet forwarding rate curve on the first The target segment whose center time is closest to the center time of the packet forwarding rate curve of each of the other target switches is denoted as the i-th segment. The target switch packet forwarding rate curve on the first The comparison segments corresponding to each target segment;

[0032] No. The target switch packet forwarding rate curve on the first The specific formula for calculating the degree to which each target segment meets the characteristics of an emergency is as follows:

[0033]

[0034] In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each target segment conforms to the characteristics of an emergency; Indicates the first The target switch packet forwarding rate curve on the first The average degree of abnormality of all comparison segments corresponding to each target segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each target segment; Indicates the first The target switch packet forwarding rate curve on the first The average forwarding rate of all packets in each target segment; Indicates the first The target switch packet forwarding rate curve is related to the first... The mean of all packet forwarding rates in the nearest normal curve segment to the target segment; Indicates the first The packet forwarding rate of the target switch is the th The number of abnormal curve segments in each target segment.

[0035] Preferably, the specific steps for determining the probability that each target segment on the packet forwarding rate curve of the target switch is an anomaly due to a sudden event, based on the degree to which each target segment on the target switch's packet forwarding rate curve conforms to the characteristics of a sudden event, the effective abnormal segments, and the abnormal curve segments, are as follows:

[0036] In the On the target switch packet forwarding rate curve, calculate the first... The number of abnormal curve segments in the target segment and the number of segments in the target segment The product of the durations of the target segments, then calculate the duration of the first segment. The number of valid outlier segments in the target segment and the number of valid outlier segments in the target segment are related to the number of The number of abnormal curve segments in the target segment and the number of segments in the target segment The ratio of the product of the durations of the target segments to the product of the durations of the target segments, and the ratio of the product of the durations of the target segments to the product of the durations of the target segments. The normalized value of the product of the degree to which each target segment conforms to the characteristics of an emergency is denoted as the i-th. Each target segment represents the possibility of unexpected or abnormal situations.

[0037] Preferably, the specific steps for obtaining the performance degradation degree of each non-burst-abnormal switch in the test group based on the packet forwarding rate data of each non-burst-abnormal segment on the packet forwarding rate curve of the non-burst-abnormal switches in the control group and test group, and the probability of the abnormality being a burst-abnormal situation, are as follows:

[0038] The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The non-burst anomaly segment whose center time is closest to the non-burst anomaly segment center time on the packet forwarding rate curve of each non-burst anomaly switch in the control group is denoted as the nth non-burst anomaly segment in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The comparison segment corresponding to each non-sudden anomaly segment;

[0039] According to the test group The packet forwarding rate data and probability of sudden anomalies for each non-sudden anomaly segment on the packet forwarding rate curve of the non-sudden anomaly switch, as well as the packet forwarding rate data of the corresponding comparison segment in the control group, and the packet forwarding rate data of the first non-sudden anomaly segment in the control group. The packet forwarding rate data for each non-burst anomaly segment on the packet forwarding rate curve of the non-burst anomaly switch and the probability of burst anomalies are used to obtain the packet forwarding rate data of the first non-burst anomaly segment in the test group. The performance degradation degree of each non-burst anomaly segment on the packet forwarding rate curve of a non-burst anomaly switch.

[0040] When the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The performance degradation of a non-sudden anomaly segment exceeds a preset first threshold. When this occurs, it is recorded as the target valid segment;

[0041] Compare the number of all non-burst anomaly segments in the packet forwarding rate curves of all non-burst anomaly switches in the control group with the number of segments in the test group. The ratio is calculated by taking the number of all target valid segments in the packet forwarding rate curve of the non-burst anomaly switch, and then multiplying this ratio by the number of segments in the test group. The average value of the performance degradation of all target valid segments in the packet forwarding rate curve of a non-burst anomaly switch is denoted as the value of the i-th segment in the test group. The degree of performance degradation of a non-sudden abnormal switch.

[0042] Preferably, the step of testing the first group... The packet forwarding rate data and probability of sudden anomalies for each non-sudden anomaly segment on the packet forwarding rate curve of the non-sudden anomaly switch, as well as the packet forwarding rate data of the corresponding comparison segment in the control group, and the packet forwarding rate data of the first non-sudden anomaly segment in the control group. The packet forwarding rate data for each non-burst anomaly segment on the packet forwarding rate curve of the non-burst anomaly switch and the probability of burst anomalies are used to obtain the packet forwarding rate data of the first non-burst anomaly segment in the test group. The performance degradation degree of each non-burst anomaly segment on the packet forwarding rate curve of a non-burst anomaly switch is calculated using the following specific formulas:

[0043]

[0044] In the formula, Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The degree of performance degradation of each non-sudden anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The average of all packet forwarding rate data in the comparison segment corresponding to each non-sudden anomaly segment; Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first The mean of all packet forwarding rates for each non-burst anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The comparison segment corresponding to each non-sudden anomaly segment represents the probability of a sudden anomaly. Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first Each non-sudden anomaly segment represents the possibility of a sudden anomaly. It is a linear normalization function.

