Hard disk fault prediction method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a hard disk fault prediction method and device, electronic equipment and a storage medium, and relates to the technical field of reliability of data storage systems.The method comprises the steps that hard disk data are obtained, data cleaning and feature selection are conducted on the hard disk data to obtain feature data, and the feature data are stored in a database; performing standardization and normalization on the feature data to obtain a data set, using a long-short term memory network as a deep learning model, designing a model structure, introducing a regularization technology and a dropout layer to construct an initial model, training and optimizing the initial model by using the data set to obtain a trained prediction model, and performing prediction on the trained prediction model. And performing state monitoring, fault early warning and maintenance on the hard disk by using the prediction model. According to the method, efficient and accurate hard disk state monitoring, fault early warning and maintenance are achieved through data preprocessing and the LSTM network deep learning model, and the problems that in the prior art, changes of the running state of the hard disk cannot be accurately captured, and data loss and system crash cannot be prevented are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of reliability of data storage systems, and in particular to a hard disk failure prediction method, device, electronic equipment and storage medium. Background Art

[0002] With the increasing popularity of computer applications, hard disks are key devices for data storage, and their stability and reliability are directly related to data security and the normal operation of the system. Hard disk failure has become an issue that cannot be ignored. It may lead to data loss and system crash, which in turn causes huge economic losses and information security risks. At present, hard disk failure prediction technology has been developed to a certain extent and is widely used in data centers, cloud storage, data backup for large enterprises and individual users.

[0003] However, existing prediction methods are unable to accurately capture the complex changes in the operating status of hard disks, resulting in low prediction accuracy and unable to meet the requirements of data storage systems for high reliability and stability.

[0004] Specifically, current hard drive failure prediction technology faces the following shortcomings and difficulties: First, the data collection and processing capabilities are insufficient, making it difficult to obtain comprehensive and accurate hard drive working status data; second, the prediction model is not accurate enough to effectively identify early signs of hard drive failure; third, it has poor adaptability and is difficult to adapt to the prediction needs of different types and brands of hard drives.

[0005] Therefore, a hard disk failure prediction method is needed that can improve prediction accuracy and has strong adaptability and generalization capabilities. Summary of the invention

[0006] The embodiments of the present invention provide a hard disk failure prediction method to solve the problem that the prior art cannot accurately capture the change of the hard disk operation status and cannot prevent data loss and system crash. The technical solution is as follows:

[0007] According to one aspect of the present invention, a hard disk failure prediction method includes: obtaining hard disk data; the hard disk data includes temperature, vibration, current, historical data, number of read and write operations and average response time; performing data cleaning and feature selection on the hard disk data to obtain feature data, and standardizing and normalizing the feature data to obtain a data set; using a long short-term memory network as a deep learning model, designing a model structure, introducing regularization technology and a dropout layer to construct an initial model; using the data set to train and optimize the initial model to obtain a trained prediction model, and using the prediction model to perform status monitoring, fault warning and maintenance on the hard disk.

[0008] In one of the embodiments, obtaining hard disk data is achieved through the following steps: setting the frequency of data collection, installing a sensor for the hard disk, and obtaining key parameters of the hard disk through the sensor at the frequency; the key parameters include temperature, vibration, and current; reading the drive log of the hard disk at the frequency to obtain historical status information, and obtaining operating data of the hard disk through the operating system; the operating data includes the number of read and write operations and the average response time.

[0009] In one of the embodiments, obtaining hard disk data further includes the following steps: storing the collected key parameters, historical status information and operating data in an extensible database, and marking the fault data in the historical status information; the database includes a relational database and a distributed file system.

[0010] In one of the embodiments, data cleaning and feature selection are performed on the hard disk data to obtain feature data, and standardization and normalization are performed on the feature data to obtain a data set through the following steps: missing values ​​in the hard disk data are processed by interpolation or deletion, and outlier identification and processing are performed on the hard disk data using a statistical algorithm or model calculation to obtain processed hard disk data; feature selection is performed on the processed hard disk data to obtain feature data, and standardization and normalization are performed on the feature data to make the data scales of different features consistent, so as to obtain a data set.

[0011] In one of the embodiments, designing the model structure, introducing regularization technology and dropout layer to build the initial model is achieved by the following steps: designing the number of layers of the initial model and the number of nodes in each layer according to the complexity and scale of the data set, adding a regularization term to the loss function of the initial model, and setting the retention ratio to introduce a dropout layer to build the initial model.

