Service life prediction method of storage device, electronic device and storage medium

Abstract

The invention discloses a service life prediction method of a storage device, an electronic device and a storage medium, and the method comprises the steps: controlling a scanning device to scan the storage device, and obtaining an internal structure image; controlling a sound monitoring device to carry out sound monitoring on the storage equipment to obtain sound data; acquiring electrical data, temperature data and error monitoring data of the storage device; inputting various data into a trained multi-modal service life prediction model; and performing multi-modal feature fusion on the various data in a multi-modal feature fusion layer of the multi-modal service life prediction model to obtain a fused feature vector, and inputting the fused feature vector into a service life prediction decision layer to obtain the predicted service life of the storage device. According to the method and the device, the multi-modal feature fusion is carried out on various data, and the predicted service life of the storage equipment can be calculated according to the multi-modal data, so that the prediction accuracy is improved, and the requirement of accurately predicting the service life of the storage equipment in practical application is met.

Description

Technical Field

[0001] The present invention relates to the technical field of the next generation information network industry, and in particular to a method for predicting the service life of a storage device, an electronic device, and a storage medium. Background Art

[0002] With the rapid development of information technology, storage devices, as core components for data storage and management, are widely used in cloud computing data centers, enterprise server clusters, personal computers, and various mobile smart devices. With the explosive growth of data volumes, the workload of storage devices is increasing. Predicting the lifespan of storage devices is crucial for ensuring data security, optimizing device maintenance strategies, and reducing operating costs. For example, in large data centers, predicting the lifespan of storage devices in advance allows for the preemptive replacement of backup storage devices, avoiding data loss and business interruption caused by sudden storage device failures. Currently, existing storage device lifespan prediction methods only predict the lifespan based on the storage device's electrical data, temperature data, or error logs, which can all be obtained by reading the storage device.

[0003] However, the aging of storage devices is a complex process, which is affected by a variety of factors such as internal structural changes, mechanical wear and electrical performance degradation. Predicting the service life based solely on the electrical data, temperature data or error logs of the storage device will result in low accuracy of the prediction results, which is difficult to meet the needs of accurate prediction of the service life of storage devices in practical applications. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present application provides a storage device service life prediction method, electronic device and storage medium. By performing multimodal feature fusion on internal structure images, sound data, electrical data, temperature data and error monitoring data in the multimodal feature fusion layer of the multimodal service life prediction model, a fused feature vector is obtained, and the fused feature vector is input into the life prediction decision layer to obtain the predicted service life of the storage device. The predicted service life of the storage device can be calculated based on the multimodal data, thereby improving the accuracy of the prediction and meeting the needs of accurate prediction of the service life of storage devices in practical applications.

[0005] In order to solve the above problems, the present invention provides the following technical solutions: In a first aspect, an embodiment of the present application provides a method for predicting the service life of a storage device, comprising: controlling a scanning device to scan the storage device to obtain an internal structure image of the storage device; controlling a sound monitoring device to perform sound monitoring on the storage device to obtain sound data when the storage device is working; Acquiring electrical data, temperature data, and error monitoring data of the storage device, wherein the electrical data includes a voltage curve and a current curve when the storage device is operating, the temperature data includes an operating temperature curve when the storage device is operating, and the error monitoring data includes hard disk stop failure records, write error records, erase error records, read error records, and data verification error records; Inputting the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data into a trained multimodal service life prediction model; In the multimodal service life prediction model, multimodal feature fusion is performed on the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data to obtain a fused feature vector, and the fused feature vector is input into a service life prediction decision layer to obtain a predicted service life of the storage device, wherein the multimodal service life prediction model includes a multimodal data preprocessing layer, a multimodal feature fusion layer, a service life prediction decision layer, and a calibration layer; The multimodal data preprocessing layer is used to perform data preprocessing on the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data, and extract feature vectors corresponding to the plurality of data; The multimodal feature fusion layer includes a reliability weight calculation module, a multimodal feature alignment module, an error type association module and a fusion operation module. The reliability weight calculation module is used to calculate the weights corresponding to the internal structure image, the sound data, the electrical data, the temperature data and the error monitoring data. The multimodal feature alignment module is used to map the input multiple feature vectors to the semantic space and obtain multiple aligned feature vectors. The error type association module is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors. The fusion operation module is used to perform multimodal feature fusion according to the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain the fused feature vector. The life prediction decision layer is used to calculate the initial predicted service life of the storage device based on the fused feature vector; The calibration layer is used to calibrate the initial predicted service life of the storage device to obtain a final predicted service life.

[0006] In some embodiments, the reliability weight calculation module includes a spatiotemporal consistency calculation unit, a priori weight determination unit, and a weight calculation function unit. The spatiotemporal consistency calculation unit is used to calculate the spatiotemporal consistency indicators corresponding to multiple data; the priori weight determination unit is used to determine the priori weights corresponding to multiple data based on multiple physical models; the weight calculation function unit is used to use a preset weight calculation function to calculate the weight corresponding to each data based on the spatiotemporal consistency indicators and prior weights corresponding to multiple data, and the weight calculation function includes trained adjustment parameters.

[0007] In some embodiments, the spatiotemporal consistency calculation unit is configured to, when the data is two-dimensional data, divide the data into a plurality of data segments according to a preset time period, extract a feature vector of each data segment, and calculate the similarity of the feature vectors of each group of two adjacent data segments to obtain a temporal similarity sequence, and calculate the spatiotemporal consistency index of the data based on the temporal similarity sequence; The spatiotemporal consistency calculation unit is used to divide the data into multiple spatial units according to a preset spatial unit size when the data is three-dimensional data, extract the feature vector of each spatial unit and calculate the similarity of the feature vectors of each group of two adjacent spatial units to obtain a spatial similarity sequence, and calculate the spatiotemporal consistency index of the data based on the spatial similarity sequence.