[0045] Preferably, the specific steps for selecting faulty segments from non-burst-abnormal segments based on the performance degradation degree of each non-burst-abnormal switch in the test group are as follows:

[0046] In the test group, the first The packet forwarding rate curve of the non-burst anomaly switch is selected on the basis of the first... The most recent pre-set value within a non-sudden abnormal period of time. A non-sudden anomaly segment is denoted as the selected non-sudden anomaly segment;

[0047] The first in the test group The average performance degradation of all selected non-burst anomaly segments on the packet forwarding rate curve of the non-burst anomaly switch and the ratio of all selected non-burst anomaly segments to the first non-burst anomaly segment. The sum of the absolute values ​​of the performance degradation differences of each non-sudden anomaly segment is used as the ratio, and this ratio is then multiplied by the number of segments in the test group. The results of the performance degradation of each non-burst-type abnormal switch are normalized, and the normalized result is denoted as the i-th value in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The possibility of a non-sudden abnormal segment experiencing a failure;

[0048] When the test group is the The packet forwarding rate curve of the non-burst anomaly switch is on the first The probability of a non-sudden abnormal segment failing is greater than the preset second threshold. When this occurs, it is recorded as a fault segment.

[0049] The beneficial effects of the technical solution of this invention are as follows: By comparing the differences in data transmission volume of multiple switches in the same network during the same time period, switches with abnormal transmission capabilities are screened. The causes of the anomalies are analyzed by comparing the current abnormal period of the switch with its own historical transmission capabilities, and alarm information is set, thus improving the accuracy and effectiveness of switch damage detection and facilitating equipment maintenance by staff. Based on the duration, packet forwarding rate data, and fluctuation indicators of each curve segment on the packet forwarding rate curve of each switch, the degree of abnormality of each curve segment is obtained. Based on the degree of abnormality of each curve segment, several abnormal curve segments and normal curve segments are selected. Switches with abnormal curve segments on their packet forwarding rate curves are designated as target switches. Adjacent abnormal curve segments on the packet forwarding rate curves of the target switches constitute target segments. Based on the degree of abnormality of the target segments on the packet forwarding rate curves of the target switches and the difference in packet forwarding rate data between the target segments and normal curve segments, the probability of each target segment being a sudden anomaly is obtained. Based on the probability of each target segment being a sudden anomaly, non-sudden abnormal segments are selected from the target segments. Target switches with non-sudden abnormal segments on their packet forwarding rate curves are designated as non-sudden abnormal switches. It can accurately screen out switches with abnormal transmission capabilities, thus improving the accuracy of switch damage detection. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the steps of a hardware device damage detection method based on big data according to the present invention. Detailed Implementation

[0052] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a hardware device damage detection method based on big data proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0054] The following description, in conjunction with the accompanying drawings, details a specific scheme for a hardware device damage detection method based on big data provided by the present invention.

[0055] Please see Figure 1 The diagram illustrates a flowchart of a hardware device damage detection method based on big data, according to an embodiment of the present invention. The method includes the following steps:

[0056] Step S001: Obtain the packet forwarding rate curve for each switch, where the horizontal axis of the packet forwarding rate curve represents time and the vertical axis represents packet forwarding rate data; the switches belong to the control group and the test group respectively.

[0057] A network performance tester is set up for each switch in the network to obtain the packet forwarding rate of each switch. This example tests the packet forwarding rate (with the same frame length) of each switch over a 24-hour period, recording the results once per second for the following description.

[0058] It should be noted that this analysis of the data transmission capacity of switches is based on the packet forwarding rate (number of data packets forwarded per second) of multiple switches in the same network, thereby enabling the detection of switch quality impairments (using the switches in this network as the test group). A network performance tester is set up for each switch in the network to obtain the packet forwarding rate of each switch. (Unit: Mpps, i.e., how many millions of data packets are forwarded per second). Since the packet forwarding capacity of a switch varies with network load under actual operating conditions, this example tests the packet forwarding rate (with the same frame length) of each switch in a network over 24 hours to obtain the differences in data forwarding capacity under different loads, recording the test results once per second.

[0059] Construct a coordinate system where the horizontal axis represents time and the vertical axis represents packet forwarding rate data to obtain the packet forwarding rate curve of the switch;

[0060] It should be noted that the collected data were arranged in the order of collection time and connected point by point according to the time sequence to obtain the packet forwarding rate curve for each switch. A coordinate system was established with the direction of increasing packet forwarding rate as the positive direction of the vertical axis and the direction of increasing time sequence as the positive direction of the horizontal axis.

[0061] All switches in the network were selected as the test group. The control group and the test group were in the same environment, processed the same amount of data, and had the same number of switches. The only difference was that all switches in the control group were being used for the first time (i.e., there was almost no performance degradation). Subsequent operations were conducted using only one switch in the test group as an example (which could be a random switch).

[0062] Obtain the packet forwarding rate curve for each switch, where the horizontal axis of the packet forwarding rate curve represents time and the vertical axis represents the packet forwarding rate data; the switches belong to the control group and the test group respectively.

[0063] Step S002: Divide the packet forwarding rate curve of each switch into several curve segments and obtain the approximate APCA data value of each curve segment; based on the approximate APCA data value of each curve segment on the packet forwarding rate curve of each switch and the packet forwarding rate data, obtain the fluctuation index of each curve segment.

[0064] It should be noted that, as mentioned above, in actual operation, the packet forwarding rate of a switch will change with the load. Under different load conditions, the packet forwarding rate of the same switch will fluctuate, and therefore the obtained curve will also fluctuate. In order to obtain the trend of packet forwarding rate changes in different time periods, the APCA segmentation method is used to segment the switch packet forwarding rate curve to obtain multiple sequences with similar packet forwarding rate changes, and the fluctuation index of each segment is calculated.

[0065] The APCA segmentation method is used to divide the first... The packet forwarding rate curve of a switch is divided into several curve segments, and the approximate APCA data value of each curve segment is obtained.

[0066] The APCA segmentation method is a well-known technique, and its specific method will not be described here.

