A solid state hard disk health status monitoring method and system

Abstract

The present invention discloses a solid-state drive health status monitoring method and system, which relates to the technical field of solid-state drives. The method includes: collecting monitoring data of the solid-state drive, extracting write characteristics from the monitoring data and performing fluctuation analysis; identifying the excessive write load characteristics of the solid-state drive based on the fluctuation analysis results of the write characteristics; analyzing the heat transfer characteristics built into the solid-state drive based on the monitoring data, and analyzing the correlation between the heat transfer characteristics and the excessive write load to obtain abnormal temperature characteristics when the excessive write load occurs; using the abnormal temperature characteristics as input to a pre-built solid-state drive health status assessment model, predicting the health status of the solid-state drive through the health status assessment model, and formulating a data migration mechanism based on the health status. The present invention can clearly determine the write volume per unit time and its fluctuation, further analyze the changes in write frequency, and effectively optimize storage management strategies by identifying these excessive write load characteristics.

Description

Technical Field

[0001] The present invention relates to the technical field of solid-state hard disks, and in particular to a method and system for monitoring the health status of solid-state hard disks. Background Art

[0002] A solid-state drive (SSD) is a computer storage device based on flash memory (typically NAND flash) as a storage medium, designed to replace traditional mechanical hard disk drives (HDDs). Compared to traditional hard disks, SSDs lack any moving parts, resulting in higher read and write speeds, lower power consumption, greater durability, and lower noise. SSDs store data using memory chips (NAND flash memory). These chips consist of multiple memory cells, each of which stores binary information (0 or 1).

[0003] Monitoring the health of solid-state drives (SSDs), particularly their health, temperature, and write volume, is crucial. This monitoring can effectively predict drive failures and prevent data loss. By monitoring temperature, read / write errors, bad blocks, and other information, users can identify potential drive issues such as overheating and wear early on, allowing them to take measures like migrating data and reducing load to extend the drive's lifespan and safeguard data security. Failure to monitor SSD health can result in failures not being detected promptly, increasing the risk of data loss or damage. For example, overheating can lead to performance degradation or even permanent damage, while frequent writes can wear out NAND flash memory cells, potentially missing the opportunity to replace the drive.

[0004] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0005] In response to the problems in the related art, the present invention proposes a solid state drive health status monitoring method and system to overcome the above technical problems existing in the existing related art.

[0006] To this end, the specific technical solutions adopted in the present invention are as follows:

[0007] According to one aspect of the present invention, a method for monitoring the health status of a solid-state drive is provided. The method:

[0008] Collect monitoring data of the solid-state drive, extract write characteristics from the monitoring data and perform fluctuation analysis. Based on the fluctuation analysis results of the write characteristics, identify the excessive write load characteristics of the solid-state drive;

[0009] Analyze the built-in heat transfer characteristics of the solid-state drive based on monitoring data, and analyze the correlation between the heat transfer characteristics and excessive write load to obtain abnormal temperature characteristics when excessive write load occurs;

[0010] Abnormal temperature characteristics are used as input to a pre-built solid-state drive health status assessment model. The health status of the solid-state drive is predicted by the health status assessment model, and a data migration mechanism is formulated based on the health status.

[0011] Preferably, collecting monitoring data of the solid-state drive, extracting write features from the monitoring data and performing fluctuation analysis, and identifying excessive write load features of the solid-state drive based on the fluctuation analysis results of the write features include:

[0012] Use hard disk monitoring technology to collect monitoring data of solid-state hard disks, including write volume, write frequency, erase and write cycles, and temperature parameters;

[0013] Perform time series mining analysis on the write volume and write frequency of fixed hard disks to obtain the write characteristics of solid-state drives. Perform Fourier transform analysis on the write characteristics to obtain the fluctuation analysis results of the write characteristics.

[0014] The fluctuation analysis results of the write features are input into a predefined variational autoencoder to learn the normal pattern of the write features. The difference pattern generated by the reconstruction error of the variational autoencoder is compared with the normal pattern to identify the excessive write load characteristics of the solid-state drive from the write features.

[0015] Preferably, the write amount and write frequency of the fixed hard disk are respectively subjected to time series mining analysis to obtain the write characteristics of the solid state hard disk, and the write characteristics are subjected to Fourier transform analysis to obtain the fluctuation analysis results of the write characteristics, including:

[0016] A tree indexing technique is used to obtain similar time series of write volume and write frequency of a fixed hard disk within a preset period, and weights are assigned to the similar time series respectively;

[0017] Sort similar time series by weight, select the write amount and write frequency combination within the preset sequence range to generate a write feature sequence;

[0018] Performing Fourier transform on the written feature sequence, and identifying the frequency components of the written feature sequence based on the frequency fluctuation trend of the written feature sequence;

[0019] A time series data curve is established according to the frequency component of the write feature sequence, a weight is assigned to the write feature sequence based on the interest level of the time series data curve, and the time series points of the write feature sequence are calculated to obtain the fluctuation analysis results of the write feature timing.

[0020] Preferably, performing Fourier transform on the written feature sequence and identifying the frequency components of the written feature sequence based on the frequency fluctuation trend of the written feature sequence includes:

[0021] Storing the continuous writing amount and writing frequency of the writing feature sequence into a predefined channel stream buffer;

[0022] Each time point of several written features is used as an amplitude component, and the phase component is sequentially inserted between each adjacent amplitude component to obtain a composite value of the channel stream buffer;

[0023] Using multiple composite values in the time domain of the channel stream buffer as input for time decimation, transforming the time domain signal using the time decimation method to obtain multiple composite values in the frequency domain;

[0024] Calculate the modulus of the complex value in the frequency domain, convert the amplitude value into decibel value to obtain the amplitude value of several discrete frequency lines, and use the frequency corresponding to the discrete frequency line as the input in the next time period;

[0025] The calculation process of the composite values in the time domain is iteratively performed until the processed composite values are aggregated to obtain the frequency components written into the characteristic sequence.

[0026] Preferably, the expression of the timing point of writing the feature sequence is:

[0027] ;

[0028] Where, D Indicates the timing point of writing the feature sequence; A Represents a similar time series; WA Represents the weight of similar time series; Q Indicates writing feature sequence; WQ Represents the weight of writing feature sequence.