[0008] In some embodiments, the error type association module includes an error type feature library and a graph neural network unit, the error type feature library includes error types, corresponding feature vectors, associated data types and physical models, the graph neural network unit is used to use the feature vectors in the error type feature library as edge weights, and construct a multimodal feature interaction graph based on the multiple aligned feature vectors, wherein one node in the multimodal feature interaction graph is one of the aligned feature vectors; The graph neural network unit is also used to use a graph convolution layer to aggregate information of adjacent nodes in the multimodal feature interaction graph and update the alignment feature vector represented by each node to obtain multiple associated feature vectors, which contain constraint information of the error type.

[0009] In some embodiments, the life prediction decision layer is further used to input the fused feature vector into a fully connected layer, input the vector output by the fully connected layer into an activation layer, and obtain the probability of each error type occurring in the storage device output by the activation layer.

[0010] In some embodiments, the calibration layer is also used to obtain the device type, usage scenario type and usage duration of the storage device, obtain the corresponding calibration coefficient from the calibration error information based on the device type, the usage scenario type and the usage duration, and calibrate the initial predicted service life of the storage device based on the calibration coefficient to obtain the final predicted service life.

[0011] In some embodiments, the calibration error information is a calibration error table representing a multi-dimensional mapping relationship among device type, usage scenario type, usage duration, and calibration coefficient determined based on training data. In the calibration error table, a combination of a device type, usage scenario type, and a duration interval of usage duration corresponds to one calibration coefficient.

[0012] In some embodiments, the method further comprises: Controlling the three-dimensional scanning device to scan the storage device to obtain three-dimensional point cloud data of the storage device; The step of inputting the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data into a trained multimodal service life prediction model includes: The internal structure image, the sound data, the electrical data, the temperature data, the error monitoring data and the three-dimensional point cloud data are input into a trained multimodal service life prediction model.

[0013] In a second aspect, an embodiment of the present application provides an electronic device, comprising: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the service life prediction method of the storage device as described in the first aspect.

[0014] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores an executable program, and the executable program is executed by a processor to implement the service life prediction method of the storage device as described in the first aspect.

[0015] The present application provides a method for predicting the service life of a storage device, an electronic device, and a storage medium. The present application performs multimodal feature fusion on internal structure images, sound data, electrical data, temperature data, and error monitoring data in a multimodal feature fusion layer of a multimodal service life prediction model to obtain a fused feature vector, and inputs the fused feature vector into a service life prediction decision layer to obtain a predicted service life of the storage device. The predicted service life of the storage device can be calculated based on the multimodal data, thereby improving the accuracy of the prediction and meeting the demand for accurate prediction of the service life of the storage device in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for predicting the service life of a storage device provided in an embodiment of the present application.

[0017] Figure 2 It is a structural diagram of the multimodal service life prediction model provided in the embodiment of the present application.

[0018] Figure 3 It is a structural diagram of the multimodal feature fusion layer provided in an embodiment of the present application.

[0019] Figure 4 It is a structural diagram of the reliability weight calculation module provided in an embodiment of the present application.

[0020] Figure 5 It is a structural diagram of a storage device service life prediction device provided in an embodiment of the present application.

[0021] Figure 6 This is a structural diagram of an electronic device provided in an embodiment of the present application.

[0022] Figure 7 This is a structural block diagram of a computer-readable storage medium provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0024] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0025] The present application provides a storage device service life prediction method, electronic device, and storage medium. The method performs multimodal feature fusion on internal structure images, sound data, electrical data, temperature data, and error monitoring data in a multimodal feature fusion layer of a multimodal service life prediction model to obtain a fused feature vector, and inputs the fused feature vector into a service life prediction decision layer to obtain a predicted service life of the storage device. The predicted service life of the storage device can be calculated based on the multimodal data, thereby improving the accuracy of the prediction and meeting the demand for accurate prediction of the service life of the storage device in practical applications.

[0026] The storage devices referred to in this application may be mechanical hard disks, solid-state drives (SSDs), optical disks, or memory cards. The SSDs may be ball grid array (BGA) SSDs. BGA SSDs may include Flash memory chips, which may include NOR Flash and NAND Flash. The following example uses a BGA SSD as an example, but this should not be considered a limitation of this application.

[0027] The storage device of this application is a computer component, and therefore this application relates to computer component manufacturing. The storage device of this application can be used in databases and cloud computing centers, and therefore this application also relates to the fields of Internet and cloud computing, big data services, and specifically to database and cloud database services.

[0028] The storage device lifespan prediction method of this application is used to predict the lifespan of storage devices during operational preparation or idle time, allowing personnel to preemptively replace spare storage devices based on the predicted lifespan, thereby avoiding data loss and business interruption caused by sudden storage device failures. This method also effectively detects or locates faulty hardware through testing.

[0029] The following will describe in detail the storage device service life prediction method provided by the present application with reference to the accompanying drawings.

[0030] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for predicting the service life of a storage device provided in an embodiment of the present application. Figure 1As shown, the storage device service life prediction method includes: steps S100 to S500.

[0031] Step S100: controlling a scanning device to scan a storage device to obtain an internal structure image of the storage device.

[0032] In some embodiments, the internal structure image of the storage device is an X-ray image, an ultrasound image, or a magnetic resonance image.

[0033] Preferably, the internal structure image of the storage device is an ultrasonic image. In this case, the scanning device is an ultrasonic imaging device, which is small in size and has no radiation, making it easy to perform scanning operations.

[0034] In some embodiments, the internal structure image of the storage device may include multiple images obtained by continuously scanning the storage device, or may be a single image.

[0035] Step S200: controlling the sound monitoring device to perform sound monitoring on the storage device to obtain sound data when the storage device is working.

