Operation and maintenance fault identification method and device, equipment and readable storage medium

Abstract

The application belongs to the technical field of digital medical treatment, and provides a kind of operation and maintenance fault identification method and device, computer equipment and computer readable storage medium, in order to solve the problem of low accuracy of fault identification when digital medical system appears operation and maintenance fault in traditional technology, if system appears fault, the time sequence characteristic data corresponding to the preset index of system and the local subsequence feature contained therein are acquired, and the corresponding index fault classification model is acquired, the feature vector corresponding to the local subsequence feature is input into the index fault classification model, and the fault is identified, with the help of local subsequence feature and corresponding index fault classification model, the fault can be identified, including but not limited to large medical group and other digital medical IT system, improve the intelligence, accuracy and efficiency of fault identification, avoid significant loss and harm caused by the failure of digital medical IT system.

Description

Technical Field

[0001] This application relates to the field of digital healthcare technology, and in particular to a method, apparatus, computer equipment, and computer-readable storage medium for identifying operational and maintenance faults. Background Technology

[0002] In digital healthcare business systems, including but not limited to large medical groups, the development of digital healthcare technology often leads to the adoption of large and complex AI-based digital healthcare IT systems for the automation and intelligent processing of medical operations. The operation and maintenance of these digital healthcare IT systems is crucial; any malfunctions must be promptly addressed to prevent significant medical losses or harm. Therefore, in the aforementioned application scenarios, leveraging AI for the automated and intelligent operation and maintenance of digital healthcare IT systems is essential to improve their efficiency.

[0003] In root cause localization within intelligent operations and maintenance (O&M), especially for complex digital healthcare IT systems (including but not limited to large medical groups), rapid identification of system faults is crucial, particularly recurring faults. These faults frequently reappear in different locations, disrupting the normal operation of the digital healthcare IT system and constituting a significant portion of daily O&M troubleshooting. O&M engineers must dedicate substantial time and effort to handling these issues. Therefore, when a fault occurs, it is essential to quickly and accurately diagnose the fault in the digital healthcare IT system to guide O&M engineers in troubleshooting.

[0004] Because repetitive faults occur repeatedly and the variations in different faults follow certain patterns, classifiers can be used to identify them. However, fault identification in traditional digital healthcare IT systems requires extensive manual feature engineering. Due to the subjectivity of manual feature engineering, the identification effect is poor and the accuracy is low. Summary of the Invention

[0005] This application provides a method, apparatus, computer equipment, and computer-readable storage medium for identifying maintenance faults, which can solve the technical problem of low accuracy in identifying maintenance faults in digital medical IT systems in traditional technologies.

[0006] In a first aspect, this application provides a method for identifying operational and maintenance faults, comprising: if a fault is detected in the system, obtaining time series feature data corresponding to a preset indicator of the system; filtering local subsequence features from the time series feature data; obtaining the indicator type corresponding to the local subsequence features, and obtaining an indicator fault classification model corresponding to the indicator type; converting the local subsequence features into feature vectors, and inputting the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault.

[0007] Secondly, this application provides a device for identifying operational and maintenance faults, comprising: a first acquisition unit, configured to acquire time series feature data corresponding to preset indicators of the system if a fault is detected in the system; a second acquisition unit, configured to filter local subsequence features from the time series feature data; a third acquisition unit, configured to acquire the indicator type corresponding to the local subsequence features and acquire an indicator fault classification model corresponding to the indicator type; and an identification unit, configured to convert the local subsequence features into feature vectors and input the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault.

[0008] Thirdly, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for identifying maintenance faults.

[0009] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the operation and maintenance fault identification method.

[0010] This application provides a method, apparatus, computer equipment, and computer-readable storage medium for identifying operational and maintenance faults. The processing method involves, if a system fault occurs, acquiring time-series feature data corresponding to preset indicators of the system, filtering local sub-series features from the time-series feature data, acquiring the indicator type corresponding to the local sub-series features, acquiring the indicator fault classification model corresponding to the indicator type, converting the local sub-series features into feature vectors, and inputting the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault. With the help of local sub-series features and the corresponding indicator fault classification model, faults including but not limited to those occurring in digital medical IT systems can be effectively identified, improving the intelligence, accuracy, and efficiency of fault identification, thereby improving the operational and maintenance efficiency. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating the method for identifying operational and maintenance faults provided in this application embodiment;

[0013] Figure 2 This is a schematic diagram of the first sub-process of the method for identifying operation and maintenance faults provided in the embodiments of this application;

[0014] Figure 3 A schematic diagram of the second sub-process of the method for identifying operation and maintenance faults provided in the embodiments of this application;

[0015] Figure 4 A schematic block diagram of an operation and maintenance fault identification device provided in an embodiment of this application;

[0016] Figure 5 A schematic block diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0019] This application provides a method for identifying operational and maintenance faults. The method can be applied to computer devices such as smartphones, tablets, laptops, and desktop computers, and is used to identify and process operational and maintenance faults, including but not limited to digital medical IT systems.