[0012] In one embodiment, the training and optimization of the initial model using the data set to obtain a trained prediction model is achieved by the following steps: dividing the data set into a training set, a validation set, and a test set, selecting a loss function, an optimization algorithm, and a performance indicator; using the training set and the loss function to train the initial model, using the optimization algorithm to adjust parameters according to the performance of the initial model in the validation set, and obtaining a trained prediction model when the performance of the initial model in the test set reaches the performance indicator.

[0013] In one of the embodiments, the use of the prediction model to perform status monitoring, fault warning and maintenance on the hard disk is achieved through the following steps: designing a monitoring system for the hard disk, deploying the prediction model into the monitoring system, and maintaining and regularly updating the prediction model according to the actual operation of the prediction model; using the prediction model to predict the working status of the hard disk to determine whether the hard disk has a risk of failure; if the hard disk has a risk of failure, issuing a warning and taking maintenance measures; the maintenance measures include backing up data and replacing the hard disk.

[0014] According to one aspect of the present invention, a hard disk fault prediction device includes: a data acquisition module for acquiring hard disk data; the hard disk data includes temperature, vibration, current, historical data, number of read and write operations and average response time; a data processing module for performing data cleaning and feature selection on the hard disk data to obtain feature data, and standardizing and normalizing the feature data to obtain a data set; a model construction module for using a long short-term memory network as a deep learning model, and designing a model structure, introducing regularization technology and a dropout layer to construct an initial model; a hard disk fault prediction module for using the data set to train and optimize the initial model to obtain a trained prediction model, and using the prediction model to perform status monitoring, fault warning and maintenance on the hard disk.

[0015] According to one aspect of the present invention, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements a hard disk failure prediction method as described above.

[0016] According to one aspect of the present invention, a storage medium stores computer-readable instructions thereon, and the computer-readable instructions are executed by one or more processors to implement a hard disk failure prediction method as described above.

[0017] The beneficial effects brought by the technical solution provided by the present invention are:

[0018] In the above technical scheme, the present invention first obtains hard disk data, performs data cleaning and feature selection on the hard disk data to obtain feature data, standardizes and normalizes the feature data to obtain a data set, uses a long short-term memory network as a deep learning model, designs a model structure, introduces regularization technology and a dropout layer to build an initial model, uses the data set to train and optimize the initial model to obtain a trained prediction model, uses the prediction model to perform status monitoring, fault warning and maintenance on the hard disk, and realizes efficient and accurate hard disk status monitoring, fault warning and maintenance through data preprocessing and a long short-term memory network (LSTM) deep learning model, and effectively eliminates redundant information through data preprocessing and feature selection. Information is collected, which improves data quality and ensures the stability and consistency of model input. The LSTM model is used to capture the subtle features of the hard disk status over time. The regularization and dropout layers are introduced to effectively prevent model overfitting and enhance the generalization performance of the model. The trained prediction model can monitor the hard disk status in real time and accurately, and warn of potential failures in advance, providing a reliable basis for timely maintenance, significantly reducing the hard disk failure rate and extending its service life. At the same time, it also reduces the losses caused by data loss or service interruption, and improves the overall stability and reliability of the system, thereby effectively solving the problem that the existing technology cannot accurately capture the changes in the hard disk operating status and cannot prevent data loss and system crashes. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in describing the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0020] Figure 1 is a flow chart of a hard disk failure prediction method according to an exemplary embodiment;

[0021] Figure 2 is a flow chart of a hard disk failure prediction method shown in an exemplary embodiment;

[0022] Figure 3 is a flow chart of a hard disk failure prediction method shown in another exemplary embodiment;

[0023] Figure 4 yes Figure 3 Comparison diagram of the effects of the corresponding embodiment and the prior art on the test sample;

[0024] Figure 5 yes Figure 3 A comparison chart of the generalization performance of the corresponding embodiment and the prior art on the test sample;

[0025] Figure 6 is a block diagram of a hard disk failure prediction device according to an exemplary embodiment;

[0026] Figure 7 is a hardware structure diagram of an electronic device according to an exemplary embodiment;

[0027] Figure 8 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0028] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0029] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0030] To this end, the present invention provides a hard disk failure prediction method, which realizes efficient and accurate hard disk status monitoring, fault warning and maintenance through data preprocessing and long short-term memory network (LSTM) deep learning model, and can effectively solve the problem that the existing technology cannot accurately capture the changes in the hard disk operation status and cannot prevent data loss and system crash. The hard disk failure prediction method is suitable for a hard disk failure prediction device, which can be an electronic device. A hard disk failure prediction method in an embodiment of the present invention can be applied to a variety of scenarios, such as hard disk failure prediction.