[0036] When the storage device is a mechanical hard drive, the sound it makes during operation is related to the head seek operation or the platter rotation speed. Mechanical hard drives produce regular sounds during normal operation. Any abnormal noise could indicate head wear, platter scratches, or bearing failure, which can shorten the lifespan. When the storage device is an SSD, abnormal current noise could indicate a capacitor or controller failure. Therefore, it's possible to predict the lifespan of a storage device based on its operating sound data.

[0037] Step S300: Acquire electrical data, temperature data, and error monitoring data of the storage device.

[0038] Electrical data includes voltage and current curves during storage device operation. Temperature data includes the operating temperature curve during storage device operation. Error monitoring data includes error records such as hard drive stop failure records, write error records, erase error records, read error records, and data verification error records.

[0039] Optionally, the temperature data also includes an ambient temperature curve when the storage device is operating.

[0040] Step S400: Input the internal structure image, sound data, electrical data, temperature data and error monitoring data into the trained multimodal service life prediction model.

[0041] Step S500: In the multimodal service life prediction model, multimodal feature fusion is performed on the internal structure image, sound data, electrical data, temperature data and error monitoring data to obtain a fused feature vector, and the fused feature vector is input into the service life prediction decision layer to obtain the predicted service life of the storage device.

[0042] See also Figure 2 , Figure 2 Schematic diagram of the structure of the multimodal service life prediction model provided in the embodiment of the present application. Figure 2 As shown, in some embodiments, the multimodal service life prediction model 1 includes a multimodal data preprocessing layer 10, a multimodal feature fusion layer 20, a service life prediction decision layer 30 and a calibration layer 40.

[0043] In some embodiments, the multimodal data preprocessing layer 10 is used to perform data preprocessing on internal structure images, sound data, electrical data, temperature data, and error monitoring data, extract feature vectors corresponding to multiple data, and input the feature vectors corresponding to the multiple data into the multimodal feature fusion layer 20.

[0044] In some embodiments, the multimodal data preprocessing layer includes an image preprocessing module and an image feature extraction module. The image preprocessing module is used to perform image processing operations such as enhancement and denoising on the internal structure image, and the image feature extraction module is used to extract feature vectors corresponding to the processed image.

[0045] In some embodiments, the image feature extraction module is a convolutional neural network model, which can extract feature vectors corresponding to the processed image through convolution operations, pooling operations, and fully connected layers.

[0046] As described above, the internal structure images of the storage device may be multiple images obtained by continuously scanning the storage device. In this case, the image feature extraction module may include a trunk branch module, a detail branch module, and a result vector module. The trunk branch module is used to extract two-dimensional feature vectors from the multiple internal structure images, and the detail branch module is used to extract three-dimensional feature vectors based on the multiple internal structure images. The result vector module is used to concatenate the two-dimensional feature vectors and the three-dimensional feature vectors to obtain a feature vector corresponding to the internal structure image.

[0047] In some embodiments, the result vector module is further used to reduce the dimension of the vector obtained by concatenating the two-dimensional feature vector and the three-dimensional feature vector to obtain a feature vector corresponding to the internal structure image.

[0048] In some embodiments, when the internal structure image is a single image, the feature extraction module may include only a trunk branch module and not a detail branch module. In this case, the trunk branch module is used to extract a two-dimensional feature vector of the single internal structure image.

[0049] In some embodiments, the trunk-branch module includes multiple convolutional layers and fully connected layers connected in sequence. The trunk-branch module is configured to progressively extract two-dimensional features of the internal structure image through the multiple convolutional layers to obtain a feature map. The two-dimensional features may include, for example, the morphological features of chip solder joints and the appearance features of capacitors. The fully connected layer is configured to convert the feature map output by the convolutional layer into a two-dimensional feature vector.

[0050] In some embodiments, the detail branch module may be a 3D image neural network model. The detail branch module includes multiple 3D convolutional layers, multiple 3D pooling layers, and multiple fully connected layers connected in sequence. The 3D convolutional layers and the 3D pooling layers are used to extract 3D geometric features based on multiple internal structure images. The 3D geometric features may include 3D geometric features between stacked layers of storage units.

[0051] In some embodiments, the multimodal data preprocessing layer further includes a sound data preprocessing module and a sound feature extraction module. The sound data preprocessing module is configured to perform spectral analysis on the sound data to obtain spectral data. The sound feature extraction module is configured to extract, based on the spectral data, sound signals in the sound data whose frequencies fall within any one of a plurality of preset frequency bands, thereby obtaining sound signals corresponding to the plurality of preset frequency bands, and extract a feature vector for each sound signal to obtain a feature vector corresponding to the sound data.

[0052] Optionally, the multiple preset frequency bands include a normal operating sound band and various types of abnormal operating sound bands. The various types of abnormal operating sound bands may include an overload operating sound band and a short circuit operating sound band. If the storage device is a mechanical hard disk, the various types of abnormal operating sound bands may also include operating sound bands caused by head wear, platter scratches, and bearing failures.

[0053] In some embodiments, the multimodal data preprocessing layer further includes an electrical data preprocessing module configured to extract characteristic vectors of voltage and current curves in the electrical data and concatenate the characteristic vectors of the voltage and current curves into a characteristic vector corresponding to the electrical data.

[0054] In some embodiments, the electrical data preprocessing module is a time-recursive neural network model. A time-recursive neural network model is a recurrent neural network model specifically designed for processing time series data. By introducing recurrent connections between hidden layers, a time-recursive neural network model enables information to be transferred between different time steps. This design enables the time-recursive neural network model to capture dependencies and patterns in time series.

[0055] Optionally, the characteristic vectors of the voltage curve and the current curve include periodic fluctuation characteristic vectors and abnormal fluctuation characteristic vectors.

[0056] Optionally, in the time recursive neural network model, a sliding window of a preset length is used to extract the periodic fluctuation feature vectors of the voltage curve and the current curve.