[0029] Preferably, the built-in heat transfer characteristics of the solid-state drive are analyzed based on the monitoring data, and the correlation between the heat transfer characteristics and the excessive write load is analyzed to obtain abnormal temperature characteristics during excessive write load, including:

[0030] The built-in area of the solid-state drive is divided, and the temperature rise rate of each area in the solid-state drive in each time window is calculated based on the temperature parameters;

[0031] A temperature gradient transfer mechanism is established based on the temperature rise rate, and combined with the physical structure of the SSD and the thermal conductivity of the materials in each area, the heat transfer characteristics inside the SSD are extracted;

[0032] Calculating the correlation between the heat transfer feature and the excessive write load using the Pearson correlation coefficient, and comparing the correlation with a preset threshold;

[0033] If the correlation degree is greater than or equal to the prediction threshold, it means that the temperature rise rate increases abnormally when the write load is excessive, and the load balancing algorithm is used to regulate the temperature parameters of each area built into the solid-state drive. Otherwise, it means that the temperature rise rate is not affected when the write load is excessive.

[0034] Preferably, the calculation formula of the Pearson correlation coefficient is:

[0035] ;

[0036] Where, r represents the Pearson correlation coefficient; represents the mean value of heat transfer characteristics; Indicates excessive write load; Indicates the i heat transfer characteristics; Indicates the i Excessive write load; Indicates the weight value; n Represents the total number of features.

[0037] Preferably, a temperature gradient transfer mechanism is established based on the temperature rise rate, and combined with the physical structure of the solid-state drive and the thermal conductivity of the materials in each region, the heat transfer characteristics of the heat extracted in the solid-state drive include:

[0038] Using finite element analysis software to establish a finite element model of the solid state drive, and the finite element model includes the layout and contact interfaces of various regions of the solid state drive;

[0039] Define the physical properties and boundary conditions of each region of the SSD in the finite element model, use finite element analysis software to perform heat conduction simulation, obtain the temperature distribution of the SSD under normal working conditions, and obtain the preliminary temperature field of the SSD;

[0040] Based on the preliminary temperature field of the solid-state drive, the heat transfer path of the solid-state drive's built-in heat in each area is generated. The fitness of each heat transfer path is estimated using a genetic algorithm to select the optimal heat transfer path.

[0041] The temperature distribution of each area in the solid-state drive under the optimal heat transfer path is obtained, and based on the temperature distribution, the temperature field of the solid-state drive is updated using finite element analysis software to obtain the heat transfer characteristics built into the solid-state drive.

[0042] Preferably, using a genetic algorithm to estimate the fitness of each heat transfer path and screen out the optimal heat transfer path includes:

[0043] Encode the heat transfer path of the solid-state drive into individuals to obtain several initial individuals;

[0044] Define a fitness function and evaluate the initial individuals based on the fitness function. Select individuals with high fitness to enter the next generation based on the fitness evaluation results and the predefined screening strategy.

[0045] Cross the selected individuals to generate new individuals, and mutate the individuals after crossover, introducing small random changes to avoid falling into local optimality;

[0046] The fitness of the new generation of individuals is evaluated, the individuals in the population are updated, and the crossover and mutation process is iteratively performed until the termination condition is reached, and the heat transfer path with the best fitness is obtained.

[0047] According to another aspect of the present invention, a solid state drive health status monitoring system is provided, the system comprising:

[0048] A write load feature recognition module is used to collect monitoring data of the solid-state drive, extract write features from the monitoring data and perform fluctuation analysis, and identify excessive write load features of the solid-state drive based on the fluctuation analysis results of the write features;

[0049] A temperature characteristic analysis module is used to analyze the heat transfer characteristics of the solid-state drive based on the monitoring data, and analyze the correlation between the heat transfer characteristics and excessive write load to obtain abnormal temperature characteristics under excessive write load;

[0050] The health status assessment module is used to use abnormal temperature characteristics as input to a pre-built solid-state drive health status assessment model, predict the health status of the solid-state drive through the health status assessment model, and formulate a data migration mechanism based on the health status.

[0051] The beneficial effects of the present invention are:

[0052] 1. The present invention helps identify trends and anomalies in write operations by collecting data including write volume, write frequency, temperature, etc. By extracting write characteristics, it is possible to clearly identify the write volume per unit time and its fluctuations, and further analyze changes in write frequency. By identifying these excessive write load characteristics, it is possible to effectively optimize storage management strategies, reduce unnecessary write operations, and extend the service life of the hard disk. Excessive writing not only accelerates the wear of storage cells in SSDs and reduces the life of the hard disk, but may also lead to performance degradation and the risk of data loss. Timely identification and processing can help prevent this from happening. In addition, identifying excessive write load characteristics can also improve device performance and avoid problems caused by performance bottlenecks and heat accumulation.

[0053] 2. The present invention identifies the path and pattern of heat transfer by analyzing the heat transfer process inside the hard disk. By analyzing the characteristics of heat transfer, it can help identify these high-temperature areas and the direction of heat flow, and reveal the heat accumulation phenomenon caused by excessive write load. Further analysis of the correlation between heat transfer characteristics and excessive write load shows that when excessive write load is applied, certain areas of the hard disk may produce abnormal temperature fluctuations due to continuous high-frequency write operations. Therefore, by identifying the abnormal temperature characteristics under excessive write load, it can provide an important basis for the temperature control and heat dissipation design of the hard disk, optimize the heat dissipation system of the hard disk, avoid failures caused by overheating, and effectively extend the service life of the hard disk.

[0054] 3. The present invention uses abnormal temperature characteristics as input and predicts the health status of the hard disk through a pre-built solid-state hard disk health status assessment model. It can effectively identify the potential failure risks of the hard disk in advance, so as to discover performance degradation or damage problems that may be caused by temperature abnormalities and take timely measures. Based on the results of the health status assessment, a data migration mechanism can be formulated to migrate data to another hard disk with better health status, thereby avoiding data loss or hard disk damage. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Picture 1 is a flow chart of a method for monitoring the health status of a solid-state drive according to an embodiment of the present invention;

[0057] Picture 2 This is a principle block diagram of a solid state drive health status monitoring system according to an embodiment of the present invention.

[0058] In the picture:

[0059] 1. Write load feature recognition module; 2. Temperature feature analysis module; 3. Health status assessment module. DETAILED DESCRIPTION

[0060] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and the advantages of the present invention.

[0061] According to an embodiment of the present invention, a method and system for monitoring the health status of a solid-state drive are provided.

[0062] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Picture 1 As shown, according to the solid state hard disk health status monitoring method of an embodiment of the present invention, the method:

[0063] S1. Collect monitoring data of the solid-state drive, extract write features from the monitoring data and perform fluctuation analysis, and identify excessive write load features of the solid-state drive based on the fluctuation analysis results of the write features.

[0064] The monitoring data of the solid-state drive is collected, and the write characteristics are extracted from the monitoring data and fluctuation analysis is performed. Based on the fluctuation analysis results of the write characteristics, the excessive write load characteristics of the solid-state drive are identified, including:

[0065] Hard disk monitoring technology is used to collect monitoring data of solid-state hard disks, including write volume, write frequency, erase and write cycles, and temperature parameters.