[0020] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0021] Please see Figure 1 , Figure 1 This is a flowchart illustrating the method for identifying operational and maintenance faults provided in an embodiment of this application. Figure 1 As shown, the method includes, but is not limited to, the following steps S11-S14:

[0022] S11. If a system fault is detected, the time series feature data corresponding to the preset indicators of the system are obtained.

[0023] Specifically, in digital healthcare IT systems, including but not limited to, indicators are pre-set based on the failure scenarios of system components, especially those failures that are prone to recurrence. For example, for system components such as CPU, disk, and hard drive that may fail, indicators related to CPU failure, such as CPU utilization and temperature, are pre-set as pre-set indicators. Or, indicators related to failures such as read / write head damage, circuit board problems, chip information loss, motor failure, and chip burnout are pre-set as pre-set indicators for hard drives.

[0024] If a system failure is detected, time-series characteristic data corresponding to several preset indicators of the system are obtained according to pre-set parameters. For example, for the CPU, the CPU utilization rate or temperature in the period before the system failure occurs are obtained. For instance, in a large medical group's digital healthcare IT system, if a failure is detected, time-series characteristic data corresponding to several preset indicators of potential failures in the digital healthcare IT system, including but not limited to the CPU, disk, and hard drive of medical computer equipment, are obtained according to pre-set parameters. For example, for the CPU of medical computer equipment, the CPU utilization rate or temperature in the period before the digital healthcare IT system failure occurs are obtained.

[0025] S12. Local subsequence features are obtained by filtering from the time series feature data.

[0026] Specifically, in digital healthcare IT systems, including but not limited to, local subsequence features are selected from time series feature data, particularly discriminative local subsequence features. These discriminative local subsequence features describe typical characteristics that represent local features of the time series feature data. For example, in one instance, based on Shapelet, the most discriminative local subsequence features are selected from the time series feature data; these can be called Shapelet features. Shapelet can be broken down into Shape and let; Shape means shape, and let indicates a suffix for "small." Therefore, Shapelet represents a "small shape" in the time series, i.e., a subsequence. This subsequence is a specific subsequence within the time series data that expresses the most prominent features of the time series data. In other examples, for the time series feature data corresponding to CPU utilization, the CPU utilization can be compared with the average CPU utilization over a period of time to obtain the local subsequence features corresponding to the local utilization rates that are greater than or equal to the average utilization rate in the CPU time series feature data. Alternatively, the CPU utilization can be sorted by size based on the CPU time series feature data, and several utilization rates with larger utilization rates can be obtained as local subsequence features, thereby filtering out the local subsequence features corresponding to the CPU utilization from the CPU time series feature data.

[0027] S13. Obtain the index type corresponding to the local subsequence features, and obtain the index fault classification model corresponding to the index type.

[0028] Specifically, in digital medical IT systems, including but not limited to, for different preset indicators of systems such as CPUs, disks, and hard drives in medical computer equipment, corresponding preset indicator fault classification models are set. These preset indicator fault classification models are used to identify one or more types of preset indicator faults. For example, for the CPU of a medical computer equipment, a corresponding preset indicator fault classification model can be set. This model can be used solely to identify CPU faults, handling CPU indicators such as CPU memory usage or CPU temperature. In other embodiments, the PCU and hard drive can also be set together with corresponding preset indicator fault classification models. In this case, the preset indicator fault classification model is used to identify both CPU and disk faults, handling CPU indicators such as CPU memory usage or CPU temperature, and handling disk indicators such as disk usage or temperature.

[0029] Based on the above settings, the indicator type corresponding to the local subsequence feature is obtained. The indicator type describes the type of the preset indicator corresponding to the local subsequence feature. For example, the indicator type corresponding to the local subsequence feature is a preset indicator such as CPU indicator, disk indicator or hard disk indicator. Based on the indicator type, the corresponding preset indicator fault classification model is obtained.

[0030] S14. Convert the local subsequence features into feature vectors, and input the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault.