[0057] Optionally, the preset length is a median of erase and write cycles of the storage device.

[0058] Exemplarily, the time recursive neural network model may be a long short-term memory network model.

[0059] In some embodiments, the multimodal data preprocessing layer also includes a temperature data preprocessing module, which is used to calculate the self-heating temperature curve of the storage device based on the ambient temperature curve and the operating temperature curve in the temperature data, and extract the characteristic vector corresponding to the temperature data based on the self-heating temperature curve.

[0060] In some embodiments, in the temperature data preprocessing module, for each time point, the temperature value of the working temperature curve at that time point is subtracted from the temperature value of the ambient temperature curve at the same time point, and the resulting difference is the self-heating temperature of the storage device at that time point, thereby obtaining the self-heating temperature curve of the storage device.

[0061] In some embodiments, the multimodal data preprocessing layer further includes an error monitoring data preprocessing module configured to calculate a feature vector corresponding to the error monitoring data based on the type and time of the error records in the error monitoring data. The feature vector corresponding to the error monitoring data represents the frequency of various errors occurring in the storage device during various usage time periods.

[0062] See also Figure 3 , Figure 3 Schematic diagram of the structure of the multimodal feature fusion layer provided in the embodiment of the present application. Figure 3 As shown, in some embodiments, the multimodal feature fusion layer 20 includes a reliability weight calculation module 21 , a multimodal feature alignment module 22 , an error type association module 23 and a fusion operation module 24 .

[0063] Among them, the reliability weight calculation module 21 is used to calculate the weights corresponding to the internal structure image, sound data, electrical data, temperature data and error monitoring data. The multimodal feature alignment module 22 is used to map the input multiple feature vectors to the semantic space and obtain multiple aligned feature vectors. The error type association module 23 is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors. The fusion operation module 24 is used to perform multimodal feature fusion based on the weights corresponding to the multiple data output by the reliability weight calculation module 21 and the multiple associated feature vectors output by the error type association module 23 to obtain a fused feature vector.

[0064] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of the reliability weight calculation module provided in the embodiment of the present application. Figure 4 As shown, in some embodiments, the reliability weight calculation module 21 includes a spatiotemporal consistency calculation unit 211 , a priori weight determination unit 212 and a weight calculation function unit 213 .

[0065] The spatiotemporal consistency calculation unit 211 is configured to calculate spatiotemporal consistency indicators corresponding to multiple data sets. The prior weight determination unit 212 is configured to determine prior weights corresponding to the multiple data sets based on multiple physical models. The weight calculation function unit 213 is configured to use a preset weight calculation function to calculate the weight corresponding to each data set based on the spatiotemporal consistency indicators and prior weights corresponding to the multiple data sets. The weight calculation function includes trained adjustment parameters. The multiple data sets include internal structure images, sound data, electrical data, temperature data, and error monitoring data.

[0066] In some implementations, for two-dimensional data, the spatiotemporal consistency index can be used to reflect the degree of similarity between features at adjacent time points in the temporal dimension. For three-dimensional data, the spatiotemporal consistency index can be used to reflect the similarity of features across different regions in the spatial dimension. A higher value for this index indicates greater spatiotemporal consistency of the modal data and higher data reliability.

[0067] Alternatively, electrical data, sound data, temperature data, and error monitoring data may be classified as two-dimensional data.

[0068] Optionally, when the internal structure image includes a plurality of images, the internal structure image may be classified as three-dimensional data.

[0069] In some embodiments, the spatiotemporal consistency calculation unit is used to divide the data into multiple data segments according to a preset time period when the data is two-dimensional data, extract the feature vector of each data segment and calculate the similarity of the feature vectors of each group of two adjacent data segments to obtain a time similarity sequence, and calculate the spatiotemporal consistency index of the data based on the time similarity sequence.

[0070] Optionally, the cosine similarity of the feature vectors of each group of two adjacent data segments is calculated to obtain a time similarity sequence including multiple cosine similarities, and the average value of the cosine similarities in the time similarity sequence is used as the spatiotemporal consistency indicator of the data.

[0071] Optionally, when the internal structure image includes multiple images, the spatiotemporal consistency calculation unit is further configured to obtain an internal structure model of the storage device based on the multiple internal structure images. The internal structure model of the storage device is three-dimensional data. When the data is three-dimensional data, the spatiotemporal consistency calculation unit is configured to divide the data into multiple spatial units according to a preset spatial unit size, extract the feature vector of each spatial unit, and calculate the similarity between the feature vectors of each pair of adjacent spatial units to obtain a spatial similarity sequence, and calculate the spatiotemporal consistency index of the data based on the spatial similarity sequence.

[0072] Optionally, the cosine similarity of the feature vectors of each group of two adjacent spatial units is calculated to obtain a spatial similarity sequence including multiple cosine similarities, and the average value of the cosine similarities in the spatial similarity sequence is used as the spatiotemporal consistency indicator of the data.

[0073] In some embodiments, the prior weight determination unit has multiple built-in physical models, which may include a model determined in advance based on physical principles for describing the relationship between storage unit failure and temperature and time, and a model for describing the relationship between solder joint fatigue and temperature, etc.

[0074] In some embodiments, the prior weight determination unit is configured to determine a usage scenario type of the storage device based on at least one of the plurality of data, and determine the prior weights corresponding to the plurality of data based on a physical model corresponding to the usage scenario type. For example, when the operating temperature of the storage device is determined to be greater than a preset temperature based on the temperature data, the usage scenario type of the storage device is determined to be a high-temperature scenario, and a model corresponding to the high-temperature scenario, which describes the relationship between storage unit failure, temperature, and time, is selected to determine the prior weights corresponding to the plurality of data. In a high-temperature scenario, the prior weight corresponding to the temperature data is higher than the prior weight corresponding to a general scenario.