[0031] See also Figure 1 , an embodiment of the present invention provides a hard disk failure prediction method, which is applicable to electronic equipment.

[0032] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0033] like Figure 1 As shown, the method may include the following steps:

[0034] Step 110, obtaining hard disk data.

[0035] Among them, hard disk data includes temperature, vibration, current, historical data, number of read and write operations, and average response time.

[0036] In one possible implementation, the frequency of data collection is set, a sensor is installed for the hard disk, key parameters of the hard disk are obtained through the sensor according to the frequency, the drive log of the hard disk is read according to the frequency to obtain historical status information, and the operation data of the hard disk is obtained through the operating system.

[0037] Among them, key parameters include temperature, vibration, and current, and operating data includes the number of read and write operations and the average response time.

[0038] In a possible implementation, the collected key parameters, historical status information, and operating data are stored in an expandable database, and the fault data in the historical status information is marked.

[0039] Among them, the database includes a relational database and a distributed file system.

[0040] Step 130 , performing data cleaning and feature selection on the hard disk data to obtain feature data, and standardizing and normalizing the feature data to obtain a data set.

[0041] In one possible implementation, missing values ​​in hard disk data are processed by interpolation or deletion, outliers in hard disk data are identified and processed using statistical algorithms or model calculations to obtain processed hard disk data, feature selection is performed on the processed hard disk data to obtain feature data, and the feature data is standardized and normalized to make the data scales of different features consistent, thereby obtaining a data set.

[0042] Step 150, using the long short-term memory network as a deep learning model, and designing the model structure to introduce regularization technology and a dropout layer to build an initial model.

[0043] In one possible implementation, the number of layers and the number of nodes in each layer of the initial model are designed according to the complexity and scale of the dataset, a regularization term is added to the loss function of the initial model, and a retention ratio is set to introduce a dropout layer to construct the initial model.

[0044] Step 170: Use the data set to train and optimize the initial model to obtain a trained prediction model, and use the prediction model to perform status monitoring, fault warning and maintenance on the hard disk.

[0045] In one possible implementation, the data set is divided into a training set, a validation set, and a test set, a loss function, an optimization algorithm, and a performance indicator are selected, the initial model is trained using the training set and the loss function, the optimization algorithm is used to adjust the parameters according to the performance of the initial model in the validation set, and the trained prediction model is obtained when the performance of the initial model in the test set reaches the performance indicator.

[0046] In one possible implementation, a prediction model is used to predict the working status of the hard disk to determine whether the hard disk has a risk of failure. If the hard disk has a risk of failure, an early warning is issued and maintenance measures are taken.

[0047] Among them, maintenance measures include backing up data and replacing hard drives.

[0048] Through the above process, the embodiment of the present invention first obtains hard disk data, performs data cleaning and feature selection on the hard disk data to obtain feature data, standardizes and normalizes the feature data to obtain a data set, uses a long short-term memory network as a deep learning model, designs a model structure, introduces regularization technology and a dropout layer to build an initial model, uses the data set to train and optimize the initial model to obtain a trained prediction model, uses the prediction model to perform status monitoring, fault warning and maintenance on the hard disk, and realizes efficient and accurate hard disk status monitoring, fault warning and maintenance through data preprocessing and a long short-term memory network (LSTM) deep learning model, and effectively eliminates redundant Information is collected, which improves data quality and ensures the stability and consistency of model input. The LSTM model is used to capture the subtle features of the hard disk status over time. The regularization and dropout layers are introduced to effectively prevent model overfitting and enhance the generalization performance of the model. The trained prediction model can monitor the hard disk status in real time and accurately, and warn of potential failures in advance, providing a reliable basis for timely maintenance, significantly reducing the hard disk failure rate and extending its service life. At the same time, it also reduces the losses caused by data loss or service interruption, and improves the overall stability and reliability of the system, thereby effectively solving the problem that the existing technology cannot accurately capture the changes in the hard disk operating status and cannot prevent data loss and system crashes.