[0031] Specifically, in digital healthcare IT systems, including but not limited to, converting local subsequence features into feature vectors can be based on shapelet feature transformation. For example, the shapelet feature set is S = <s1,s2,…s m >wherein, s j Let S represent the j-th Shapelet, j = 1, 2, 3, ..., m. Using set S, each T in the time series dataset of fault indicators can be represented... i Transform into feature vectors in, T represents i With s j The distance between them, and the transformed feature vector of the fault indicator time series dataset T, can be directly input into the pre-defined indicator fault classification model based on supervised learning to identify faults and obtain the corresponding fault types. Thus, based on the trend of local sub-sequence features of the pre-defined indicators, the capture of local features is more accurate than the capture of global features. It can avoid the interference of non-discriminatory feature data in time series feature data, determine the corresponding fault types, improve the accuracy of fault identification, and quickly take repair measures for system faults.

[0032] In this embodiment, by acquiring local subsequence features and their corresponding indicator types, and obtaining a corresponding preset indicator fault classification model based on the indicator type, the local subsequence features are converted into feature vectors. The feature vectors are then input into the preset indicator fault classification model corresponding to the indicator type to identify faults and obtain the fault class corresponding to the fault. Thus, by leveraging local subsequence features and the corresponding preset indicator fault classification model, the accuracy of fault identification, including but not limited to those occurring in digital medical IT systems, can be improved. Furthermore, the preset indicator fault classification model is only used to identify faults of one or a few preset indicator types, which reduces the computational load of the preset indicator fault classification model and improves the efficiency of fault identification. Therefore, it can effectively identify faults, including but not limited to those in digital medical IT systems, improving the intelligence, accuracy, and efficiency of fault identification, thereby improving the operational efficiency of fault maintenance.

[0033] In one embodiment, please refer to Figure 2 , Figure 2 This is a schematic diagram of the first sub-process of the method for identifying operational and maintenance faults provided in an embodiment of this application. For example... Figure 2 As shown, in this embodiment, the local subsequence features are features based on the Shapelet feature algorithm, and the process of filtering local subsequence features from the time series feature data includes:

[0034] S21. The time series feature data corresponding to several preset indicators of the system are combined into a fault indicator time series dataset.

[0035] S22. Based on the fault index time series dataset, obtain the initial sub-feature sequence and form the initial sub-sequence feature set from the initial sub-feature sequence;

[0036] S23. Based on the initial subsequence feature set, classify the initial subsequence features to obtain a single-class initial subsequence feature set, and calculate the similarity between the initial subsequence features contained in the single-class initial subsequence feature set and the corresponding time series feature data.

[0037] S24. Based on the similarity, calculate the statistics for each of the initial subsequence features, wherein the statistics describe the classification ability of the initial subsequence features;

[0038] S25. Based on the similarity, for initial subsequence features belonging to the same time series feature data, filter out initial subsequence features with high similarity to obtain several target subsequence features;

[0039] S26. Based on the statistical measure, sort all the target subsequence features and obtain the target subsequence features with the strongest classification ability corresponding to the statistical measure as local subsequence features.

[0040] Further, the step of obtaining the initial sub-feature sequence based on the fault index time series dataset includes:

[0041] The fault index time series dataset is traversed by sliding windows of different lengths for each time series feature data.

[0042] Extract the feature data corresponding to different sliding windows, and use the feature data with the same window length as the initial sub-feature sequence.

[0043] Specifically, in digital healthcare IT systems, including but not limited to, fault indicator time-series datasets are compiled from the time-series feature data corresponding to several preset indicators of the system's components, including but not limited to the CPU, disk, hard disk, circuit board, chip, and motor of computer equipment. For example, in one example, CPU utilization is used as one of the preset indicators, and the time-series feature data of CPU utilization values ​​at a series of time points is used to compile the corresponding fault indicator time-series dataset. In another example, CPU temperature is used as one of the preset indicators, and the time-series feature data of CPU temperature values ​​at a series of time points is used to compile the corresponding fault indicator time-series dataset. The fault indicator time-series dataset can be described using the following T: T = {T1, ..., T} i ,…,T n}, where the fault indicator time series dataset can be described by T, and the i-th time series feature data contained in the fault indicator time series dataset is described by T. i The description describes how to obtain initial sub-feature sequences based on the time-series dataset T of fault indicators, and then assemble these initial sub-feature sequences into an initial local sub-sequence feature set. For example, a sliding window of different lengths is used to traverse each time-series feature data in the fault indicator time-series dataset. During the traversal, feature data corresponding to the sliding window is extracted from each time-series feature data, and feature data with the same window length are used as initial sub-feature sequences. Thus, the initial sub-feature sequences are obtained based on the sliding of the sliding window. For each time-series feature data T... iDifferent initial sub-feature sequences are obtained. The initial sub-feature sequences can adopt Shapelet features. Based on the fault index time series dataset T, all initial sub-feature sequences with lengths between min and max can be found. Then, all initial sub-feature sequences are combined into an initial sub-sequence feature set, which is the candidate Shapelet set. The sliding window is a two-pointer-based approach. A window is formed between the elements pointed to by the two pointers. The size of the sliding window can be described by the window length, which can be described by L. The sliding window can be described as a window of length L. The minimum value of length L can be described by min, and the maximum value of length L can be described by max. min and max are determined by the maintenance personnel of digital medical IT systems according to different maintenance needs, or min and max can be determined by the maintenance personnel based on experience.