[0075] In some embodiments, the weight calculation function is: ,in, Indicates the The weight corresponding to each data point, represents the adjustment parameter, Indicates the The spatiotemporal consistency index corresponding to the data, represents the historical average spatiotemporal consistency index, Indicates the The prior weight of each data point.

[0076] Optionally, and It can be obtained by pre-training the multimodal service life prediction model.

[0077] In some embodiments, the role of the adjustment parameter is to control the weight of and The larger the value of the adjustment parameter, the steeper the function curve, which means that when and When the difference in changes slightly, the weight will change significantly.

[0078] In some embodiments, the multimodal feature alignment module is used to map feature vector quantities corresponding to multiple data to a semantic space and obtain multiple aligned feature vectors, each of which has the same spatial dimension for subsequent calculations.

[0079] In some implementations, since the sampling rates of multiple data may be different, there may be a problem of misalignment in the time dimension. In this case, the multimodal feature alignment module is also used to map the feature vector quantities corresponding to the multiple data to the semantic space, and then use the dynamic time warping algorithm to flexibly match the time series of the data, so that the two feature vectors can be aligned on the time axis through scaling and translation, thereby achieving alignment of each feature vector in the time dimension.

[0080] In some embodiments, the error type association module 23 includes an error type feature library and a graph neural network unit. The error type feature library includes error types, corresponding feature vectors, associated data types, and physical models. The graph neural network unit is configured to use feature vectors in the error type feature library as edge weights and construct a multimodal feature interaction graph based on multiple aligned feature vectors. A node in the modal feature interaction graph is an aligned feature vector.

[0081] In some embodiments, the error types include memory cell failure, circuit open, circuit short, solder joint fatigue, and exceeding a safe temperature threshold.

[0082] For example, data types associated with storage cell failure include error monitoring data. Data types associated with circuit breaks include electrical data, error monitoring data, and internal structure images. Data types associated with circuit shorts include electrical data, error monitoring data, and internal structure images. Data types associated with solder joint fatigue include electrical data and internal structure images. Data types associated with exceeding a safety temperature threshold include temperature data and sound data. This is because when the safety temperature threshold is exceeded, the operating noise of the storage device is often relatively high.

[0083] Optionally, the graph neural network unit is also used to use a graph convolution layer to aggregate the information of adjacent nodes in the multimodal feature interaction graph and update the aligned feature vector represented by each node to obtain multiple associated feature vectors. Among them, the associated feature vector contains constraint information of the error type. In this way, the fusion of multimodal features can be preliminarily achieved. Moreover, the associated feature vector can be calculated based on prior knowledge, and the associated feature vector contains constraint information of the error type, so that the output of the multimodal service life prediction model can conform to the laws of physics.

[0084] Optionally, there are multiple graph convolution layers, which are connected in sequence, and each graph convolution layer updates the alignment feature vector represented by each node by aggregating information from adjacent nodes. In some embodiments, the fusion operation module is configured to perform multimodal feature fusion based on the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain a fused feature vector.

[0085] Optionally, the fusion operation module is used to perform weighted summation on the associated feature vectors corresponding to each data according to the weight corresponding to each data to obtain a fused feature vector. In this way, the fusion of multimodal features can be ultimately achieved.

[0086] In some embodiments, the lifespan prediction decision layer is used to calculate an initial predicted service life of the storage device based on the fused feature vector.

[0087] Optionally, the life prediction decision layer includes a main task regression module, which includes multiple fully connected layers and activation functions connected in sequence. The first fully connected layer of the main task regression module is used to input the fused feature vector and perform calculations. The activation function of the main task regression module is used to determine the initial predicted service life of the storage device based on the output of the last fully connected layer.

[0088] In some embodiments, the life prediction decision layer is further used to input the fused feature vector into the fully connected layer, input the vector output by the fully connected layer into the activation layer, and obtain the probability of each error type occurring in the storage device output by the activation layer.

[0089] Optionally, the life prediction decision layer is a multi-task learning architecture and also includes an auxiliary task module, which includes multiple fully connected layers and activation layers connected in sequence. The first fully connected layer of the auxiliary task module is used to input the fused feature vector and perform calculations, and the activation layer of the auxiliary task module is used to determine the probability of each error type occurring in the storage device based on the output of the last fully connected layer.

[0090] In some embodiments, when training a multimodal service life prediction model, the loss function used by the main task regression module of the service life prediction decision layer is a mean square error loss function or a root mean square error loss function, etc.

[0091] In some embodiments, when training a multimodal service life prediction model, the auxiliary task module of the service life prediction decision layer uses a cross-entropy loss function. The cross-entropy loss function is used to evaluate the difference between the probability distribution of the probability of each error type occurring in the storage device during training and the true distribution.

[0092] Optionally, the fully connected layer includes multiple groups of neurons, each group of neurons corresponding to one error type.

[0093] As described above, in some embodiments, error types include memory cell failure, circuit breakage, circuit short circuit, solder joint fatigue, and exceeding a safe temperature threshold. In this way, the calculation results of the multimodal service life prediction model can be interpreted, allowing users to understand which errors affect the service life of the storage device.

[0094] In some embodiments, the calibration layer is used to calibrate the initial predicted service life of the storage device to obtain a final predicted service life.

[0095] In some implementations, the calibration layer is further used to obtain the device type, usage scenario type, and usage duration of the storage device. Based on the device type, usage scenario type, and usage duration, a corresponding calibration coefficient is obtained from the calibration error information. The initial predicted service life of the storage device is calibrated based on the calibration coefficient to obtain a final predicted service life. In this way, the accuracy of the prediction can be further improved.

[0096] Optionally, log data of the storage device may be obtained, and the device type and usage duration of the storage device may be determined based on the log data. For example, it may be determined that the storage device is an SSD and has been in use for 1000 hours.