[0049] In an exemplary embodiment, the present invention provides a hard disk failure prediction method for hard disk failure prediction.

[0050] like Figure 2 As shown, the following steps may be specifically included:

[0051] Step 210: Data collection and preprocessing.

[0052] In this step, the hard disk working status data is comprehensively collected through various methods to provide reliable data support for subsequent deep learning model training.

[0053] Specifically, sensor data collection: Install sensors on hard disk devices to monitor and collect key parameters of hard disk operation in real time, such as temperature, vibration, current, etc. These data can directly reflect the working status and performance changes of the hard disk.

[0054] Furthermore, drive log collection: by reading the drive log of the hard disk, the historical working status information of the hard disk is obtained. This information includes the use time of the hard disk, read and write operation records, etc., which is helpful for analyzing the life and failure trend of the hard disk.

[0055] Furthermore, operating system data collection: The operating system provides a wealth of hard disk related data, such as the number of read and write operations, average response time, etc. These data can reflect the performance of the hard disk at the operating system level.

[0056] Furthermore, real-time monitoring data collection: set a reasonable data collection frequency to ensure that enough data points are obtained for training and validating the model. Collect these real-time monitoring data through network protocols or local interfaces to ensure the integrity and accuracy of the data.

[0057] Step 220: Data preprocessing.

[0058] After data collection is completed, data preprocessing is performed, including data cleaning and feature extraction. Data cleaning aims to deal with missing values ​​and outliers, and improves the data set through interpolation methods or deletion of missing values; outliers are identified and processed through statistical methods or model-based methods. Feature selection is to select the most representative hard disk operation features from many features to reduce the complexity of the model and improve the prediction performance.

[0059] Specifically, data cleaning: in addition to processing missing values ​​and outliers, it is also necessary to ensure the integrity and consistency of the data and remove duplicate or invalid data; feature selection: use feature selection algorithms (such as XGBoost, etc.) to extract the features that have the greatest impact on hard disk life prediction, reduce the input dimensions of the model, and improve prediction efficiency; standardization and normalization: standardize and normalize features of different scales to ensure that their importance is balanced during the model training process, which is conducive to improving the prediction accuracy of the model.

[0060] Step 230: Model selection.

[0061] In this step, the long short-term memory network (LSTM) suitable for sequence data is selected as the deep learning model.

[0062] Specifically, the deep learning architecture: LSTM is a special recurrent neural network (RNN) with long-term memory capabilities, which can capture long-term dependencies in sequence data and is suitable for time series data analysis tasks such as hard disk life prediction; according to the complexity and scale of the data, the number of layers and nodes of each layer of the LSTM model are determined, and regularization technology and dropout layers are introduced to prevent overfitting.

[0063] Step 240: Model training.

[0064] Specifically, in the model training phase, the LSTM model is trained using historical data to optimize model parameters. The data set is divided into a training set, a validation set, and a test set, which are used for model training, parameter adjustment, and performance evaluation, respectively. An appropriate loss function (such as binary cross entropy) is selected to measure the difference between the model output and the true fault label. A gradient descent optimization algorithm (such as Adam or RMSprop) is used to optimize model parameters and improve prediction accuracy. A learning rate adjustment strategy (such as learning rate decay) is used to ensure that the model converges to the optimal solution faster during training.

[0065] Step 250: Model evaluation and optimization.

[0066] Specifically, in the model evaluation and optimization stage, the model parameters are adjusted by evaluating the model's performance indicators (such as accuracy, recall rate, precision, etc.) to improve the model's generalization ability; appropriate performance indicators are selected according to task requirements to evaluate the model's performance; model parameters (such as the number of layers, number of nodes, learning rate, etc.) are adjusted according to the performance of the validation set to improve the model's predictive performance; and ensemble learning methods (such as voting or stacking) are used to further improve the model's predictive performance.

[0067] Step 260: Deploy monitoring.

[0068] Specifically, in the deployment and monitoring phase, the trained LSTM model is deployed to the hard disk monitoring system to perform real-time hard disk failure prediction. The trained model is integrated into the hard disk monitoring system to realize the real-time fault prediction function. The monitoring system is designed to monitor the performance of the model in actual operation in real time, and adjust the model parameters or update the model according to the feedback to adapt to the changes in the hard disk operation data.