[0066] It should be noted that the monitoring data of the solid-state hard disk collected by hard disk monitoring technology includes the write amount, write frequency, erase and write cycle and temperature parameters including:

[0067] Step 1: Build a hard drive monitoring system: Use hard drive monitoring tools or APIs (such as smartctl, CrystalDiskInfo, HWMonitor, or monitor through custom scripts and operating system interfaces). These tools can access the health status and performance indicators of hard drives through SMART (Self-Monitoring, Analysis, and Reporting Technology) data.

[0068] Step 2: Data collection:

[0069] Collect write data: Use the "Total Host Writes" indicator in the SMART metrics or other compatible items to obtain the amount of data written to the drive. This is usually expressed in bytes, megabytes (MB), or gigabytes (GB). Collect the write data generated by the drive within a specified time period (such as hourly or daily), recording the cumulative value of each write operation.

[0070] Collect write frequency: Write frequency usually refers to the number of writes per unit time. It can be obtained by real-time monitoring of the hard disk's I / O operations or read / write times per second (IOPS, Input / Output Operations Per Second).

[0071] Collecting Erase Cycles: Erase cycles refer to the number of times an SSD block is completely erased and rewritten. After reaching a certain number of erase cycles, an SSD's performance and durability degrade. Erase cycles can be obtained using the "Program / Erase Cycle" function in SMART. By monitoring changes in erase cycles and recording the number of times each SSD block is erased, you can understand the drive's durability and health.

[0072] Collecting temperature parameters: SSDs typically have built-in temperature sensors that report the drive's operating temperature in real time. Hard drive monitoring tools (such as smartctl) can directly read the drive's temperature data. It's important to collect and record the drive's operating temperature, including maximum and current temperatures, and track temperature fluctuations to prevent high temperatures from impacting drive performance.

[0073] Step 3: Data processing and storage:

[0074] Data Storage: Store monitoring data (write volume, write frequency, erase / write cycles, and temperature) in chronological order in a database or local storage. SQL databases (such as MySQL and PostgreSQL) or NoSQL databases (such as MongoDB) can be used to store this time series data.

[0075] Data cleaning and filtering: Clean the collected data to remove errors or outliers. Ensure data accuracy, filter out invalid data points, and set thresholds and alarm mechanisms to prevent extreme values from interfering with data analysis.

[0076] The write volume and write frequency of the fixed hard disk are analyzed through time series mining to obtain the write characteristics of the solid-state drive. The write characteristics are then analyzed through Fourier transform to obtain the fluctuation analysis results of the write characteristics.

[0077] Among them, the write volume and write frequency of the fixed hard disk are analyzed by time series mining to obtain the write characteristics of the solid-state hard disk. The write characteristics are then analyzed by Fourier transform to obtain the fluctuation analysis results of the write characteristics, including:

[0078] A tree indexing technique is used to obtain similar time series of write volume and write frequency of a fixed hard disk within a preset period, and weights are assigned to the similar time series respectively.

[0079] It should be noted that the tree indexing technology is used to obtain similar time series of write volume and write frequency of a fixed hard disk within a preset period, and weights are assigned to the similar time series respectively, including:

[0080] Step 1: Create a tree index structure:

[0081] Tree-based indexes such as R-trees, KD-trees, and quadtrees can be used to index high-dimensional data. Here, you can choose R-trees or KD-trees, as they are particularly well-suited for spatial indexing of time series data and can effectively organize and query similar time series data.

[0082] Build index:

[0083] The collected time series data of write volume and write frequency are used to construct a tree index based on timestamps. Each time series data point (such as write volume and frequency) is considered a node in the tree.

[0084] The tree structure can efficiently query and retrieve similar time series data, thereby quickly finding sequences that are similar to the current time series data within a time period.

[0085] Step 2: Similar time series retrieval:

[0086] Use methods such as dynamic time warping (DTW), Euclidean distance, and cosine similarity to define the similarity between time series data. DTW is a commonly used similarity metric in time series analysis that can measure the nonlinear matching of time series. For time series data with high write volume and write frequency, selecting an appropriate similarity metric can measure the similarity of data sequences within different time periods.

[0087] Retrieving similar time series: Using the constructed tree index, we calculate similarity with the current time series (such as the amount of writes or frequency series for a certain period) and retrieve the most similar time series from the tree. These similar time series can help analyze long-term trends and patterns in the data.

[0088] Step 3: Assign weights to similar time series:

[0089] Calculate weights: Based on the similarity calculation results, assign a weight to each similar time series. The weight is usually proportional to the similarity: the more similar the sequence, the greater the weight.

[0090] In order to control the computational complexity and improve the accuracy of the model, a minimum similarity threshold can be set for similar sequences. Time series below this threshold will be ignored and will not participate in subsequent analysis.

[0091] Sort similar time series by weight, select the write amount and write frequency combination within the preset sequence range to generate a write feature sequence;

[0092] The written feature sequence is subjected to Fourier transform, and the frequency components of the written feature sequence are identified based on the frequency fluctuation trend of the written feature sequence.

[0093] The Fourier transform is performed on the written feature sequence, and the frequency components of the written feature sequence are identified based on the frequency fluctuation trend of the written feature sequence. The method includes:

[0094] The continuous writing amount and writing frequency of the writing feature sequence are stored in a predefined channel stream buffer.

[0095] It should be noted that a channel stream refers to a channel or stream for data transmission, especially in data stream processing. For example, in multi-channel audio processing, each channel can be considered an independent signal stream. A cache is an area for temporary data storage. The purpose of the cache is to increase the processing speed of the system and avoid frequent calculation or storage operations. If these features are to be stored in the "channel stream cache", these written features are stored as a sequence of time domain signals, and it is planned to perform a Fourier transform (or similar signal analysis) on this data to identify frequency features, which can then be used to optimize storage management or predict the health status of the hard drive.

[0096] Each time point of several written features is used as an amplitude component, and the phase component is sequentially inserted between each adjacent amplitude component to obtain a composite value of the channel stream buffer;

[0097] A plurality of complex values in the time domain in the channel stream buffer are used as the input of time decimation, and the time domain signal is transformed by using the time decimation method to obtain a plurality of complex values in the frequency domain.

[0098] It should be noted that several composite values in the time domain in the channel stream buffer are used as input for time decimation. The time domain signal is transformed using the time decimation method to obtain several composite values in the frequency domain, including:

[0099] Step 1: Create a channel stream cache:

[0100] Obtain a set of time series data representing the amount or frequency of disk writes (e.g., the amount or frequency of writes per second). This data is stored in the channel stream buffer. The data value at each time point can be represented as a complex number (complex form), where the amplitude represents the signal strength and the phase component represents the signal phase information.

[0101] Data representation:

[0102] The collected data can be composed of composite values in the time domain, where each data point can be a complex number. For example, the time series data points representing the hard disk write load x ( t ) is represented as a complex number: ,in A ( t ) represents the amplitude; Indicates phase.