[0067] No. The packet forwarding rate curve of the first switch The calculation method for the volatility index of each curve segment is as follows:

[0068]

[0069] In the formula, Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch Approximate APCA data values ​​for each curve segment; Indicates the first The mean of the approximate APCA data values ​​of all curve segments on the packet forwarding rate curve of a switch; Indicates the first The packet forwarding rate curve of the first switch The number of all data points within each curve segment; Indicates the first The packet forwarding rate curve of the first switch Within the curve segment, the th Packet forwarding rate data for each data point; It is an absolute value function.

[0070] It should be noted that, Indicates the first The packet forwarding rate curve of the first switch The greater the difference in packet forwarding rate between each curve segment and the overall packet forwarding rate of the current curve, the greater the fluctuation. Indicates the first The packet forwarding rate curve of the first switch The sum of the differences between all data points within a curve segment and the packet forwarding rate of that segment is the largest value, indicating the greater the fluctuation of that segment.

[0071] Thus, the fluctuation index of each curve segment on the packet forwarding rate curve of each switch is obtained.

[0072] Step S003: Based on the duration of each curve segment, packet forwarding rate data, and fluctuation indicators on the packet forwarding rate curve of each switch, obtain the degree of abnormality of each curve segment; based on the degree of abnormality of each curve segment, filter out several abnormal curve segments and normal curve segments; and record the switches with abnormal curve segments on the packet forwarding rate curve as target switches.

[0073] It should be noted that network load (sudden or sustained high load causing network congestion, resulting in a decrease in the amount of data processed by the switch) and performance degradation can both lead to a decrease in the data transmission capacity of the switch (i.e., a decrease in the amount of data processed and an increase in processing time, etc.). Therefore, the switch with abnormality is identified by comparing the differences in packet forwarding rates among multiple switches in the network.

[0074] It should be noted that, taking the first curve as an example... Taking the current curve segment as an example (when comparing it with other switch curves, select the curve segment with the smallest difference in x-coordinate from the center point of the current curve segment as the comparison segment), compare the current curve with the segments of the other curves that are similar in timing, because similar timing means similar load and similar amount of data processed, which makes the comparison meaningful; traverse all curves and determine the segments of each curve that are similar in timing to the current curve segment. Each curve segment corresponds to a comparison segment. If the current curve segment takes longer and has a lower packet forwarding rate compared to the comparison segments of other switches, the greater the difference in the fluctuation coefficient of the curve segment, the more abnormal the current curve segment is. Calculate its degree of abnormality and mark the switch corresponding to the curve segment as an abnormal switch.

[0075] Select the first The packet forwarding rate curve of the first switch The curve segment whose center time is closest to the center time of each of the other switch packet forwarding rate curves is denoted as the i-th curve segment. The packet forwarding rate curve of the first switch The corresponding comparison segment for each curve segment;

[0076] No. The packet forwarding rate curve of the first switch The method for calculating the degree of anomaly of each curve segment is as follows:

[0077]

[0078] In the formula, Indicates the first The packet forwarding rate curve of the first switch The degree of abnormality of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The duration of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average duration of all comparison segments corresponding to each curve segment; Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average of the volatility indicators of all comparison segments corresponding to each curve segment. Indicates the first The packet forwarding rate curve of the first switch The maximum value of packet forwarding rate data in each curve segment; Indicates the first The packet forwarding rate curve of the first switch The corresponding curve segment is the th The maximum value of packet forwarding rate data in each comparison segment; Indicates the first The packet forwarding rate curve of the first switch The number of comparison segments corresponding to each curve segment; It is an absolute value function; This is the normalization function.

[0079] It should be noted that when the denominator in the formula is 0, let the denominator be 1. This will be used as an example to ensure that the formula is true.

[0080] It should be noted that, Indicates the first The first curve The duration of the curve segment and its relationship with the length of the first curve segment among all curves. The greater the difference between the average duration of the comparison segments corresponding to each curve and the average duration of the current curve segment, the greater the degree of abnormality of that segment. Indicates the first The first curve The curve segment and all curves in the current curve are in the same order. The larger the difference in the mean of the volatility index of the corresponding comparison segments, the greater the deviation of the current curve segment from the overall volatility, and the greater the degree of abnormality of that segment. Indicates the first The first curve The maximum packet forwarding rate in each curve segment and the curve with respect to the first curve. The sum of the maximum packet forwarding rates of the corresponding comparison segments of each curve segment indicates that the larger the value, the more abnormal the curve segment is relative to the corresponding segments of the other curves, and therefore the greater its degree of abnormality.

[0081] At this point, the degree of anomaly is obtained for each segment of the packet forwarding rate curve for each switch.

[0082] Preset first threshold A value of 0.5 indicates that the anomaly level on each switch's packet forwarding rate curve exceeds a preset first threshold. The curve segment with an abnormality level less than or equal to a preset first threshold is designated as the abnormal curve segment. The curve segment that is not a curve segment is called the normal curve segment.

[0083] Switches with abnormal curve segments on their packet forwarding rate curves are designated as target switches.

[0084] Step S004: Construct target segments from adjacent abnormal curve segments on the packet forwarding rate curve of the target switch; based on the degree of abnormality of the target segments on the packet forwarding rate curve of the target switch and the difference in packet forwarding rate data between the target segments and the normal curve segments, obtain the probability that each target segment is an abnormal burst; based on the probability that each target segment is an abnormal burst, filter out non-abrupt abnormal segments from the target segments; Target switches with non-abrupt abnormal segments on the packet forwarding rate curve are denoted as non-abrupt abnormal switches.