[0097] Optionally, the usage scenario type may be determined by the a priori weight determination unit based on at least one data among a plurality of data, or may be acquired according to a received user instruction.

[0098] In some embodiments, the calibration error information is a calibration error table determined based on training data and representing a multi-dimensional mapping relationship between device type, usage scenario type, usage duration, and calibration coefficient.

[0099] Optionally, in the calibration error table, a combination of a device type, a usage scenario type, and a duration interval of the usage duration corresponds to a calibration coefficient.

[0100] In some embodiments, the training data is divided into three dimensions according to device type, usage scenario type, and usage duration, and a calibration error table is obtained by pre-training based on part of the training data corresponding to the combination of each device type, usage scenario type, and usage duration interval.

[0101] Optionally, the duration interval may be preset or obtained through training.

[0102] For example, when the storage device is an SSD, the usage scenario type is a general scenario, and the usage time is 2000 hours, the usage time interval can be determined to be 1500 hours to 2500 hours according to the calibration error table, and the calibration coefficient corresponding to the combination of the current device type, usage scenario type, and usage time interval is 0.1.

[0103] In some embodiments, the initial calibration formula for predicting service life is: ,in, represents the final predicted service life, represents the initial predicted service life, represents the calibration coefficient, Represents the base of natural logarithms.

[0104] in, Can be any value. The value can be obtained through training.

[0105] In some embodiments, when training a multimodal service life prediction model, the loss function used in the calibration layer is a mean square error loss function, a logarithmic loss function, or a mean absolute error loss function.

[0106] In some embodiments, the method further includes step S600.

[0107] Step S600: controlling the three-dimensional scanning device to scan the storage device to obtain three-dimensional point cloud data of the storage device.

[0108] Three-dimensional point cloud data can be classified as three-dimensional data.

[0109] At this time, step S400 includes: inputting the internal structure image, sound data, electrical data, temperature data, error monitoring data and three-dimensional point cloud data into the trained multimodal service life prediction model.

[0110] The processing of multiple data in the multimodal service life prediction model refers to the description in step S500.

[0111] In summary, the storage device service life prediction method provided by the embodiment of the present application has the following advantages: 1. By performing multimodal feature fusion on internal structure images, sound data, electrical data, temperature data, and error monitoring data in the multimodal feature fusion layer of the multimodal service life prediction model, a fused feature vector is obtained, and the fused feature vector is input into the service life prediction decision layer to obtain the predicted service life of the storage device. The predicted service life of the storage device can be calculated based on the multimodal data, thereby improving the accuracy of the prediction and meeting the needs of accurate prediction of the service life of storage devices in practical applications.

[0112] 2. The graph neural network unit is also used to aggregate information from adjacent nodes in the multimodal feature interaction graph using a graph convolutional layer and update the aligned feature vectors representing each node to obtain multiple associated feature vectors, which can initially achieve multimodal feature fusion. Furthermore, the associated feature vectors can be calculated based on prior knowledge. The associated feature vectors contain constraint information about the error type, thus ensuring that the output of the multimodal service life prediction model conforms to physical laws.

[0113] 3. The life prediction decision layer is also used to input the fused feature vector into the fully connected layer, and the vector output by the fully connected layer is input into the activation layer. The probability of each error type occurring in the storage device is obtained from the output of the activation layer. This method can explain the calculation results of the multimodal service life prediction model, allowing users to know which errors affect the service life of the storage device.

[0114] See also Figure 5 , Figure 5 This is a schematic diagram of the structure of the device for predicting the service life of a storage device provided in an embodiment of the present application. Figure 5 As shown, the storage device service life prediction apparatus 300 includes an acquisition module 310 and a processing module 320 .

[0115] In some embodiments, the acquisition module 310 is used to control the scanning device to scan the storage device to obtain an internal structure image of the storage device; control the sound monitoring device to perform sound monitoring on the storage device to obtain sound data when the storage device is working; obtain the electrical data, temperature data and error monitoring data of the storage device, the electrical data includes the voltage curve and current curve when the storage device is working, the temperature data includes the working temperature curve when the storage device is working, and the error monitoring data includes hard disk stop failure records, write error records, erase error records, read error records and data verification error records.

[0116] In some embodiments, the processing module 320 is used to input the internal structure image, sound data, electrical data, temperature data and error monitoring data into a trained multimodal service life prediction model; in the multimodal feature fusion layer of the multimodal service life prediction model, the internal structure image, sound data, electrical data, temperature data and error monitoring data are multimodally fused to obtain a fused feature vector, and the fused feature vector is input into the life prediction decision layer to obtain a predicted service life of the storage device, wherein the multimodal service life prediction model includes a multimodal data preprocessing layer, a multimodal feature fusion layer, a life prediction decision layer and a calibration layer; the multimodal data preprocessing layer is used to perform data preprocessing on the internal structure image, sound data, electrical data, temperature data and error monitoring data, and extract feature vectors corresponding to multiple data; the multimodal feature fusion layer includes a reliability weight calculation module Block, multimodal feature alignment module, error type association module and fusion operation module, the reliability weight calculation module is used to calculate the weights corresponding to the internal structure image, sound data, electrical data, temperature data and error monitoring data, the multimodal feature alignment module is used to map the multiple input feature vectors to the semantic space and obtain multiple aligned feature vectors, the error type association module is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors, the fusion operation module is used to perform multimodal feature fusion according to the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain a fused feature vector; the life prediction decision layer is used to calculate the initial predicted service life of the storage device based on the fused feature vector; the calibration layer is used to calibrate the initial predicted service life of the storage device to obtain the final predicted service life.

[0117] See also Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 As shown, the electronic device 400 includes: one or more processors 410 and a memory 420, Figure 6 A processor 410 is taken as an example.