[0103] Create a channel stream cache: collect a set of time domain data into the cache. Assume that each time point stores a complex value, and the cache stores one or more such time series sequences.

[0104] Step 2: Time selection operation:

[0105] Select some time series data extracted from the time domain signal (i.e., time decimation). This data will be used as input for frequency domain analysis. Time decimation usually refers to selecting a portion of the data from the entire time domain signal for Fourier transform processing. Decimation methods include:

[0106] 1. Select a fixed-length timing window:

[0107] Select continuous time segments (windows) in the channel stream buffer. Assume that the sampling frequency of the time domain signal is f s , then the length of a time window is T The corresponding number of samples is N = T × f s .

[0108] For example, you can select the hard disk write frequency data per second, or select data within a fixed period as a time series window for processing.

[0109] 2. Sliding Window:

[0110] Using the sliding window technique, the window is slid in fixed steps to select different parts of the signal in sequence. Each time a part of the data is selected, it will be processed as input.

[0111] 3. Selection interval:

[0112] The time interval for data sampling can be fixed (such as every second or every minute) or based on certain rules (for example, sampling when the write load reaches a certain threshold).

[0113] Step 3: Transform the time signal into a frequency domain signal:

[0114] After time decimation, the time-domain signal is converted to the frequency domain using the Discrete Fourier Transform (DFT), and specifically its efficient implementation, the Fast Fourier Transform (FFT). The FFT quickly computes the frequency-domain representation of the time-domain signal, yielding its spectrum.

[0115] FFT process:

[0116] Perform FFT transformation on the time domain data after time decimation to convert the time domain signal into a frequency domain signal.

[0117] The calculation result will be a series of complex values, where the modulus of each complex value represents the amplitude of the signal at a specific frequency, and the phase part of the complex number represents the phase information of the frequency component. The Fourier transform expression is:

[0118] ;

[0119] Where, f k represents the frequency point in the frequency domain, x ( t ) represents time domain data, N Indicates the length of the window, t Indicates time.

[0120] Step 4: Get the complex value in the frequency domain:

[0121] The frequency domain data obtained by FFT transformation can be expressed as several complex values. These complex values represent the different frequency components of the signal, including its amplitude and phase information. Specifically:

[0122] Frequency domain amplitude and phase: Each complex value in the frequency domain can be extracted from its amplitude | X ( f k )| and phase arg( X ( f k )); amplitude | X ( f k )| describes the intensity of the frequency component and reflects the energy of the signal at this frequency; the phase arg( X ( f k )) describes the phase information of this frequency component.

[0123] Spectrum: Plots the amplitude values of all frequency domain data points as a spectrum. The spectrum reflects the intensity distribution of the signal at each frequency.

[0124] Step 5: Analyze and use frequency domain data:

[0125] By analyzing the frequency distribution of the signal through the spectrum diagram, we can identify which frequency components are most significant in the hard drive's write load. This can be used to monitor the hard drive's operating mode and identify high-load periods or abnormal frequency patterns.

[0126] Frequency domain analysis helps extract certain specific frequency components in hard drive operation and further analyze whether these frequency components are related to hard drive failure, performance degradation, or overload.

[0127] We collected data on the write volume of a hard drive for 30 consecutive minutes at a sampling rate of 1 Hz, recording the write volume once per second. Assume that over a certain period of time, the write volume data (in GB / s) for the hard drive is: [1.2, 1.3, 1.5, 1.4, 1.6].

[0128] Time decimation: A 10-second window is selected for time decimation. Suppose the data from the 1st to the 10th second is selected as a time series window for FFT transformation.

[0129] Perform FFT: Apply FFT to the selected 10 seconds of data to obtain a complex value in the frequency domain. .

[0130] Frequency domain analysis: Calculates the magnitude and phase of complex values. For example, ,∣ X ( f k )|=3.6,arg( X ( f k ))=tan -1 (3 / 2), indicating the intensity and phase information of the frequency component.

[0131] Calculate the modulus of the complex value in the frequency domain, convert the amplitude value into decibel value to obtain the amplitude value of several discrete frequency lines, and use the frequency corresponding to the discrete frequency line as the input in the next time period;

[0132] The calculation process of the composite values in the time domain is iteratively performed until the processed composite values are aggregated to obtain the frequency components written into the characteristic sequence.

[0133] A time series data curve is established according to the frequency component of the write feature sequence, a weight is assigned to the write feature sequence based on the interest level of the time series data curve, and the time series points of the write feature sequence are calculated to obtain the fluctuation analysis results of the write feature timing.

[0134] The expression for the timing point of writing the feature sequence is:

[0135] ;

[0136] Where, D Indicates the timing point of writing the feature sequence; A Represents a similar time series; WA Represents the weight of similar time series; Q Indicates writing feature sequence; WQ Represents the weight of writing feature sequence.

[0137] The fluctuation analysis results of the write features are input into a predefined variational autoencoder to learn the normal pattern of the write features. The difference pattern generated by the reconstruction error of the variational autoencoder is compared with the normal pattern to identify the excessive write load characteristics of the solid-state drive from the write features.

[0138] It should be noted that the fluctuation analysis results of the write features are input into a predefined variational autoencoder to learn the normal pattern of the write features. The difference pattern generated by the variational autoencoder's reconstruction error is then compared with the normal pattern. The characteristics of excessive write load on the SSD identified from the write features include:

[0139] Step 1: Write Fluctuation Analysis: Calculate the fluctuation or change in write volume within each time window. For example, you can calculate the change in write volume per second or minute compared to the previous moment to measure the volatility of the write load.

[0140] For example, assuming the write volume sequence: Write Volume = [1.2, 1.3, 1.5, 1.4, 1.6], calculate its fluctuation (difference) as: .

[0141] Step 2: Feature extraction: Extract time domain features (such as mean, standard deviation, maximum, minimum, etc.) from the original data, and extract additional features (such as amplitude and periodicity of fluctuations) based on fluctuation analysis. These features are vectorized and input into the variational autoencoder for learning.

[0142] Step 3: Train a Variational Autoencoder (VAE): A variational autoencoder (VAE) is a deep learning model that learns the underlying distribution of data and automatically generates low-dimensional representations of features during training. In this scenario, the VAE is used to learn the normal pattern of hard drive write features, including:

[0143] 1. Build and configure the variational autoencoder:

[0144] Encoder: The encoder compresses the input written feature data into the latent space and learns the low-dimensional representation of the normal written features.

[0145] Decoder: The decoder reconstructs the input data from the representation in the latent space.

[0146] Regularization: The variational autoencoder regularizes the latent space through the KL divergence (Kullback-Leibler Divergence) to ensure that the learned latent representation has good distribution properties.