[0085] It should be noted that because a switch's data processing capacity is affected by load, when the network load suddenly increases, causing congestion, the amount of data processed by the switch at that moment will decrease (packet forwarding rate decreases), and the data processing time will also increase. After adjustment by the switch's congestion control mechanism, its data transmission capacity will gradually recover. If the above conditions are met, the current anomaly is considered to be caused by a sudden event (i.e., the decrease in packet forwarding rate caused by a sudden event is temporary, and the switch itself is not abnormal). However, the packet forwarding rate drop caused by performance degradation is more significant and difficult to recover to normal. Therefore, we analyze whether the current anomaly is caused by a sudden event by examining the differences in the remaining segments of the packet forwarding rate curve corresponding to the current switch.

[0086] It should be noted that, therefore, based on the comparison of the packet forwarding rate changes in the abnormal segment of the current test group switch curve with those in other time periods, the possibility that the current abnormal segment was caused by a sudden event is analyzed. From the above analysis, it can be seen that the anomaly caused by a sudden event is manifested in the curve as follows: the better the current segment's ability to respond to load changes (the smaller the drop in packet forwarding rate, the shorter the duration of the abnormal segment, and the smaller the fluctuation of the curve), the smaller the degree of anomaly, and the curve segment after the current segment gradually returns to normal. Conversely, the anomaly caused by performance degradation is manifested in the curve as follows: the packet forwarding rate of the current segment drops significantly, the degree of anomaly is greater, and it is more difficult to recover to a normal state.

[0087] Since there may be multiple consecutive abnormal curve segments in the same curve, adjacent abnormal curve segments are merged to obtain the target segment. The degree of abnormality of each target segment is the average of the degree of abnormality of all abnormal curve segments in the target segment.

[0088] Select the first The target switch packet forwarding rate curve on the first The target segment whose center time is closest to the center time of the packet forwarding rate curve of each of the other target switches is denoted as the i-th segment. The target switch packet forwarding rate curve on the first The comparison segments corresponding to each target segment;

[0089] No. The target switch packet forwarding rate curve on the first The calculation method for the degree to which a target segment meets the characteristics of an emergency is as follows:

[0090]

[0091] In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each target segment conforms to the characteristics of an emergency; Indicates the first The target switch packet forwarding rate curve on the first The average degree of abnormality of all comparison segments corresponding to each target segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each target segment; Indicates the first The target switch packet forwarding rate curve on the first The average forwarding rate of all packets in each target segment; Indicates the first The target switch packet forwarding rate curve is related to the first... The mean of all packet forwarding rates in the nearest normal curve segment to the target segment; Indicates the first The packet forwarding rate of the target switch is the th The number of abnormal curve segments in each target segment.

[0092] It should be noted that, Indicates the first On the packet forwarding rate curve of the target switch, that is The difference between the packet forwarding rate of a target segment and the previous normal curve segment is positive and the smaller the difference is, the less the current target segment deviates from the normal state. This indicates the degree of anomaly of the current target segment relative to the comparison segments among all curves. The smaller the degree of anomaly, the less the current target segment deviates from the normal state, meaning that the switch's ability to respond to load changes is better under the current state. The greater the degree to which a target segment conforms to the characteristics of an emergency, the better. Indicates the current curve's th The number of abnormal segments in a target segment; the larger this value, the more abnormal the target segment.

[0093] It should be noted that the drop in packet forwarding rate caused by the sudden situation is temporary. After the switch congestion control is adjusted, the packet forwarding rate will gradually return to normal. Therefore, analyzing the degree of change of the abnormal segment within the target segment can help determine the possibility that the abnormality of the target segment is due to a sudden situation.

[0094] No. The target switch packet forwarding rate curve on the first The method for calculating the degree to which an abnormal curve segment conforms to the characteristics of a sudden event is as follows:

[0095]

[0096] In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each abnormal curve segment conforms to the characteristics of a sudden situation; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment.

[0097] It should be noted that, Indicates the current curve's first... The abnormal curve segment and the first The greater the difference in the degree of abnormality between the segments of the abnormal curve, the stronger the abnormality. The abnormal segment is compared to the first The abnormal segments are more likely to be in a normal state; Indicates the current curve's first... The abnormal curve segment and the first The difference in packet forwarding rate among the data points of the anomaly curve segment, when the result is positive, the larger the value, the stronger the anomaly curve segment. The first abnormal segment compared to the first The better the data transmission capability of an abnormal segment, the closer it is to the normal state.

[0098] In the On the packet forwarding rate curves of each target switch, abnormal curve segments with a positive degree of conformity to the characteristics of a sudden event are designated as primary abnormal curve segments. Min-max normalization is used to normalize the degree of conformity to the characteristics of a sudden event for all primary abnormal curve segments. Normalized values ​​of conformity to the characteristics of a sudden event that are greater than a preset first threshold are considered normalized segments. The main anomaly curve segment is denoted as the effective anomaly segment.

[0099] In the On the target switch packet forwarding rate curve, calculate the first... The number of abnormal curve segments in the target segment and the number of segments in the target segment The product of the durations of the target segments, then calculate the duration of the first segment. The number of valid outlier segments in the target segment and the number of valid outlier segments in the target segment are related to the number of The number of abnormal curve segments in the target segment and the number of segments in the target segment The ratio of the product of the durations of the first target segment to the product of the durations of the second target segment, and the ratio of this ratio to the product of the durations of the third target segment. The normalized value of the product of the degree to which each target segment conforms to the characteristics of an emergency is denoted as the i-th. Each target segment represents the possibility of unexpected or abnormal situations.