[0044] Based on the initial subsequence feature set, the initial subsequence features are classified to obtain a single-class initial subsequence feature set. The similarity between each initial subsequence feature in the single-class initial subsequence feature set and its corresponding time series feature data is calculated. For example, if the initial subsequence features are classified according to the length L of the sliding window, initial subsequence features of the same length L are grouped into one class, resulting in a single-class initial subsequence feature set. The minimum value of L can be described using `min`, and the maximum value of L can be described using `max`. The set of L can be described using `{min, max}`, for example, L can be different time lengths such as 10 minutes, 30 minutes, or 1 hour, which can be determined according to different operational needs, including but not limited to digital healthcare IT systems. Furthermore, the similarity between the initial subsequence feature Shapelet with a sliding window length of L and each time series feature data T of length L in the fault indicator time series dataset T can be calculated. i The similarity between them is determined by comparing the initial subsequence feature shapes of the same time series length with the time series feature data T in the fault index time series dataset T. i Similarity can be described by distance. If distance is used to describe similarity, the distance calculation formula can be calculated using the following formula (1):

[0045]

[0046] Among them, W i,l For T i R is the set of all time series features of length l in W. i,l The time series features of length l in the data are defined by the dist function, which is a formula for calculating the distance between two time series of equal length. A distance vector, Shapelet = , can be obtained through similarity measurement. <d s,1 ,d s,2 ,…,ds,n >,d s,i Used to represent the Shapelet and the corresponding T i The distance between them can be calculated using Euclidean distance, cosine distance, information entropy, Manhattan distance, etc. Euclidean distance, cosine distance, information entropy, Manhattan distance, etc. are all existing calculation methods, which will not be elaborated here.

[0047] Based on similarity, the classification ability of each candidate Shapelet can be evaluated. A statistic F can be introduced to describe the classification ability of the Shapelet. The statistic F corresponding to the classification ability of each initial subsequence feature can be calculated using the following formula (2):

[0048]

[0049] Among them, D i This represents the set of all time series and Shapelet distances under a certain label category; D represents i The average value; C represents the number of categories; n represents the size of the time series dataset.

[0050] Based on similarity, for initial subsequence features belonging to the same time series feature data, features with high similarity are filtered out to obtain several target subsequence features. According to the statistic F, all target subsequence features are ranked, and the target subsequence features with the strongest classification ability corresponding to statistic F are selected as discriminative subsequence features. For example, for shapelet filtering, for features from the same time series feature data... i The set of Shapelets is processed by removing highly similar Shapelets based on a distance formula, sorting them according to the statistic F, and finally returning the k Shapelets with the strongest classification ability in the time series dataset T.

[0051] In this embodiment, an initial sub-feature sequence is obtained based on a fault indicator time-series dataset. Specifically, the initial sub-feature sequence is obtained by traversing each time-series feature data contained in the fault indicator time-series dataset through a sliding window. The obtained initial sub-feature sequence is then filtered to obtain discriminative local sub-sequence features. Furthermore, the capture of local features of time-series feature data is more accurate than the capture of global features, thereby determining the fault type corresponding to faults, including but not limited to those occurring in digital medical IT systems. This approach effectively identifies faults, including but not limited to those occurring in digital medical IT systems, improving the intelligence, accuracy, and efficiency of fault identification, and ultimately enhancing the operational efficiency of maintaining faults, including but not limited to those occurring in digital medical IT systems.

[0052] In one embodiment, after obtaining the fault type corresponding to the fault, the method further includes:

[0053] Obtain the fault category corresponding to each of the k local sub-sequence features, where k is a positive integer;

[0054] The preset content corresponding to the k local sub-sequence features is displayed on a preset screen.

[0055] Further, displaying the preset content corresponding to the k local sub-sequence features on a preset display screen includes:

[0056] The feature vector and fault category corresponding to each of the local sub-sequence features are displayed on a preset display screen;

[0057] The Shapelet feature corresponding to each of the local subsequence features is displayed graphically on the preset display screen.

[0058] Specifically, the fault categories corresponding to each of the k local sub-sequence features are obtained, where k is a positive integer, such as 3, 5, or 8. The feature vectors corresponding to the k local sub-sequence features and the preset content such as fault categories are displayed on a preset display screen. For example, in a medical scenario corresponding to a digital healthcare IT system, the preset content corresponding to the medical computer equipment is displayed on the display screen of the server used by the maintenance personnel of the digital healthcare IT system or on the display screen of the terminal connected to the server.