[0147] 2. Data input and training:

[0148] The write features obtained through fluctuation analysis and feature extraction are fed into the VAE, and the VAE model is trained using large-scale normal data (i.e., normal write features without excessive writes). The goal is to enable the model to learn the underlying distribution of normal write patterns.

[0149] Step 4: Reconstruction and error calculation:

[0150] After VAE training is complete, the model is able to reconstruct the input data through the decoder. In practical applications, new input data (including potential abnormal data) is fed into the VAE and the reconstruction error is calculated to identify excessive write load.

[0151] 1. Enter new write feature data:

[0152] During the detection process, the real-time write feature data collected is fed into the trained VAE. If the reconstruction error is large, it indicates that the current input data is significantly different from the normal pattern, possibly indicating an abnormality caused by excessive write load.

[0153] 2. Comparison with normal mode:

[0154] The reconstruction error is compared to the error threshold of a normal write pattern. If the error exceeds the preset threshold, it indicates that the input write data exhibits characteristics of excessive write load, triggering an alarm or further analysis.

[0155] Step 5: Identify excessive write load characteristics:

[0156] By analyzing reconstruction errors, excessive write load characteristics can be identified in write signatures. If the cumulative reconstruction error exceeds a certain threshold within a time window, it can be determined that the drive's current write load is abnormal, further identifying possible excessive write load characteristics. Once excessive write load is detected, the system can trigger an alarm, notifying the administrator to take appropriate measures (such as reducing the load and adjusting the write strategy).

[0157] S2. Analyze the built-in heat transfer characteristics of the solid-state drive based on the monitoring data, and analyze the correlation between the heat transfer characteristics and the excessive write load to obtain the abnormal temperature characteristics under the excessive write load.

[0158] The analysis of the built-in heat transfer characteristics of the solid-state drive based on monitoring data and the correlation between the heat transfer characteristics and excessive write loads revealed the following abnormal temperature characteristics during excessive write loads:

[0159] The built-in area of the solid-state drive is divided, and the temperature rise rate of each area in the solid-state drive in each time window is calculated based on the temperature parameters;

[0160] A temperature gradient transfer mechanism is established based on the temperature rise rate, and combined with the physical structure of the solid-state drive and the thermal conductivity of the materials in each area, the heat transfer characteristics built into the solid-state drive are extracted.

[0161] Among them, the temperature gradient transfer mechanism is established based on the temperature rise rate. Combined with the physical structure of the solid-state drive and the thermal conductivity of the materials in each area, the heat transfer characteristics of the heat extracted in the solid-state drive include:

[0162] Using finite element analysis software to establish a finite element model of the solid state drive, and the finite element model includes the layout and contact interfaces of various regions of the solid state drive;

[0163] The physical properties and boundary conditions of each area of the SSD are defined in the finite element model. The heat conduction simulation is performed using finite element analysis software to obtain the temperature distribution of the SSD under normal working conditions and obtain the preliminary temperature field of the SSD.

[0164] It should be noted that the physical properties and boundary conditions of each region of the SSD are defined in the finite element model. The finite element analysis software is used to perform heat conduction simulation to obtain the temperature distribution of the SSD under normal working conditions. The preliminary temperature field of the SSD includes:

[0165] Step 1: Create a geometric model of the SSD, including its internal structure and various components. Different areas may have different material properties:

[0166] 1. Determine the geometry of the hard drive:

[0167] The SSD's physical form factor typically includes components such as the housing, PCB, memory chips, and circuits. The HDD's geometric model should include the dimensions and relative positions of these different areas.

[0168] Segment each region: A hard drive may have different internal regions (e.g., memory chips, controller, heat sink, etc.), each with different material properties and thermal conductivity. Therefore, the geometry model needs to be segmented according to these regions.

[0169] 2. Create a 3D model:

[0170] Use finite element software (such as ANSYS, COMSOL, Abaqus, etc.) to create a 3D geometric model of the hard drive. You can create the model based on the actual hard drive design drawings or import it from a CAD model.

[0171] Different areas of the hard disk (such as memory chips, PCB, power modules, etc.) are divided into different grid units to form a finite element grid.

[0172] Step 2: Define material properties:

[0173] The physical properties of each region (such as thermal conductivity, specific heat capacity, density, etc.) should be defined according to the material type. Different materials have different thermal conductivities and thermal responses, which have a significant impact on the temperature distribution.

[0174] 1. Definition of material thermal properties:

[0175] Thermal conductivity: Describes the ability of heat to be transferred. Different materials (such as metals, silicon, plastics, etc.) have different thermal conductivities.

[0176] Specific heat capacity: The amount of heat required to raise a unit mass of a substance by 1°C.

[0177] Density: affects heat capacity calculation and heat conduction.

[0178] Thermal expansion coefficient: Considering the expansion effect of materials at high temperatures, it may be necessary to consider it in some simulations with high accuracy requirements.

[0179] The physical properties of these areas are defined based on the actual material choices for each part of the hard drive. For example:

[0180] Memory chip: silicon material, thermal conductivity is about 148W / m·K;

[0181] PCB (Printed Circuit Board): Glass fiber, epoxy resin composite material, thermal conductivity is about 0.2W / m·K;

[0182] Housing: Metal material with high thermal conductivity, usually aluminum or stainless steel.

[0183] 2. Material property assignment:

[0184] In the model, assign the material's thermophysical properties to the geometry of each region. For example:

[0185] The thermal conductivity of silicon is imparted to the memory chip area;

[0186] The PCB area imparts thermal conductivity to the glass fiber composite material;

[0187] The shell area gives the metal its thermal conductivity.

[0188] Step 3: Define boundary conditions:

[0189] In heat transfer simulations, boundary conditions are very important because they define how heat transfer interacts with the environment.

[0190] 1. Set external boundary conditions:

[0191] Convection boundary conditions: Given that SSDs are typically installed in enclosed environments, heat transfer between the enclosure and the outside air is often achieved through convection. Therefore, the convection heat transfer coefficient with the ambient air must be defined. Assuming minimal air flow, a standard convection heat transfer coefficient (typically 5-20 W / m²·K) can be used. If cooling devices such as fans are used, the coefficient may be higher.

[0192] Radiation boundary conditions: If radiation effects are considered (especially in high temperature environments), radiation heat transfer conditions need to be applied to the external surfaces. The calculation of radiation heat transfer usually depends on the surface emissivity of the object.

[0193] 2. Set internal boundary conditions:

[0194] Internal heat sources: The SSD's memory chips, controller, and other electronic components generate heat during operation. Internal heat sources need to be defined based on actual power consumption (such as power consumption during write and read operations). For example, memory chips may generate significant heat under high load.

[0195] Estimate the internal heat source from the hard drive's power consumption data (such as the performance indicators provided by the manufacturer).