[0100] No. The target switch packet forwarding rate curve on the first The method for calculating the probability of an unexpected anomaly in each target segment is as follows:

[0101]

[0102] In the formula, Indicates the first The target switch packet forwarding rate curve on the first Each target segment represents a possibility of unforeseen and abnormal situations. Indicates the first The target switch packet forwarding rate curve on the first The degree to which each target segment conforms to the characteristics of an emergency; Indicates the first The target switch packet forwarding rate curve on the first The number of valid outlier segments in each target segment; Indicates the first The packet forwarding rate of the target switch is the th The number of abnormal curve segments in each target segment; Indicates the first The target switch packet forwarding rate curve on the first The duration of each target segment; This is the normalization function.

[0103] It should be noted that, The larger the value, the more... The more valid abnormal segments there are in a target segment, the more it matches the characteristics of an abnormal sudden situation; Indicates the first The smaller the duration of each target segment, the stronger the switch's ability to cope with load changes; This indicates the switch's ability to recover to a normal state under the current condition. The larger the value, the more it matches the characteristics of a sudden abnormal situation, and the more the current target segment matches the characteristics of a sudden abnormal situation.

[0104] Thus, the probability of each target segment on the packet forwarding rate curve of each target switch being an abnormal sudden situation is obtained.

[0105] On the packet forwarding rate curve of each target switch, the probability of sudden abnormal situations is less than or equal to a preset first threshold. The target segment is denoted as the non-sudden anomaly segment;

[0106] The target switch with non-burst abnormal segments on the packet forwarding rate curve is denoted as the non-burst abnormal switch.

[0107] Step S005: Based on the packet forwarding rate data of each non-burst abnormal segment on the packet forwarding rate curve of the non-burst abnormal switch in the control group and the test group, and the probability of it being a burst abnormality, obtain the performance degradation degree of each non-burst abnormal switch in the test group.

[0108] It should be noted that under normal circumstances, switches have a good ability to respond to load changes (manifested as small fluctuations in packet forwarding rate under different loads, short processing time for abnormal situations, and quick recovery to normal). As the switch is used for a long time, the internal hardware of the switch will gradually age, which will lead to a gradual decline in the switch's data transmission capacity, a poor ability to respond to load changes, and difficulty in restoring the data transmission capacity of the switch in the corresponding state to normal. In other words, the performance of the switch will gradually degrade with the length of use.

[0109] It should be noted that the data transmission capabilities of the switches in the test group are compared with those in the corresponding control group (the control group consists of switches being used for the first time, with the least performance degradation), to analyze the degree of performance degradation of the switches in the test group. If the packet forwarding rate of the switches in the test group is lower than that of their counterparts in the control group, and their ability to respond to load changes is worse, then the performance degradation of the switches in the current state is greater.

[0110] It should be noted that, taking an abnormal segment in the curve corresponding to a non-burst abnormal switch as an example, the degree of performance degradation of the switch under the current state is obtained by analyzing the difference between the non-burst abnormal segment in the abnormal switch curve and the corresponding switch in the control group during the same period, as well as their ability to cope with load changes.

[0111] The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first Taking a non-sudden abnormal segment as an example;

[0112] The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The non-burst anomaly segment whose center time is closest to the non-burst anomaly segment center time on the packet forwarding rate curve of each non-burst anomaly switch in the control group is denoted as the nth non-burst anomaly segment in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The comparison segment corresponding to each non-sudden anomaly segment.

[0113] The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The calculation method for the performance degradation degree of each non-sudden anomaly segment is as follows:

[0114]

[0115] In the formula, Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The degree of performance degradation of each non-sudden anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The average of all packet forwarding rates in all comparison segments corresponding to a non-burst anomaly segment; Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first The mean of all packet forwarding rates for each non-burst anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The average probability of a sudden anomaly is represented by all the comparison segments corresponding to each non-sudden anomaly segment. Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first Each non-sudden anomaly segment represents the possibility of a sudden anomaly. It is a linear normalization function.

[0116] It should be noted that, The larger the value, the smaller the switch's ability to recover to normal status at the time corresponding to the current non-sudden abnormal segment, and the longer it takes to cope with load changes; This indicates the difference in packet forwarding rates between the control group and the test group within the same time period. The greater the difference, the worse the current data transmission capability. The larger the value, the greater the difference in the ability of the control group and the test group to cope with load changes within the same time period, and the greater the degree of performance degradation of the switch in the current state.

[0117] When the test group is the The packet forwarding rate curve of the non-burst anomaly switch is on the first The performance degradation of a non-sudden anomaly segment exceeds a preset first threshold. When the time is right, it is recorded as the target valid segment.

[0118] Compare the number of all non-burst anomaly segments in the packet forwarding rate curves of all non-burst anomaly switches in the control group with the number of segments in the test group. The ratio of the number of all target valid segments in the packet forwarding rate curve of the non-burst anomaly switch, multiplied by the number of segments in the test group. The result of the average performance degradation of all target effective segments in the packet forwarding rate curve of a non-burst anomaly switch is denoted as the i-th segment in the test group. The degree of performance degradation of a non-burst abnormal switch;

[0119] The first in the test group The method for calculating the performance degradation of a non-burst-type abnormal switch is as follows:

[0120]

[0121] In the formula, Indicates the first in the test group The degree of performance degradation of a non-burst abnormal switch; This represents the number of all non-burst abnormal segments in the packet forwarding rate curves of all non-burst abnormal switches in the control group. Indicates the first in the test group The number of all target valid segments in the packet forwarding rate curve of a non-burst anomaly switch; Indicates the first in the test group The average performance degradation of all target valid segments in the packet forwarding rate curve of a non-burst anomaly switch.