[0069] Step 270: Predict failure.

[0070] Specifically, the trained LSTM model is used to predict the current working status of the hard disk and determine whether there is a risk of failure. The current working status data of the hard disk is input into the model, and the model outputs the prediction result based on the input data, that is, the possibility of hard disk failure, and analyzes whether the hard disk has a risk of failure based on the prediction result.

[0071] Step 280: Early warning and maintenance.

[0072] Specifically, when the prediction results show that the hard disk has a risk of failure, an early warning is issued in time and corresponding maintenance measures are taken. Early warning notification: Early warning notification is issued to relevant personnel through email, SMS, etc. Maintenance measures: Take corresponding maintenance measures according to the early warning results, such as backing up data, replacing hard disks, etc., to avoid data loss and system crashes.

[0073] Through the above process, the embodiment of the present invention provides a comprehensive and reliable data foundation for analysis and prediction. The subsequent data preprocessing further improves the data quality and ensures the accuracy of model training. Model selection and model training use deep learning algorithms to effectively capture the operating characteristics of the hard disk and improve the accuracy of prediction. Model evaluation and optimization ensure that the model performance is optimal, and deployment monitoring realizes real-time prediction. Ultimately, through predicting faults and early warning and maintenance, timely measures can be taken to avoid data loss and system crashes, significantly improving the reliability and stability of the data storage system.

[0074] In another exemplary embodiment, the present invention provides a hard disk failure prediction method for hard disk failure prediction.

[0075] like Figure 3 As shown, the following steps may be specifically included:

[0076] Step 310: Data preprocessing.

[0077] Specifically, the collected hard disk working status data is preprocessed, including data cleaning, such as missing value processing (using interpolation methods or deleting missing values) and outlier processing (using statistical methods or model-based methods to identify and process). In addition, feature selection is required to select the most representative hard disk operation features to reduce the complexity of the model. Finally, the data is standardized and normalized to ensure that the scales of different features are consistent, which is conducive to subsequent model training.

[0078] Step 320: Feature analysis.

[0079] Specifically, the preprocessed data is analyzed using algorithms such as XGBoost to extract key features that have a significant impact on the hard drive life prediction. This step helps to further optimize the model and improve the accuracy of the prediction.

[0080] Step 330: Convert to 3D data.

[0081] Specifically, the feature-analyzed data is converted into a 3D data format suitable for deep learning models such as LSTM. This step is necessary before model training to ensure that the data can match the model architecture.

[0082] Step 340: Build a model.

[0083] Specifically, according to the characteristics of the hard disk working data and the prediction requirements, select a suitable deep learning model, such as the long short-term memory network (LSTM). Determine the number of layers and the number of nodes in each layer of the model, and introduce regularization and dropout techniques to prevent overfitting.

[0084] Step 350: Model fitting.

[0085] Specifically, the established deep learning model is trained using historical data, and the model parameters are adjusted through optimization algorithms (such as Adam or RMSprop) to improve the prediction accuracy. During the training process, the data set is divided (training set, validation set, and test set), and a suitable loss function (such as binary cross entropy) is selected to measure the difference between the model output and the actual fault label.

[0086] Step 360: Model prediction.

[0087] Specifically, the trained model is applied to new hard disk working data to predict failures. The model output is used to determine whether the hard disk has a failure risk, providing a decision-making basis for subsequent early warning and maintenance.

[0088] Step 370: generalization ability test.

[0089] Specifically, the generalization ability of the model is tested using data from hard drives of different models or brands to verify the versatility and applicability of the model. The generalization performance of the model is evaluated by comparing its performance on the test samples, and the model is further optimized and adjusted based on the test results.

[0090] Furthermore, after determining the optimal parameters of the model, a long short-term memory network (LSTM) model with a time step of 20 days was selected to predict the remaining life of the hard disk for 40 days.

[0091] like Figure 4 As shown, on the test samples, the predicted value (LSTM) of the prediction model of the embodiment of the present invention is relatively close to the true value (Actual RUL) as a whole, with an average absolute error of 1.85 on the test samples, while the average absolute error of the RNN model (RNN) on the test samples is 2.66.