[0118] In some embodiments, the processor 410 and the memory 420 may be connected via a bus or other means. Figure 6 The bus connection is taken as an example.

[0119] In some embodiments, the processor 410 is used to control the scanning device to scan the storage device to obtain an internal structure image of the storage device; control the sound monitoring device to perform sound monitoring on the storage device to obtain sound data of the storage device when it is working; obtain electrical data, temperature data and error monitoring data of the storage device, the electrical data includes a voltage curve and a current curve when the storage device is working, the temperature data includes an operating temperature curve when the storage device is working, and the error monitoring data includes a hard disk stop failure record, a write error record, an erase error record, a read error record and a data verification error record; input the internal structure image, sound data, electrical data, temperature data and error monitoring data into a trained multimodal service life prediction model; perform multimodal feature fusion on the internal structure image, sound data, electrical data, temperature data and error monitoring data in the multimodal feature fusion layer of the multimodal service life prediction model to obtain a fused feature vector, and input the fused feature vector into a life prediction decision layer to obtain a predicted service life of the storage device, wherein the multimodal service life prediction model includes a multimodal data preprocessing layer, a multimodal feature fusion layer, a life prediction layer and a service life prediction layer. Decision layer and calibration layer; the multimodal data preprocessing layer is used to preprocess the internal structure image, sound data, electrical data, temperature data and error monitoring data, and extract the feature vectors corresponding to the multiple data; the multimodal feature fusion layer includes a reliability weight calculation module, a multimodal feature alignment module, an error type association module and a fusion operation module. The reliability weight calculation module is used to calculate the weights corresponding to the internal structure image, sound data, electrical data, temperature data and error monitoring data. The multimodal feature alignment module is used to map the input multiple feature vectors to the semantic space and obtain multiple aligned feature vectors. The error type association module is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors. The fusion operation module is used to perform multimodal feature fusion based on the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain a fused feature vector; the life prediction decision layer is used to calculate the initial predicted service life of the storage device based on the fused feature vector; the calibration layer is used to calibrate the initial predicted service life of the storage device to obtain the final predicted service life.

[0120] In some embodiments, memory 420, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules of the storage device service life prediction method in the embodiments of the present application. Processor 410 executes the non-volatile software programs, instructions, and modules stored in memory 420 to execute various functional applications and data processing of electronic device 400, thereby implementing the storage device service life prediction method in the above-mentioned method embodiment.

[0121] In some embodiments, the memory 420 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device 400, etc. In addition, the memory 420 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 420 may optionally include a memory remotely located relative to the processor 410, and these remote memories may be connected to the controller via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0122] In some embodiments, one or more modules are stored in the memory 420, and when executed by one or more processors 410, perform the storage device service life prediction method in any of the above method embodiments, for example, perform the above described Figure 1 Method steps S100 to S500.

[0123] Please refer to Figure 7 , Figure 7 The computer-readable storage medium 500 stores program code 510, which can be called by a processor to execute the storage device service life prediction method described in the above method embodiment.

[0124] Computer-readable storage medium 500 can be an electronic memory such as flash memory, electrically erasable programmable read-only memory (EEPROM), a hard disk, or read-only memory (ROM). Alternatively, the computer-readable storage medium includes non-volatile computer-readable media. Computer-readable storage medium 500 has storage space for program code that executes any of the steps in the above-described method for predicting the useful life of a storage device. This program code can be read from or written to one or more computer program products. The program code can be compressed, for example, in a suitable format.

[0125] The present application also provides a computer program product, including a computer program, which implements the above-mentioned storage device service life prediction method when executed by a processor.

[0126] In summary, the present application provides a method for predicting the service life of a storage device, an electronic device, and a storage medium. The method for predicting the service life of the storage device includes: controlling a scanning device to scan the storage device to obtain an internal structure image of the storage device; controlling a sound monitoring device to perform sound monitoring on the storage device to obtain sound data when the storage device is working; obtaining electrical data, temperature data, and error monitoring data of the storage device, the electrical data including a voltage curve and a current curve when the storage device is working, the temperature data including an operating temperature curve when the storage device is working, and the error monitoring data including a hard disk stop failure record, a write error record, an erase error record, a read error record, and a data verification error record; inputting the internal structure image, sound data, electrical data, temperature data, and error monitoring data into a trained multimodal service life prediction model; performing multimodal feature fusion on the internal structure image, sound data, electrical data, temperature data, and error monitoring data in the multimodal feature fusion layer of the multimodal service life prediction model to obtain a fused feature vector, and inputting the fused feature vector into a life prediction decision layer to obtain a predicted service life of the storage device. The multimodal service life prediction model includes a multimodal data preprocessing layer, a multimodal feature fusion layer, a service life prediction decision layer and a calibration layer; the multimodal data preprocessing layer is used to preprocess the internal structure image, sound data, electrical data, temperature data and error monitoring data, and extract the feature vectors corresponding to the multiple data; the multimodal feature fusion layer includes a reliability weight calculation module, a multimodal feature alignment module, an error type association module and a fusion operation module, the reliability weight calculation module is used to calculate the weights corresponding to the internal structure image, sound data, electrical data, temperature data and error monitoring data, the multimodal feature alignment module is used to map the input multiple feature vectors to the semantic space and obtain multiple aligned feature vectors, the error type association module is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors, and the fusion operation module is used to perform multimodal feature fusion based on the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain a fused feature vector; the service life prediction decision layer is used to calculate the initial predicted service life of the storage device based on the fused feature vector; the calibration layer is used to calibrate the initial predicted service life of the storage device to obtain the final predicted service life. This application performs multimodal feature fusion on internal structure images, sound data, electrical data, temperature data and error monitoring data in the multimodal feature fusion layer of the multimodal service life prediction model to obtain a fused feature vector, and inputs the fused feature vector into the life prediction decision layer to obtain the predicted service life of the storage device. The predicted service life of the storage device can be calculated based on the multimodal data, thereby improving the accuracy of the prediction and meeting the demand for accurate prediction of the service life of the storage device in practical applications.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting the service life of a storage device, characterized in that: include: Controlling the scanning device to scan the storage device to obtain an internal structure image of the storage device; controlling a sound monitoring device to perform sound monitoring on the storage device to obtain sound data when the storage device is working; Acquiring electrical data, temperature data, and error monitoring data of the storage device, wherein the electrical data includes a voltage curve and a current curve when the storage device is operating, the temperature data includes an operating temperature curve when the storage device is operating, and the error monitoring data includes hard disk stop failure records, write error records, erase error records, read error records, and data verification error records; Inputting the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data into a trained multimodal service life prediction model; In the multimodal service life prediction model, multimodal feature fusion is performed on the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data to obtain a fused feature vector, and the fused feature vector is input into a service life prediction decision layer to obtain a predicted service life of the storage device, wherein the multimodal service life prediction model includes a multimodal data preprocessing layer, a multimodal feature fusion layer, a service life prediction decision layer, and a calibration layer; The multimodal data preprocessing layer is used to perform data preprocessing on the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data, and extract feature vectors corresponding to the plurality of data; The multimodal feature fusion layer includes a reliability weight calculation module, a multimodal feature alignment module, an error type association module and a fusion operation module. The reliability weight calculation module is used to calculate the weights corresponding to the internal structure image, the sound data, the electrical data, the temperature data and the error monitoring data. The multimodal feature alignment module is used to map the input multiple feature vectors to the semantic space and obtain multiple aligned feature vectors. The error type association module is used to associate the multiple aligned feature vectors with the error type to obtain multiple associated feature vectors. The fusion operation module is used to perform multimodal feature fusion according to the weights corresponding to the multiple data output by the reliability weight calculation module and the multiple associated feature vectors output by the error type association module to obtain the fused feature vector. The life prediction decision layer is used to calculate the initial predicted service life of the storage device based on the fused feature vector; The calibration layer is used to calibrate the initial predicted service life of the storage device to obtain a final predicted service life.