[0196] Heat conduction model: Within each region, a heat conduction model for the SSD needs to be defined. Fourier's heat conduction law is usually used, which describes the heat conduction process based on the thermal conductivity and temperature gradient of the material. Its expression is: ,in q represents the heat flux density; k Indicates the thermal conductivity of the material; Represents the temperature gradient.

[0197] Step 4: Meshing: Meshing is a key step in finite element analysis, which directly affects the accuracy and computational efficiency of simulation results. It specifically includes:

[0198] 1. Gridding:

[0199] The SSD's geometric model is divided into a finite number of cells. Each cell can be a tetrahedron, hexahedron, or other shape, depending on the model's complexity and computational accuracy requirements. For areas requiring higher accuracy (such as heat-intensive memory chips or circuit boards), a finer mesh is required.

[0200] 2. Mesh quality check:

[0201] Ensure the quality of meshing and avoid irregular meshes (such as long strips of cells, very small cells, etc.), which can lead to calculation errors and instability.

[0202] Step 5: Heat conduction simulation and calculation:

[0203] 1. Select the appropriate simulation type:

[0204] Select either steady-state or transient heat transfer analysis. Steady-state analysis is suitable for temperature distribution under long-term stable operation, while transient analysis is suitable for temperature changes of the hard disk under different operating conditions.

[0205] 2. Run the simulation:

[0206] Run a heat conduction simulation to calculate the temperature distribution in different areas of the SSD.

[0207] The software calculates the temperature of each node (or element) based on the input physical properties, boundary conditions, and meshing.

[0208] 3. Temperature field result output:

[0209] Output simulation results, including temperature distribution diagrams of various areas of the hard disk.

[0210] The temperature values of different areas (such as memory chips, PCB, and casing) can be obtained to understand the thermal characteristics of the hard disk under normal working conditions.

[0211] Based on the preliminary temperature field of the solid-state drive, the heat transfer path of the built-in heat of the solid-state drive in each area is generated. The fitness of each heat transfer path is estimated using a genetic algorithm to screen out the optimal heat transfer path.

[0212] It should be noted that the heat transfer paths of the SSD's built-in heat in various areas generated based on the SSD's preliminary temperature field include:

[0213] The heat transfer path is the path from the heat source (such as memory chips, controllers, power modules, etc.) to the outside of the hard drive or other cooler areas. Heat is transferred between the hard drive and the external environment through three primary methods: conduction, convection, and radiation.

[0214] Heat conduction: Heat is transferred through molecular collisions within a material, typically occurring within solids. Areas of high thermal conductivity within SSDs (such as the metal casing, memory chips, and PCB) are the primary channels for heat transfer.

[0215] Convection: Heat transfer through the flow of gaseous or liquid fluids, typically between an SSD and the external environment (air or cooling system).

[0216] Radiation: Heat is transferred in the form of electromagnetic waves (radiation). Although radiation has little effect inside the hard drive, at higher temperatures, radiation can become an important way of transferring heat.

[0217] Among them, the fitness of each heat transfer path is estimated by using genetic algorithm, and the optimal heat transfer path is selected, including:

[0218] Encode the heat transfer path of the solid-state drive into individuals to obtain several initial individuals;

[0219] Define a fitness function and evaluate the initial individuals based on the fitness function. Select individuals with high fitness to enter the next generation based on the fitness evaluation results and the predefined screening strategy.

[0220] Cross the selected individuals to generate new individuals, and mutate the individuals after crossover, introducing small random changes to avoid falling into local optimality;

[0221] The fitness of the new generation of individuals is evaluated, the individuals in the population are updated, and the crossover and mutation process is iteratively performed until the termination condition is reached, and the heat transfer path with the best fitness is obtained.

[0222] The temperature distribution of each area in the solid-state drive under the optimal heat transfer path is obtained, and based on the temperature distribution, the temperature field of the solid-state drive is updated using finite element analysis software to obtain the heat transfer characteristics built into the solid-state drive.

[0223] The correlation between the heat transfer signature and the excessive write load is calculated using the Pearson correlation coefficient, and the correlation is compared with a preset threshold.

[0224] The calculation formula of Pearson correlation coefficient is:

[0225] ;

[0226] Where, r represents the Pearson correlation coefficient; represents the mean value of heat transfer characteristics; Indicates excessive write load; Indicates the i heat transfer characteristics; Indicates the i Excessive write load; Indicates the weight value; n Represents the total number of features.

[0227] If the correlation degree is greater than or equal to the prediction threshold, it means that the temperature rise rate increases abnormally when the write load is excessive, and the load balancing algorithm is used to regulate the temperature parameters of each area built into the solid-state drive. Otherwise, it means that the temperature rise rate is not affected when the write load is excessive.

[0228] It should be noted that the load balancing algorithm is used to control the temperature parameters of each area inside the solid-state drive, including:

[0229] Step 1: Determine excessive write load:

[0230] If the correlation (such as the relationship between load and temperature rise rate) is greater than or equal to the set prediction threshold, it indicates that there is excessive write load and the load balancing mechanism needs to be enabled.

[0231] Temperature Rise Rate: Calculates the current temperature rise rate based on real-time temperature changes. The temperature rise rate is the ratio of temperature change to time change. If the temperature rise rate increases abnormally, it indicates that the area may be overheating due to excessive write load.

[0232] Dynamic load adjustment: Based on the temperature rise rate and load data, the load distribution in each area is calculated and the load distribution is dynamically adjusted.

[0233] High-load areas: If the temperature in certain areas rises rapidly, it means that the load in these areas is too large and the load needs to be reduced or dispersed.

[0234] Low-load area: Transfer the load from the high-temperature area to the low-temperature area to alleviate the heat accumulation in the high-load area.

[0235] Step 2: Control mechanism:

[0236] 1. Data migration: Migrate data writing tasks from overloaded areas to less loaded areas. For example, when the temperature of a memory chip is too high, some data can be written to a less loaded area or cache.

[0237] Write scheduling: By optimizing the write scheduling algorithm, write operations are made more evenly distributed to avoid over-concentration of writes in a certain area. For example, a round-robin or priority-based scheduling algorithm can be used to distribute write tasks.

[0238] 2. Dynamic thermal control:

[0239] Temperature control strategy: By monitoring the temperature and load of each zone, the system dynamically adjusts the temperature control measures for each zone. For example, it increases heat dissipation in high-temperature zones or restricts write operations at excessively high temperatures to reduce the load in high-temperature zones.

[0240] Internal heat dissipation management: This control adjusts the heat sink's internal performance to prevent heat transfer from high-temperature areas to other areas. Active cooling measures can be implemented for high-temperature areas, such as activating fans and adjusting heat sinks.