[0059] Furthermore, based on the Shapelet feature algorithm, when using Shapelet features for fault identification, the Shapelet features of the time series feature data are obtained, namely local subsequence features. There can be k local subsequence features, where k is a positive integer. The fault category corresponding to each of the k local subsequence features is obtained. The feature vector and fault category corresponding to each local subsequence feature are displayed on a preset display screen. The Shapelet feature corresponding to each local subsequence feature is also displayed graphically on the preset display screen. Thus, the preset content such as the k local subsequence features, fault category, the result of Shapelet feature transformation, and the line graph of Shapelet feature visualization are displayed on the display screen used by the operation and maintenance personnel.

[0060] In this embodiment, by obtaining the fault categories corresponding to each of k local subsequence features, and displaying the feature vectors corresponding to the k local subsequence features and the fault categories, especially when the local subsequence features are based on the Shapelet algorithm (which can be called Shapelet features), maintenance personnel can identify the k Shapelets with the highest time series similarity to the current fault indicator by viewing the results of Shapelet feature transformation and the visualized line graph of Shapelets, and find the categories corresponding to these k Shapelets. By comparing the similarity between different categories of Shapelets and fault indicators, the cause of the fault classification result output by the model can be accurately determined. Compared with traditional technologies, this method offers significant advantages. The classification algorithm in the traditional model can only provide a classification result, which cannot explain the differences between categories or understand the internal characteristics of the data. It also cannot allow maintenance personnel to interpret the model's output. This application's embodiment visualizes the k local subsequence features, feature vectors, and fault categories. In addition to being able to input them into the machine learning model to achieve higher classification accuracy, Shapelets can also make the model results more interpretable. Since each Shapelet can be visualized and the similarity between each Shapelet and the fault indicator time series can be observed, maintenance personnel of digital medical IT systems can better understand the data and model output results related to medical equipment, thereby helping maintenance personnel of digital medical IT systems and related systems to better troubleshoot and repair faults.

[0061] Furthermore, the step of displaying the Shapelet feature corresponding to each of the local sub-sequence features graphically on the preset display screen further includes:

[0062] Obtain the target fault and its corresponding type as specified by the user;

[0063] The target fault, the target fault type, and the corresponding local subsequence features are used as training samples to train the index fault classification model.

[0064] Specifically, in digital healthcare IT systems, including but not limited to, the system displays pre-defined information such as the fault categories corresponding to k local sub-sequence features, the results of Shapelet feature transformation, and line graphs visualizing Shapelet features on a monitor used by maintenance personnel. Users then compare the k local sub-sequence features with their corresponding pre-defined information and select the system fault based on their judgment. The fault corresponding to the selected local sub-sequence features is designated as the target fault, and the type of the target fault is designated as the target fault type. Target fault types include, but are not limited to, CPU faults, disk faults, and hard disk faults. The system then obtains the user-determined target fault and its corresponding target fault type. The target fault, target fault type, and corresponding local sub-sequence features are used as training samples to train an indicator fault classification model. This increases the accuracy of the indicator fault classification model in fault identification.

[0065] In one embodiment, please refer to Figure 3 , Figure 3 This is a schematic diagram of the second sub-process of the method for identifying operational and maintenance faults provided in the embodiments of this application. For example... Figure 3 As shown in this embodiment, before obtaining the index fault classification model corresponding to the index type, the method further includes:

[0066] S31. Construct the original classification model and obtain the initial indicator sample feature data based on time series corresponding to the sample faults;

[0067] S32. Based on the initial index sample feature data, obtain the Shapelet sample features using the Shapelet feature algorithm;

[0068] S33. Convert the Shapelet sample features into feature vectors to obtain sample feature vectors;

[0069] S34. The original classification model is trained using the sample feature vector to obtain the index fault classification model.

[0070] Specifically, in digital healthcare IT systems, including but not limited to, existing classification models such as logistic regression, support vector machines, random forests, and gradient boosting trees are used to construct an initial classification model. Then, initial indicator sample feature data based on time series corresponding to the sample faults are obtained. This initial indicator sample feature data consists of data with time series characteristics corresponding to a preset indicator over a certain period; that is, the time series sample feature data corresponding to each pre-set fault are obtained as training samples. For example, for computer equipment in digital medical IT systems, including but not limited to those in which poor heat dissipation leads to crashes or automatic shutdowns, the initial temperature sample feature data corresponding to the poor heat dissipation can be pre-set based on time series. For example, the temperature data of the computer equipment every 5 minutes within half an hour before the system failure occurs. For computer equipment crashes or automatic shutdowns caused by insufficient memory, the initial CPU utilization rate sample feature data can be pre-set. For example, the CPU utilization rate data every 1 minute within 10 minutes before the CPU failure occurs. For CPU damage failures such as burnout caused by the motherboard power supply exceeding the voltage that the CPU can withstand, the initial voltage sample feature data corresponding to the voltage can be pre-set. For example, the voltage value data every 2 minutes within 20 minutes before the CPU damage failure such as burnout occurs.