[0092] Among them, the embodiment of the present invention can detect hard disk failure 40 days in advance, which is significantly earlier than the prior art. Since the embodiment of the present invention solves the defects of long-term dependence and gradient explosion in the prior art, the prediction effect of the embodiment of the present invention is better than the prior art. In addition, the present invention is general and focuses on exploring the selection of hard disk characteristics and important parameters of deep learning networks, which can be used for fault prediction of other models of hard disks.

[0093] like Figure 5 As shown in the figure, the generalization performance of the model was tested using hard disk data produced by another manufacturer. The MAE of the pre-model (LSTM) and RNN model (RNN) of the embodiment of the present invention on the test samples were 7.94 and 10.86 respectively. It can be seen that the prediction effect of the embodiment of the present invention is significantly better than that of the prior art RNN model.

[0094] Through the above process, the embodiment of the present invention effectively improves the data quality by cleaning and organizing the original data, providing a more accurate and reliable data basis for subsequent analysis, and further screens out the most representative hard disk operation characteristics through algorithms such as XGBoost, reducing the complexity of model training and improving the accuracy of prediction. The model parameters are optimized through model fitting, so that it can better capture the long-term dependencies in the hard disk operation data, providing enterprises and individuals with a valuable time window to take timely measures for maintenance or replacement, thereby avoiding serious consequences such as data loss and system crashes. The generalization ability test verifies the model's prediction ability for hard disks of different models, demonstrating its strong adaptability and broad application prospects, and making important contributions to improving the reliability and stability of data storage systems.

[0095] The following is an embodiment of the device of the present invention, which can be used to execute a hard disk failure prediction method involved in the present invention. For details not disclosed in the embodiment of the device of the present invention, please refer to the method embodiment of a hard disk failure prediction method involved in the present invention.

[0096] See also Figure 6 , an embodiment of the present invention provides a hard disk failure prediction device 800.

[0097] The device 800 includes but is not limited to: a data acquisition module 810 , a data processing module 830 , a model building module 850 and a hard disk failure prediction module 870 .

[0098] The data acquisition module 810 is used to acquire hard disk data; the hard disk data includes temperature, vibration, current, historical data, number of read and write operations, and average response time.

[0099] The data processing module 830 is used to perform data cleaning and feature selection on the hard disk data to obtain feature data, and to standardize and normalize the feature data to obtain a data set.

[0100] The model building module 850 is used to use the long short-term memory network as a deep learning model, design the model structure, introduce regularization technology and dropout layer to build an initial model.

[0101] The hard disk fault prediction module 870 is used to use the data set to train and optimize the initial model to obtain a trained prediction model, and use the prediction model to perform status monitoring, fault warning and maintenance on the hard disk.

[0102] It should be noted that the hard disk failure prediction provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the hard disk failure prediction device will be divided into different functional modules to complete all or part of the functions described above.

[0103] In addition, the hard disk fault prediction device provided in the above embodiment and an embodiment of a hard disk fault prediction method belong to the same concept, and the specific manner in which each module performs operations has been described in detail in the method embodiment and will not be repeated here.

[0104] Figure 7 The present invention schematically illustrates the structure of an electronic device according to an exemplary embodiment.

[0105] It should be noted that the electronic device is only an example adapted to the present invention and cannot be considered to provide any limitation on the scope of use of the present invention. The electronic device cannot be interpreted as requiring dependence on or having Figure 7 One or more components of exemplary electronic device 2000 are shown.

[0106] The hardware structure of the electronic device 2000 may vary greatly due to differences in configuration or performance, such as Figure 7 As shown, the electronic device 2000 includes: a power supply 210 , an interface 230 , at least one memory 250 , and at least one central processing unit (CPU) 270 .

[0107] Specifically, the power supply 210 is used to provide operating voltage for each hardware device on the electronic device 2000 .

[0108] The interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. Of course, in other examples adapted by the present invention, the interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input-output interface 235, and at least one USB interface 237, etc. Figure 7 As shown, this is not intended to be a specific limitation.

[0109] The memory 250 is a carrier for storing resources, which may be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon include an operating system 251, an application 253 and data 255, etc. The storage method may be temporary storage or permanent storage.