2. The method for predicting the service life of a storage device according to claim 1, wherein: The reliability weight calculation module includes a spatiotemporal consistency calculation unit, a priori weight determination unit and a weight calculation function unit. The spatiotemporal consistency calculation unit is used to calculate the spatiotemporal consistency indicators corresponding to multiple data; the priori weight determination unit is used to determine the priori weights corresponding to multiple data based on multiple physical models; the weight calculation function unit is used to use a preset weight calculation function to calculate the weight corresponding to each data based on the spatiotemporal consistency indicators and priori weights corresponding to multiple data. The weight calculation function includes trained adjustment parameters.

3. The method for predicting the service life of a storage device according to claim 2, wherein: The spatiotemporal consistency calculation unit is configured to, when the data is two-dimensional data, divide the data into a plurality of data segments according to a preset time period, extract a feature vector of each data segment, calculate the similarity of the feature vectors of each group of two adjacent data segments, obtain a temporal similarity sequence, and calculate the spatiotemporal consistency index of the data based on the temporal similarity sequence; The spatiotemporal consistency calculation unit is used to divide the data into multiple spatial units according to a preset spatial unit size when the data is three-dimensional data, extract the feature vector of each spatial unit and calculate the similarity of the feature vectors of each group of two adjacent spatial units to obtain a spatial similarity sequence, and calculate the spatiotemporal consistency index of the data based on the spatial similarity sequence.

4. The method for predicting the service life of a storage device according to claim 1, wherein: The error type association module includes an error type feature library and a graph neural network unit, wherein the error type feature library includes error types, corresponding feature vectors, associated data types, and physical models, and the graph neural network unit is used to use the feature vectors in the error type feature library as edge weights and construct a multimodal feature interaction graph based on the multiple aligned feature vectors, wherein a node in the multimodal feature interaction graph is one of the aligned feature vectors; The graph neural network unit is also used to use a graph convolution layer to aggregate information of adjacent nodes in the multimodal feature interaction graph and update the alignment feature vector represented by each node to obtain multiple associated feature vectors, which contain constraint information of the error type.

5. The method for predicting the service life of a storage device according to claim 1, wherein: The life prediction decision layer is further used to input the fused feature vector into a fully connected layer, input the vector output by the fully connected layer into an activation layer, and obtain the probability of each error type occurring in the storage device output by the activation layer.

6. The method for predicting the service life of a storage device according to claim 1, wherein: The calibration layer is also used to obtain the device type, usage scenario type and usage duration of the storage device, obtain the corresponding calibration coefficient from the calibration error information based on the device type, the usage scenario type and the usage duration, and calibrate the initial predicted service life of the storage device based on the calibration coefficient to obtain the final predicted service life.

7. The method for predicting the service life of a storage device according to claim 6, wherein: The calibration error information is a calibration error table representing a multi-dimensional mapping relationship among device type, usage scenario type, usage duration and calibration coefficient determined based on training data. In the calibration error table, a combination of a device type, a usage scenario type and a duration interval of usage duration corresponds to one calibration coefficient.

8. The method for predicting the service life of a storage device according to claim 1, wherein: The method further comprises: Controlling the three-dimensional scanning device to scan the storage device to obtain three-dimensional point cloud data of the storage device; The step of inputting the internal structure image, the sound data, the electrical data, the temperature data, and the error monitoring data into a trained multimodal service life prediction model includes: The internal structure image, the sound data, the electrical data, the temperature data, the error monitoring data and the three-dimensional point cloud data are input into a trained multimodal service life prediction model.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the storage device service life prediction method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an executable program, and the executable program is executed by a processor to implement the storage device service life prediction method according to any one of claims 1 to 8.

Images