[0071] Based on the initial indicator sample feature data, and using the Shapelet feature algorithm, the identifiable time series subsequences contained in the initial indicator sample feature data are obtained as Shapelet features, i.e., Shapelet sample features. By using the Shapelet sample features, the time series can be accurately classified by identifying local features.

[0072] The feature vectors of shapelet samples are converted into feature vectors. This can be done using the following formula: Ax = cx, where A is a matrix, c is the eigenvalue, and x is the feature vector. Multiplying matrix A by x represents performing a transformation (rotation or stretching) on ​​vector x (a linear transformation), and the effect of this transformation is the same as multiplying vector x by a constant c (i.e., only stretching). Then, the original classification model is trained using these feature vectors to obtain the fault classification model.

[0073] In this embodiment, an original classification model is constructed, and initial indicator sample feature data based on time series corresponding to sample faults are obtained. Based on the initial indicator sample feature data, Shapelet sample features are obtained using the Shapelet feature algorithm. The Shapelet sample features are converted into feature vectors to obtain sample feature vectors. The original classification model is trained using the sample feature vectors to obtain an indicator fault classification model. By leveraging the local subsequence features corresponding to the discriminative Shapelet sample features and the changing patterns of preset indicators corresponding to different faults, the original classification model is trained based on the local subsequence features. This effectively trains the original classification model to obtain an accurate indicator fault classification model, improving the intelligence, accuracy, and efficiency of the indicator fault classification model for fault identification in, but not limited to, digital medical IT systems, thereby improving the operational efficiency of, but not limited to, digital medical IT system faults.

[0074] It should be noted that the methods for identifying operation and maintenance faults described in the above embodiments can be recombined with the technical features included in different embodiments as needed to obtain a combined implementation scheme, but all of them are within the protection scope claimed in this application.

[0075] Please see Figure 4 , Figure 4 This is a schematic block diagram of an operation and maintenance fault identification device provided in an embodiment of this application. Corresponding to the above-described operation and maintenance fault identification method, this application also provides an operation and maintenance fault identification device. Figure 4 As shown, the maintenance fault identification device includes a unit for performing the aforementioned maintenance fault identification method. This maintenance fault identification device can be configured in computer equipment, including but not limited to digital healthcare IT systems. Specifically, please refer to... Figure 4 The maintenance fault identification device 40 includes a first acquisition unit 41, a second acquisition unit 42, a third acquisition unit 43, and an identification unit 44.

[0076] The first acquisition unit 41 is used to acquire time series feature data corresponding to the preset indicators of the system if a system fault is detected.

[0077] The second acquisition unit 42 is used to filter local subsequence features from the time series feature data;

[0078] The second acquisition unit 43 is used to acquire the index type corresponding to the local subsequence features and acquire the index fault classification model corresponding to the index type.

[0079] The identification unit 44 is used to convert the local subsequence features into feature vectors, and input the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault.

[0080] In one embodiment, the local subsequence features are features based on a Shapelet feature algorithm; the second acquisition unit 42 includes:

[0081] The sub-unit is used to combine the time series feature data corresponding to several preset indicators of the system into a fault indicator time series dataset.

[0082] The first acquisition subunit is used to acquire an initial sub-feature sequence based on the fault index time series dataset, and to form an initial sub-sequence feature set from the initial sub-feature sequence.

[0083] The calculation subunit is used to classify the initial subsequence features according to the initial subsequence feature set to obtain a single-class initial subsequence feature set, and to calculate the similarity between the initial subsequence features contained in the single-class initial subsequence feature set and the corresponding time series feature data.

[0084] A statistical subunit is used to calculate a statistic for each of the initial subsequence features based on the similarity, wherein the statistic describes the classification ability of the initial subsequence features;

[0085] The filtering subunit is used to filter out initial subsequence features with high similarity based on the similarity, and obtain several target subsequence features.

[0086] The sorting subunit is used to sort all the target subsequence features according to the statistics, and to obtain the target subsequence features with the strongest classification ability corresponding to the statistics as local subsequence features.

[0087] In one embodiment, the first acquisition subunit includes:

[0088] The traversal sub-unit is used to traverse each time series feature data of the fault index time series dataset through sliding windows of different window lengths;

[0089] Extraction sub-units are used to extract feature data corresponding to different sliding windows, and feature data with the same window length are used as the initial sub-feature sequence.