[0122] It should be noted that, The larger the value, the more curve segments that conform to performance degradation, and the more severe the degradation of the current curve.

[0123] At this point, the performance degradation level of each non-burst anomaly switch in the test group was obtained.

[0124] Step S006: Based on the performance degradation of each non-burst abnormal switch in the test group, select the faulty segment from the non-burst abnormal segment.

[0125] It should be noted that since performance degradation to a certain extent will cause the switch to malfunction, the greater the performance degradation of the switch in the current state, the greater the probability of the switch malfunctioning. Therefore, the analysis based on the degradation trend of the current curve segment determines whether the switch needs maintenance and sets alarm thresholds. When the performance degradation is greater in adjacent time periods and the difference in performance degradation between segments is smaller, it indicates that the switch is in a process of rapid performance degradation, and therefore the probability of malfunction is also greater. Alarm thresholds are obtained based on the above characteristics.

[0126] Default value It is 5, the number in the test group The packet forwarding rate curve of the non-burst anomaly switch is selected on the basis of the first... The most recent pre-set value within a non-sudden abnormal period of time. A non-sudden anomaly segment is denoted as the selected non-sudden anomaly segment;

[0127] The first in the test group The average performance degradation of all selected non-burst anomaly segments on the packet forwarding rate curve of a non-burst anomaly switch is compared with that of all selected non-burst anomaly segments and the first non-burst anomaly segment. The sum of the absolute values ​​of the performance degradation differences of each non-sudden anomaly segment is used as the ratio, which is then multiplied by the number of segments in the test group. The results of the performance degradation of each non-burst-type abnormal switch are normalized, and the normalized result is denoted as the i-th value in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The possibility of a non-sudden abnormal segment experiencing a failure;

[0128] The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The method for calculating the probability of a non-sudden anomaly segment failing is as follows:

[0129]

[0130] In the formula, Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The possibility of a non-sudden abnormal segment experiencing a failure; Indicates the first in the test group The degree of performance degradation of a non-burst abnormal switch; Indicates the first in the test group The average performance degradation of all selected non-burst abnormal segments on the packet forwarding rate curve of a non-burst abnormal switch. Indicates the first in the test group All selected non-burst anomaly segments on the packet forwarding rate curve of the non-burst anomaly switch are related to the first... The sum of the absolute values ​​of the performance degradation differences of each non-sudden anomaly segment; This is the normalization function.

[0131] It should be noted that, This indicates the degree of performance degradation of the current switch; the higher the value, the greater the likelihood that the current switch will fail. The smaller the value, the more consistent the performance degradation of the selected segments, and the greater the possibility of a failure in that segment. The larger the value, the more severe the performance degradation of the switch in the current state.

[0132] Preset second threshold It is 0.8, the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The probability of a non-sudden abnormal segment failing is greater than the preset second threshold. When this occurs, it is recorded as a fault segment;

[0133] When a faulty segment is identified, the maximum value of the packet forwarding rate data in that segment is set as an alarm threshold to facilitate the detection of abnormal situations.

[0134] It should be noted that the current status is fed back to the detection device according to the set alarm information. If the detected abnormal data transmission capability is a sudden abnormality, no alarm is required. If the detected abnormal data transmission capability is caused by performance degradation and the degree of performance degradation is large, then the warning information is output to the staff according to the alarm threshold obtained in the current time period and the performance degradation degree of the abnormal switch is displayed, so that the staff can perform equipment maintenance in a timely manner and reduce losses.

[0135] This concludes the embodiment.

[0136] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A hardware device damage detection method based on big data, characterized in that, The method includes the following steps: Obtain the packet forwarding rate curve for each switch, where the horizontal axis of the packet forwarding rate curve represents time and the vertical axis represents the packet forwarding rate data; the switches belong to the control group and the test group respectively. The packet forwarding rate curve of each switch is divided into several curve segments, and the approximate APCA data value of each curve segment is obtained. Based on the approximate APCA data value of each curve segment on the packet forwarding rate curve of each switch and the packet forwarding rate data, the fluctuation index of each curve segment is obtained. Based on the duration of each curve segment, packet forwarding rate data, and fluctuation indicators on the packet forwarding rate curve of each switch, the degree of abnormality of each curve segment is obtained; based on the degree of abnormality of each curve segment, several abnormal curve segments and normal curve segments are selected; switches with abnormal curve segments on the packet forwarding rate curve are recorded as target switches. The target segment is formed by taking adjacent abnormal curve segments on the packet forwarding rate curve of the target switch, and the average abnormality of all abnormal curve segments in the target segment is taken as the abnormality of the target segment. Based on the degree of anomaly of the target segment on the packet forwarding rate curve of the target switch and the difference in packet forwarding rate data between the target segment and the normal curve segment, the probability of each target segment being an anomaly due to a sudden situation is obtained; based on the probability of each target segment being an anomaly due to a sudden situation, non-sudden anomaly segments are selected from the target segments; target switches with non-sudden anomaly segments on the packet forwarding rate curve are denoted as non-sudden anomaly switches. Based on the packet forwarding rate data of each non-burst abnormal segment on the packet forwarding rate curve of the non-burst abnormal switch in the control group and the test group, as well as the probability of the abnormality being a burst situation, the performance degradation degree of each non-burst abnormal switch in the test group is obtained. Based on the performance degradation of each non-burst-abnormal switch in the test group, faulty segments are selected from the non-burst-abnormal segments.