[0241] 3. Load optimization objectives:

[0242] Thermal balance optimization: Through load shifting and scheduling, the heat load on the hard disk is evenly distributed, ensuring that the temperature changes in each area are within a safe range. By adjusting the load distribution, the temperature field is optimized.

[0243] Performance and Temperature Tradeoff: When optimizing loads, the balance between performance and temperature must be considered. For example, distributing too much load to cooler areas can reduce the temperature rise rate in hotter areas, but it may also affect the read and write speeds of the hard drive.

[0244] It's important to note that combining the temperature gradient transfer mechanism with finite element analysis and genetic algorithms can offer significant advantages in solving complex heat transfer problems. Finite element analysis (FEA) is a numerical calculation method commonly used to simulate heat transfer in complex structures. It can accurately calculate the temperature field distribution, helping to understand how heat flows in solid-state drives or other complex systems. Through finite element analysis, the geometric structure of a hard drive or device is discretized into small units, and the temperature change of each unit is calculated, resulting in global temperature field data.

[0245] However, FEA alone cannot directly optimize thermal design. Therefore, combining it with a genetic algorithm (GA) allows for further optimization. A GA is an optimization method that mimics natural selection and genetic inheritance. By simulating operations such as selection, crossover, and mutation, it can find the optimal solution among a large number of possible solutions. Combining FEA with a GA allows for multi-objective optimization of heat transfer problems.

[0246] Through genetic algorithm optimization, the optimal thermal management configuration during temperature gradient transfer can be explored, thermal efficiency can be improved and energy consumption can be reduced. At the same time, the service life of the equipment can be extended, and its stability and reliability under high-load working conditions can be ensured.

[0247] S3. Use the abnormal temperature characteristics as input to a pre-built solid-state drive health status assessment model, predict the health status of the solid-state drive through the health status assessment model, and formulate a data migration mechanism based on the health status.

[0248] It should be noted that abnormal temperature characteristics are used as input to the pre-built SSD health status assessment model. The health status assessment model is used to predict the health status of the SSD. The data migration mechanism based on the health status includes:

[0249] Step 1: Build a health status assessment model: The health status assessment model can predict the health status of the hard drive based on machine learning, statistical analysis, or deep learning methods. The model includes:

[0250] Supervised learning models: such as logistic regression, support vector machine (SVM), decision tree, etc.

[0251] Neural network models: such as deep neural networks (DNNs) or long short-term memory (LSTM) networks, are suitable for predicting time series data.

[0252] Hybrid model: Combines multiple different machine learning models to enhance prediction accuracy.

[0253] Step 2: Dataset preparation and training:

[0254] Feature selection: Select temperature-related features such as temperature rise rate, temperature peak, load fluctuation, and other hard drive health indicators (such as read / write error rate and power consumption).

[0255] Data labeling: Marks the health status based on historical data, usually labeled as "normal", "warning" or "fault".

[0256] Normal status: The hard disk temperature is normal and the temperature rise rate is within the expected range.

[0257] Warning status: The temperature is approaching the upper limit and the temperature rise rate is high.

[0258] Fault status: The hard drive is overheating and the temperature rise rate is abnormal, which may cause hard drive failure.

[0259] Training process:

[0260] The labeled dataset (including abnormal temperature features and health status labels) is used to train the model.

[0261] Use cross-validation and optimization algorithms (such as grid search and random search) to adjust model parameters and select the optimal model.

[0262] For example, a support vector machine (SVM) model is used for training, by selecting features such as the temperature rise rate and peak temperature from the input data, and training a classification model that can determine the health status of the hard drive based on the input temperature features.

[0263] Step 3: Health status prediction and analysis:

[0264] 1. Input abnormal temperature characteristics:

[0265] In actual use, when the solid-state drive is running, the temperature data of each area can be obtained in real time, and its temperature rise rate and temperature peak can be calculated.

[0266] Assume that the current temperature of a hard disk is 72°C and the temperature rise rate is 5°C / min.

[0267] These features are input into the trained health status assessment model, and the model will predict the health status of the hard drive based on the rules learned during training.

[0268] 2. Health status assessment:

[0269] Model output: The model outputs the health status of the hard disk, such as "normal", "warning", or "failure".

[0270] If the output is "Warning", it means that the hard disk temperature is close to or exceeds the safety threshold, and there is a risk of overheating.

[0271] If the output is "failure", it means that the hard disk has experienced serious temperature abnormalities and may fail.

[0272] For example, a threshold can be set so that when the health status assessment model outputs "Warning" or "Failure," the next load scheduling and data migration mechanism is immediately triggered. For example, if the temperature rise rate of a hard drive is 4°C / min and the model predicts the hard drive's health status is "Warning," the data migration mechanism is triggered.

[0273] It should be noted that the dynamic data migration mechanism based on health status includes:

[0274] Step 1: Triggering conditions for data migration:

[0275] High-temperature or overloaded areas: When the health assessment model detects a hard drive in a "warning" or "failure" state, it can initiate a data migration mechanism. This requires migrating data from overloaded or high-temperature areas to less-loaded or cooler areas.

[0276] For example, among multiple areas of an SSD (such as storage chips, cache, controller, etc.), areas with higher temperatures may cause overheating, so data needs to be migrated to areas with lower temperatures.

[0277] Step 2: Data migration strategy:

[0278] Priority-based data migration: Set priorities based on the load and temperature of each area of the hard drive. Prioritize data migration in areas with heavy loads and high temperatures.

[0279] For example, data from storage chips with high write loads and excessively high temperatures can be migrated to controllers or cache areas with lower temperatures.

[0280] Temperature-load joint optimization: In addition to the load, the temperature distribution of the hard disk can also be considered. A temperature-load joint optimization strategy is adopted to ensure that the load and temperature of each area are within a reasonable range after data migration.

[0281] Dynamic migration and cache management: During data migration, the hard drive's built-in cache management mechanism is combined to reduce the performance impact of migration while ensuring data security.

[0282] Step 3: Data migration process:

[0283] Migration strategy execution: When the hard drive health status is abnormal, data migration strategies are implemented. For example, all data in a storage area can be migrated to another area with lower temperature, or the workload of an overheating area can be reduced by distributing some write load to other areas.

[0284] Real-time monitoring and adjustment: Monitor the hard drive status after migration in real time and adjust the migration strategy to ensure hard drive health.

[0285] Step 4: Migration effect evaluation:

[0286] Temperature balance: After the migration is completed, monitor the temperature changes in each area to ensure temperature balance and avoid temperature concentration again.

[0287] Performance impact assessment: Evaluate the impact of data migration on hard drive performance to ensure that the migration operation does not significantly affect the read / write performance of the hard drive.