[0090] In one embodiment, the identification unit 43 further includes:

[0091] The second acquisition subunit is used to acquire the fault category corresponding to each of the k local subsequence features, where k is a positive integer;

[0092] The display subunit is used to display the preset content corresponding to the k local subsequence features on a preset display screen.

[0093] In one embodiment, the display subunit includes:

[0094] The first display subunit is used to display the feature vector and fault category corresponding to each of the local subsequence features on a preset display screen;

[0095] The visualization display subunit is used to display the Shapelet feature corresponding to each of the local subsequence features graphically on the preset display screen.

[0096] In one embodiment, the identification unit 43 further includes:

[0097] The third acquisition subunit is used to acquire the target fault and the corresponding target fault type determined by the user.

[0098] The training subunit is used to train the index fault classification model using the target fault, the target fault type, and the corresponding local subsequence features as training samples.

[0099] In one embodiment, the maintenance fault identification device 40 further includes:

[0100] The building unit is used to build the original classification model and obtain the initial indicator sample feature data based on time series corresponding to sample faults;

[0101] The fourth acquisition unit is used to acquire Shapelet sample features based on the initial index sample feature data and the Shapelet feature algorithm.

[0102] A conversion unit is used to convert the Shapelet sample features into feature vectors to obtain sample feature vectors;

[0103] The training unit is used to train the original classification model using the sample feature vector to obtain the index fault classification model.

[0104] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned operation and maintenance fault identification device and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0105] Meanwhile, the division and connection methods of the various units in the above-mentioned operation and maintenance fault identification device are only for illustrative purposes. In other embodiments, the operation and maintenance fault identification device can be divided into different units as needed, and the various units in the operation and maintenance fault identification device can be connected in different sequences and methods to complete all or part of the functions of the above-mentioned operation and maintenance fault identification device.

[0106] The aforementioned fault identification device can be implemented as a computer program, which can, for example... Figure 5 The computer device shown can be running on a computer device that may be, but is not limited to, a computer device in a digital healthcare IT system.

[0107] Please see Figure 5 , Figure 5 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a desktop computer or a server, or it can be a component or part of other devices.

[0108] See Figure 5 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504, or it may be a volatile storage medium.

[0109] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, it causes the processor 502 to perform a method for identifying the aforementioned operational and maintenance faults.

[0110] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0111] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a method for identifying the above-mentioned maintenance faults.

[0112] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 5 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device 500 to which the present application is applied. A specific computer device 500 may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. For example, in some embodiments, the computer device may include only a memory and a processor. In such embodiments, the structure and function of the memory and processor are different from those shown in the figures. Figure 5 The embodiments shown are consistent and will not be described again here.

[0113] The processor 502 is used to run a computer program 5032 stored in a memory to implement the steps of the operation and maintenance fault identification method described above.

[0114] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0115] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program, which can be stored in a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0116] Therefore, this application also provides a computer-readable storage medium. This computer-readable storage medium can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, storing a computer program that, when executed by a processor, causes the processor to perform the following steps:

[0117] A computer program product, when run on a computer, causes the computer to perform the steps of the operation and maintenance fault identification method described in the above embodiments.

[0118] The computer-readable storage medium can be an internal storage unit of the aforementioned device, such as the device's hard drive or memory. The computer-readable storage medium can also be an external storage device of the device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the device.

[0119] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0120] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing computer programs.

[0121] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0122] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0123] The steps in the methods of this application embodiment can be adjusted, merged, or deleted according to actual needs. The units in the apparatus of this application embodiment can be merged, divided, or deleted according to actual needs. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0124] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0125] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An operation and maintenance fault identification method, characterized in that, include: If a system malfunction is detected, time-series feature data corresponding to preset indicators of the system are obtained. The preset indicators include indicators corresponding to at least one of the following: CPU, disk, hard disk, damaged read / write head, circuit board problem, chip information loss, motor not turning, and chip burnout. Local subsequence features are obtained by filtering from the time series feature data; Obtain the index type corresponding to the local subsequence features, and obtain the index fault classification model corresponding to the index type. The index type includes at least one of the following: CPU type index, disk type index, hard disk type index, head damage type index, circuit board problem type index, chip information loss type index, motor not turning type index, and chip burnout type index. The local subsequence features are converted into feature vectors, and the feature vectors are input into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault. The local subsequence features are features based on the Shapelet feature algorithm; the process of filtering local subsequence features from the time series feature data includes: The time series feature data corresponding to several preset indicators of the system are combined to form a fault indicator time series dataset; Based on the fault index time series dataset, an initial sub-feature sequence is obtained, and the initial sub-feature sequences are combined into an initial sub-sequence feature set. Based on the initial subsequence feature set, the initial subsequence features are classified to obtain single-class initial subsequence features, and the similarity between the initial subsequence features of the single-class initial subsequence features and the corresponding time series feature data is calculated. Based on the similarity, a statistic is calculated for each of the initial subsequence features, and the statistic describes the classification ability of the initial subsequence features; Based on the similarity, for initial subsequence features belonging to the same time series feature data, the initial subsequence features with high similarity are filtered out to obtain several target subsequence features; Based on the statistical measure, all the target subsequence features are sorted, and the target subsequence features with the strongest classification ability corresponding to the statistical measure are obtained as local subsequence features.