2. The hardware device damage detection method based on big data according to claim 1, characterized in that, The fluctuation index for each curve segment is obtained based on the approximate APCA data value and packet forwarding rate data of each curve segment on the packet forwarding rate curve of each switch. The specific formulas are as follows: In the formula, Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch Approximate APCA data values ​​for each curve segment; Indicates the first The mean of the approximate APCA data values ​​of all curve segments on the packet forwarding rate curve of a switch; Indicates the first The packet forwarding rate curve of the first switch The number of all data points within a curve segment; Indicates the first The packet forwarding rate curve of the first switch Within the curve segment, the th Packet forwarding rate data for each data point; It is an absolute value function.

3. The hardware device damage detection method based on big data according to claim 1, characterized in that, The specific steps for determining the degree of anomaly for each curve segment based on the duration of each curve segment, packet forwarding rate data, and fluctuation indicators on the packet forwarding rate curve of each switch are as follows: Select the first The packet forwarding rate curve of the first switch The curve segment whose center time is closest to the center time of each of the other switch packet forwarding rate curves is denoted as the i-th curve segment. The packet forwarding rate curve of the first switch The corresponding comparison segment for each curve segment; No. The packet forwarding rate curve of the first switch The specific formula for calculating the degree of anomaly of each curve segment is as follows: In the formula, Indicates the first The packet forwarding rate curve of the first switch The degree of abnormality of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The duration of each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average duration of all comparison segments corresponding to each curve segment; Indicates the first The packet forwarding rate curve of the first switch Fluctuation indicators for each curve segment; Indicates the first The packet forwarding rate curve of the first switch The average of the volatility indicators of all comparison segments corresponding to each curve segment. Indicates the first The packet forwarding rate curve of the first switch The maximum value of packet forwarding rate data in each curve segment; Indicates the first The packet forwarding rate curve of the first switch The corresponding curve segment is the th The maximum value of packet forwarding rate data in each comparison segment; Indicates the first The packet forwarding rate curve of the first switch The number of comparison segments corresponding to each curve segment; It is an absolute value function; This is the normalization function.

4. The hardware device damage detection method based on big data according to claim 1, characterized in that, The specific steps involved in filtering out several abnormal curve segments and normal curve segments based on the degree of abnormality of each curve segment are as follows: On each switch packet forwarding rate curve, the anomaly level exceeding a preset first threshold will be considered. The curve segment with an abnormality level less than or equal to a preset first threshold is designated as the abnormal curve segment. The curve segment that is not a curve segment is called the normal curve segment.

5. The hardware device damage detection method based on big data according to claim 1, characterized in that, The specific steps for determining the probability of each target segment being an anomaly due to a sudden occurrence, based on the degree of anomaly of the target segment on the target switch's packet forwarding rate curve and the difference in packet forwarding rate data between the target segment and the normal curve segment, are as follows: Based on the degree of anomaly of each target segment on the packet forwarding rate curve of the target switch and the difference in packet forwarding rate data between the target segment and the normal curve segment, the degree to which each target segment conforms to the characteristics of a sudden situation is obtained. No. The target switch packet forwarding rate curve on the first The method for calculating the degree to which an abnormal curve segment conforms to the characteristics of a sudden event is as follows: In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each abnormal curve segment conforms to the characteristics of a sudden situation; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment; Indicates the first The target switch packet forwarding rate curve on the first The mean of all packet forwarding rate data in each abnormal curve segment; In the On the packet forwarding rate curves of each target switch, abnormal curve segments with a positive degree of conformity to the characteristics of a sudden event are designated as primary abnormal curve segments. Min-max normalization is used to normalize the degree of conformity to the characteristics of a sudden event for all primary abnormal curve segments. Normalized values ​​of conformity to the characteristics of a sudden event that are greater than a preset first threshold are considered normalized segments. The main anomaly curve segment is denoted as the valid anomaly segment; Based on the degree to which each target segment on the packet forwarding rate curve of the target switch conforms to the characteristics of a sudden situation, the effective abnormal segments, and the abnormal curve segments, the probability that each target segment on the packet forwarding rate curve of the target switch is a sudden situation abnormality can be obtained.

6. The hardware device damage detection method based on big data according to claim 5, characterized in that, The specific steps for determining the degree to which each target segment conforms to the characteristics of a sudden event, based on the degree of anomaly of each target segment on the target switch's packet forwarding rate curve and the difference in packet forwarding rate data between the target segment and the normal curve segment, are as follows: Select the first The target switch packet forwarding rate curve on the first The target segment whose center time is closest to the center time of the packet forwarding rate curve of each of the other target switches is denoted as the i-th segment. The target switch packet forwarding rate curve on the first The comparison segments corresponding to each target segment; No. The target switch packet forwarding rate curve on the first The specific formula for calculating the degree to which each target segment meets the characteristics of an emergency is as follows: In the formula, Indicates the first The target switch packet forwarding rate curve on the first The degree to which each target segment conforms to the characteristics of an emergency; Indicates the first The target switch packet forwarding rate curve on the first The average degree of abnormality of all comparison segments corresponding to each target segment; Indicates the first The target switch packet forwarding rate curve on the first The degree of abnormality of each target segment; Indicates the first The target switch packet forwarding rate curve on the first The average forwarding rate of all packets in each target segment; Indicates the first The target switch packet forwarding rate curve is related to the first... The mean of all packet forwarding rates in the nearest normal curve segment to the target segment; Indicates the first The packet forwarding rate of the target switch is the th The number of abnormal curve segments in each target segment.