[0110] Among them, the operating system 251 is used to manage and control various hardware devices and application programs 253 on the electronic device 2000 to enable the central processor 270 to calculate and process the massive data 255 in the memory 250. It can be WindowsServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0111] The application program 253 is a computer-readable instruction that performs at least one specific task based on the operating system 251, and may include at least one module ( Figure 7 (not shown), each module may respectively include computer-readable instructions for the electronic device 2000. For example, the hard disk failure prediction device may be regarded as an application 253 deployed in the electronic device 2000.

[0112] The data 255 may be signal information, etc., stored in the memory 250 .

[0113] The central processor 270 may include one or more processors and is configured to communicate with the memory 250 through at least one communication bus to read the computer-readable instructions stored in the memory 250, thereby realizing the operation and processing of the mass data 255 in the memory 250. For example, a hard disk failure prediction method is completed by the central processor 270 reading a series of computer-readable instructions stored in the memory 250.

[0114] In addition, the present invention can also be implemented by hardware circuits or hardware circuits combined with software. Therefore, the implementation of the present invention is not limited to any specific hardware circuits, software, or a combination of the two.

[0115] See also Figure 8 In an embodiment of the present invention, an electronic device 4000 is provided. The electronic device 400 may include: a desktop computer, a laptop computer, a server, etc. with sensor recognition capability.

[0116] exist Figure 8 In the embodiment, the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0117] The data interaction between the processor 4001 and the memory 4003 can be realized through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0118] Optionally, the electronic device 4000 may further include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception, etc. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0119] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0120] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program instructions or codes in the form of instructions or data structures and can be accessed by the electronic device 400, but is not limited to this.

[0121] Computer-readable instructions are stored in the memory 4003 , and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002 .

[0122] The computer-readable instructions are executed by one or more processors 4001 to implement a hard disk failure prediction method in the above-mentioned embodiments.

[0123] In addition, an embodiment of the present invention provides a storage medium on which computer-readable instructions are stored. The computer-readable instructions are executed by one or more processors to implement a hard disk failure prediction method as described above.

[0124] A computer program product is provided in an embodiment of the present invention. The computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements a hard disk failure prediction method as described above.

[0125] Compared with the related art, the present invention has the following beneficial effects:

[0126] 1. The present invention first obtains hard disk data, performs data cleaning and feature selection on the hard disk data to obtain feature data, standardizes and normalizes the feature data to obtain a data set, uses a long short-term memory network as a deep learning model, designs a model structure, introduces regularization technology and a dropout layer to build an initial model, uses the data set to train and optimize the initial model to obtain a trained prediction model, uses the prediction model to perform status monitoring, fault warning and maintenance on the hard disk, and realizes efficient and accurate hard disk status monitoring, fault warning and maintenance through data preprocessing and a long short-term memory network (LSTM) deep learning model. Redundant information is effectively eliminated through data preprocessing and feature selection, and the prediction model is improved. It improves data quality and ensures the stability and consistency of model input. It captures subtle features of the hard disk status over time through the LSTM model. It introduces regularization and dropout layers to effectively prevent model overfitting and enhance the generalization performance of the model. The trained prediction model can monitor the hard disk status in real time and accurately, and warn of potential failures in advance, providing a reliable basis for timely maintenance, significantly reducing the hard disk failure rate and extending its service life. It also reduces losses caused by data loss or service interruptions, and improves the overall stability and reliability of the system, thereby effectively solving the problem that existing technologies cannot accurately capture changes in the hard disk operating status and cannot prevent data loss and system crashes.

[0127] 2. The present invention utilizes a deep learning algorithm to model and predict the hard disk working status data, which can more accurately predict the possibility of hard disk failure and improve the accuracy of the prediction; it can adapt to the prediction needs of different types of hard disks, has strong generalization capabilities, and can be applied to multiple models of hard disks; by timely predicting hard disk failures and taking measures to maintain or replace them, data loss and system crashes can be avoided, thereby improving the reliability and stability of the data storage system.

[0128] 3. The present invention has important application value in the field of data storage and is expected to become an important technological innovation in this field; the present invention can perform preventive maintenance before hard disk failure occurs through accurate prediction, thereby reducing maintenance costs and data recovery costs caused by hard disk failure; the present invention reduces data loss and system crashes caused by hard disk failure, and can improve users' storage experience and usage satisfaction.