2. The method of claim 1, wherein, The step of obtaining the initial sub-feature sequence based on the fault index time series dataset includes: The fault index time series dataset is traversed by sliding windows of different lengths for each time series feature data. Extract the feature data corresponding to different sliding windows, and use the feature data with the same window length as the initial sub-feature sequence.

3. The method for identifying maintenance faults according to claim 1, characterized in that, After obtaining the fault type corresponding to the fault, the process further includes: Obtain the fault category corresponding to each of the k local sub-sequence features, where k is a positive integer; The preset content corresponding to the k local sub-sequence features is displayed on a preset screen.

4. The method for identifying maintenance faults according to claim 3, characterized in that, The step of displaying the preset content corresponding to the k local sub-sequence features on a preset display screen includes: The feature vector and fault category corresponding to each of the local sub-sequence features are displayed on a preset display screen; The Shapelet feature corresponding to each of the local subsequence features is displayed graphically on the preset display screen.

5. The method for identifying maintenance faults according to claim 4, characterized in that, The step of displaying the Shapelet feature corresponding to each of the local sub-sequence features graphically on the preset display screen further includes: Obtain the target fault and its corresponding type as specified by the user; The target fault, the target fault type, and the corresponding local subsequence features are used as training samples to train the index fault classification model.

6. The method for identifying maintenance faults according to claim 1, characterized in that, Before obtaining the index fault classification model corresponding to the index type, the method further includes: Construct the original classification model and obtain the initial indicator sample feature data based on time series corresponding to sample faults; Based on the initial indicator sample feature data, shapelet sample features are obtained using the shapelet feature algorithm. The shapelet sample features are converted into feature vectors to obtain the sample feature vectors; The original classification model is trained using the sample feature vectors to obtain the index fault classification model.

7. A device for identifying operational and maintenance faults, characterized in that, include: The first acquisition unit is used to acquire time series feature data corresponding to preset indicators of the system if a system failure is detected. The preset indicators include indicators corresponding to at least one of the following: CPU, disk, hard disk, damaged read / write head, circuit board problem, lost chip information, motor not turning, and chip burnout. The second acquisition unit is used to filter local subsequence features from the time series feature data; The third acquisition unit is used to acquire the index type corresponding to the local subsequence features and acquire the index fault classification model corresponding to the index type. The index type includes at least one of the following: CPU type index, disk type index, hard disk type index, head damage type index, circuit board problem type index, chip information loss type index, motor not turning type index, and chip burnout type index. The identification unit is used to convert the local subsequence features into feature vectors, and input the feature vectors into the indicator fault classification model corresponding to the indicator type to identify the fault and obtain the fault type corresponding to the fault. Wherein, the local subsequence features are features based on the Shapelet feature algorithm; the second acquisition unit includes: The sub-unit is used to combine the time series feature data corresponding to several preset indicators of the system into a fault indicator time series dataset. The first acquisition subunit is used to acquire an initial sub-feature sequence based on the fault index time series dataset, and to form an initial sub-sequence feature set from the initial sub-feature sequence. The calculation subunit is used to classify the initial subsequence features according to the initial subsequence feature set to obtain a single-class initial subsequence feature set, and to calculate the similarity between the initial subsequence features contained in the single-class initial subsequence feature set and the corresponding time series feature data. A statistical subunit is used to calculate a statistic for each of the initial subsequence features based on the similarity, wherein the statistic describes the classification ability of the initial subsequence features; The filtering subunit is used to filter out initial subsequence features with high similarity based on the similarity, and obtain several target subsequence features. The sorting subunit is used to sort all the target subsequence features according to the statistics, and to obtain the target subsequence features with the strongest classification ability corresponding to the statistics as local subsequence features.

8. A computer device, characterized in that, The computer device includes a memory and a processor connected to the memory; the memory is used to store a computer program; the processor is used to run the computer program to perform the steps of the method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, can implement the steps of the method as described in any one of claims 1-6.

Images