7. The hardware device damage detection method based on big data according to claim 5, characterized in that, The specific steps involved in determining the probability that each target segment on the packet forwarding rate curve of the target switch is an anomaly due to a sudden event, based on the degree to which each target segment conforms to the characteristics of a sudden event, the effective abnormal segments, and the abnormal curve segments, are as follows: In the On the target switch packet forwarding rate curve, calculate the first... The number of abnormal curve segments in the target segment and the number of segments in the target segment The product of the durations of the target segments is then used to calculate the duration of the first segment. The number of valid outlier segments in the target segment and the number of valid outlier segments in the target segment are related to the number of The number of abnormal curve segments in the target segment and the number of segments in the target segment The ratio of the product of the durations of the target segments to the product of the durations of the target segments, and the ratio of the product of the durations of the target segments to the product of the durations of the target segments. The normalized value of the product of the degree to which each target segment conforms to the characteristics of an emergency is denoted as the i-th. Each target segment represents the possibility of unexpected or abnormal situations.

8. The hardware device damage detection method based on big data according to claim 1, characterized in that, The process of determining the performance degradation degree of each non-burst-abnormal switch in the test group based on the packet forwarding rate data of each non-burst-abnormal segment on the packet forwarding rate curves of the control group and the test group, and the probability of the abnormality being a burst situation, includes the following specific steps: The first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The non-burst anomaly segment whose center time is closest to the non-burst anomaly segment center time on the packet forwarding rate curve of each non-burst anomaly switch in the control group is denoted as the nth non-burst anomaly segment in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The comparison segment corresponding to each non-sudden anomaly segment; According to the test group The packet forwarding rate data and probability of sudden anomalies for each non-sudden anomaly segment on the packet forwarding rate curve of the non-sudden anomaly switch, as well as the packet forwarding rate data of the corresponding comparison segment in the control group, and the packet forwarding rate data of the first non-sudden anomaly segment in the control group. The packet forwarding rate data for each non-burst anomaly segment on the packet forwarding rate curve of the non-burst anomaly switch and the probability of burst anomalies are used to obtain the packet forwarding rate data of the first non-burst anomaly segment in the test group. The performance degradation degree of each non-burst anomaly segment on the packet forwarding rate curve of a non-burst anomaly switch. When the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The performance degradation of a non-sudden anomaly segment exceeds a preset first threshold. When this occurs, it is recorded as the target valid segment; Compare the number of all non-burst anomaly segments in the packet forwarding rate curves of all non-burst anomaly switches in the control group with the number of segments in the test group. The ratio is calculated by taking the number of all target valid segments in the packet forwarding rate curve of the non-burst anomaly switch, and then multiplying this ratio by the number of segments in the test group. The average value of the performance degradation of all target valid segments in the packet forwarding rate curve of a non-burst anomaly switch is denoted as the value of the i-th segment in the test group. The degree of performance degradation of a non-sudden abnormal switch.

9. The hardware device damage detection method based on big data according to claim 8, characterized in that, According to the test group, the first The packet forwarding rate data and probability of sudden anomalies for each non-sudden anomaly segment on the packet forwarding rate curve of the non-sudden anomaly switch, as well as the packet forwarding rate data of the corresponding comparison segment in the control group, and the packet forwarding rate data of the first non-sudden anomaly segment in the control group. The packet forwarding rate data for each non-burst anomaly segment on the packet forwarding rate curve of the non-burst anomaly switch and the probability of burst anomalies are used to obtain the packet forwarding rate data of the first non-burst anomaly segment in the test group. The performance degradation degree of each non-burst anomaly segment on the packet forwarding rate curve of a non-burst anomaly switch is calculated using the following specific formulas: In the formula, Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The degree of performance degradation of each non-sudden anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The average of all packet forwarding rate data in the comparison segment corresponding to each non-sudden anomaly segment; Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first The mean of all packet forwarding rates for each non-burst anomaly segment; Indicates the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The comparison segment corresponding to each non-sudden anomaly segment represents the probability of a sudden anomaly. Indicating the number of cases in the control group The packet forwarding rate curve of the non-burst anomaly switch is on the first Each non-sudden anomaly segment represents the possibility of a sudden anomaly. It is a linear normalization function.

10. The hardware device damage detection method based on big data according to claim 1, characterized in that, The specific steps for filtering out faulty segments from non-burst-abnormal segments based on the performance degradation degree of each non-burst-abnormal switch in the test group are as follows: In the test group, the first The packet forwarding rate curve of the non-burst anomaly switch is selected on the basis of the first... The most recent pre-set value within a non-sudden abnormal period of time. A non-sudden anomaly segment is denoted as the selected non-sudden anomaly segment; The first in the test group The average performance degradation of all selected non-burst anomaly segments on the packet forwarding rate curve of the non-burst anomaly switch and the ratio of all selected non-burst anomaly segments to the first non-burst anomaly segment. The sum of the absolute values ​​of the performance degradation differences of each non-sudden anomaly segment is used as the ratio, and this ratio is then multiplied by the number of segments in the test group. The results of the performance degradation of each non-burst-type abnormal switch are normalized, and the normalized result is denoted as the i-th value in the test group. The packet forwarding rate curve of the non-burst anomaly switch is on the first The possibility of a non-sudden abnormal segment experiencing a failure; When the first in the test group The packet forwarding rate curve of the non-burst anomaly switch is on the first The probability of a non-sudden abnormal segment failing is greater than the preset second threshold. When this occurs, it is recorded as a fault segment.

Images