[0129] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0130] The above descriptions are only some embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A hard disk failure prediction method, characterized in that: The method comprises: Obtain hard disk data; the hard disk data includes temperature, vibration, current, historical data, number of read and write operations, and average response time; Performing data cleaning and feature selection on the hard disk data to obtain feature data, and standardizing and normalizing the feature data to obtain a data set; Use long short-term memory network as a deep learning model, design the model structure, introduce regularization technology and dropout layer to build the initial model; The initial model is trained and optimized using the data set to obtain a trained prediction model, and the prediction model is used to perform status monitoring, fault warning and maintenance on the hard disk.

2. A hard disk failure prediction method as claimed in claim 1, characterized in that: The obtaining of hard disk data includes: Setting a frequency of data collection, installing a sensor for the hard disk, and obtaining key parameters of the hard disk through the sensor at the frequency; the key parameters include temperature, vibration, and current; The drive log of the hard disk is read at the frequency to obtain historical status information, and the operation data of the hard disk is obtained through the operating system; the operation data includes the number of read and write operations and the average response time.

3. A hard disk failure prediction method as claimed in claim 2, characterized in that: The method further comprises: The key parameters, historical status information and operation data collected are stored in an expandable database, and the fault data in the historical status information is marked; the database includes a relational database and a distributed file system.

4. A hard disk failure prediction method as claimed in claim 1, characterized in that: The step of performing data cleaning and feature selection on the hard disk data to obtain feature data, and performing standardization and normalization on the feature data to obtain a data set includes: Processing missing values ​​in the hard disk data by interpolation or deletion, and identifying and processing abnormal values ​​of the hard disk data by using a statistical algorithm or model calculation to obtain processed hard disk data; Feature selection is performed on the processed hard disk data to obtain feature data, and the feature data is standardized and normalized to make the data scales of different features consistent, thereby obtaining a data set.

5. A hard disk failure prediction method as claimed in claim 1, characterized in that: The design of the model structure, introduction of regularization technology and dropout layer to build the initial model include: The number of layers and the number of nodes in each layer of the initial model are designed according to the complexity and scale of the data set, a regularization term is added to the loss function of the initial model, and a retention ratio is set to introduce a dropout layer to construct the initial model.

6. A hard disk failure prediction method as claimed in claim 1, characterized in that: The using the data set to train and optimize the initial model to obtain a trained prediction model includes: Dividing the data set into a training set, a validation set, and a test set, and selecting a loss function, an optimization algorithm, and a performance indicator; The initial model is trained using the training set and the loss function, and the optimization algorithm is used to adjust parameters according to the performance of the initial model in the validation set, and a trained prediction model is obtained when the performance of the initial model in the test set reaches the performance index.

7. A hard disk failure prediction method as claimed in claim 1, characterized in that: The using of the prediction model to perform status monitoring, fault warning and maintenance on the hard disk includes: Designing a monitoring system for the hard disk, deploying the prediction model into the monitoring system, and maintaining and regularly updating the prediction model according to the actual operation of the prediction model; Using the prediction model to predict the working status of the hard disk, and determining whether the hard disk has a failure risk; If the hard disk has a risk of failure, an early warning is issued and maintenance measures are taken; the maintenance measures include backing up data and replacing the hard disk.

8. A hard disk failure prediction device, characterized in that: The device comprises: A data acquisition module, used to acquire hard disk data; the hard disk data includes temperature, vibration, current, historical data, number of read and write operations, and average response time; A data processing module, used for performing data cleaning and feature selection on the hard disk data to obtain feature data, and standardizing and normalizing the feature data to obtain a data set; Model building module, which is used to use the long short-term memory network as a deep learning model, design the model structure, introduce regularization technology and dropout layer to build the initial model; The hard disk fault prediction module is used to use the data set to train and optimize the initial model to obtain a trained prediction model, and use the prediction model to perform status monitoring, fault warning and maintenance on the hard disk.

9. An electronic device, characterized in that: include: at least one processor and at least one memory, wherein: The memory has computer-readable instructions stored thereon; The computer-readable instructions are executed by one or more of the processors, so that the electronic device implements a hard disk failure prediction method as described in any one of claims 1 to 7.

10. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer-readable instructions are executed by one or more processors to implement a hard disk failure prediction method as described in any one of claims 1 to 7.

Images