[0288] For example, we conducted experiments on solid-state drives, obtained temperature data under different load conditions, and trained a health status assessment model. The following are some hypothetical data and results:

[0289] Normal operating conditions: The temperature fluctuates between 50°C and 55°C, with a temperature rise rate of 2°C / min. The health status is predicted to be "normal."

[0290] Abnormal status: Under high load conditions, the temperature exceeds 70°C, the temperature rise rate reaches 5°C / min, and the health status prediction is "Warning".

[0291] Fault status: Continuous high load, temperature reaches 80°C, temperature rise rate is 8°C / min, and health status prediction is "fault".

[0292] When the hard drive health status reaches "Warning," the hard drive's storage area is relocated to a less-loaded, cooler area. For example, some data on the memory chip, previously located in a high-temperature area, is relocated to the controller area, which offers better cooling. This effectively reduces the temperature rise and prevents failures.

[0293] According to another embodiment of the present invention, Picture 2 As shown, a solid state drive health status monitoring system is also provided, the system comprising:

[0294] A write load feature recognition module 1 is configured to collect monitoring data of the solid-state drive, extract write features from the monitoring data, perform fluctuation analysis, and identify excessive write load features of the solid-state drive based on the fluctuation analysis results of the write features;

[0295] Temperature characteristic analysis module 2 is used to analyze the heat transfer characteristics of the solid-state drive based on the monitoring data, and analyze the correlation between the heat transfer characteristics and the excessive write load to obtain the abnormal temperature characteristics under the excessive write load;

[0296] The health status assessment module 3 is used to take the abnormal temperature characteristics as the input of the pre-built solid-state drive health status assessment model, predict the health status of the solid-state drive through the health status assessment model, and formulate a data migration mechanism based on the health status.

[0297] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for monitoring the health status of a solid-state drive, characterized in that: The method: Collect monitoring data of the solid-state drive, extract write characteristics from the monitoring data and perform fluctuation analysis. Based on the fluctuation analysis results of the write characteristics, identify the excessive write load characteristics of the solid-state drive; Analyze the built-in heat transfer characteristics of the solid-state drive based on monitoring data, and analyze the correlation between the heat transfer characteristics and excessive write load to obtain abnormal temperature characteristics when excessive write load occurs; Abnormal temperature characteristics are used as input to a pre-built SSD health status assessment model. The health status assessment model is used to predict the health status of the SSD, and a data migration mechanism is developed based on the health status. Extracting write features from monitoring data and performing fluctuation analysis includes: Perform time series mining analysis on the write volume and write frequency of fixed hard disks to obtain the write characteristics of solid-state drives. Perform Fourier transform analysis on the write characteristics to obtain the fluctuation analysis results of the write characteristics. The analysis of the heat transfer characteristics built into the solid-state drive based on the monitoring data includes: The built-in area of the solid-state drive is divided, and the temperature rise rate of each area in the solid-state drive in each time window is calculated based on the temperature parameters; A temperature gradient transfer mechanism is established based on the temperature rise rate. Combined with the physical structure of the SSD and the thermal conductivity of the materials in each area, the heat transfer characteristics within the SSD are extracted. Specifically, the following are included: Using finite element analysis software to establish a finite element model of the solid state drive, and the finite element model includes the layout and contact interfaces of various regions of the solid state drive; Define the physical properties and boundary conditions of each region of the SSD in the finite element model, use finite element analysis software to perform heat conduction simulation, obtain the temperature distribution of the SSD under normal working conditions, and obtain the preliminary temperature field of the SSD; Based on the preliminary temperature field of the SSD, the heat transfer paths of the SSD's internal heat in each area are generated. The fitness of each heat transfer path is estimated using a genetic algorithm to select the optimal heat transfer path. Specifically, the following steps are performed: Encode the heat transfer path of the solid-state drive into individuals to obtain several initial individuals; Define a fitness function and evaluate the initial individuals based on the fitness function. Select individuals with high fitness to enter the next generation based on the fitness evaluation results and the predefined screening strategy. Cross the selected individuals to generate new individuals, and mutate the individuals after crossover, introducing small random changes to avoid falling into local optimality; The fitness of the new generation of individuals is evaluated, the individuals in the population are updated, and the crossover and mutation process is iteratively performed until the termination condition is reached, and the heat transfer path with the best fitness is obtained; The temperature distribution of each area in the solid-state drive under the optimal heat transfer path is obtained, and based on the temperature distribution, the temperature field of the solid-state drive is updated using finite element analysis software to obtain the heat transfer characteristics built into the solid-state drive.

2. A solid state drive health status monitoring method according to claim 1, characterized in that: The monitoring data collected from the solid-state hard drive includes: Use hard disk monitoring technology to collect monitoring data of solid-state hard disks, including write volume, write frequency, erase and write cycle and temperature parameters; The identifying of the excessive write load characteristics of the solid-state drive based on the fluctuation analysis result of the write characteristics includes: The fluctuation analysis results of the write features are input into a predefined variational autoencoder to learn the normal pattern of the write features. The difference pattern generated by the reconstruction error of the variational autoencoder is compared with the normal pattern to identify the excessive write load characteristics of the solid-state drive from the write features.

3. The method for monitoring the health status of a solid-state drive according to claim 1, wherein: The writing amount and writing frequency of the fixed hard disk are respectively subjected to time series mining analysis to obtain the writing characteristics of the solid state hard disk, and the writing characteristics are subjected to Fourier transform analysis to obtain the fluctuation analysis results of the writing characteristics, including: A tree indexing technique is used to obtain similar time series of write volume and write frequency of a fixed hard disk within a preset period, and weights are assigned to the similar time series respectively; Sort similar time series by weight, select the write amount and write frequency combination within the preset sequence range to generate a write feature sequence; Performing Fourier transform on the written feature sequence, and identifying the frequency components of the written feature sequence based on the frequency fluctuation trend of the written feature sequence; A time series data curve is established according to the frequency component of the write feature sequence, a weight is assigned to the write feature sequence based on the interest level of the time series data curve, and the time points of the write feature sequence are calculated to obtain the fluctuation analysis results of the write feature timing.

4. The method for monitoring the health status of a solid-state drive according to claim 3, wherein: The performing Fourier transform on the written feature sequence and identifying the frequency components of the written feature sequence based on the frequency fluctuation trend of the written feature sequence includes: Storing the continuous writing amount and writing frequency of the writing feature sequence into a predefined channel stream buffer; Each time point of several written features is used as an amplitude component, and the phase component is sequentially inserted between each adjacent amplitude component to obtain a composite value of the channel stream buffer; Using multiple composite values in the time domain in the channel stream buffer as input for time decimation, transforming the time domain signal using the time decimation method to obtain multiple composite values in the frequency domain; Calculate the modulus of the complex value in the frequency domain, convert the amplitude value into decibel value to obtain the amplitude value of several discrete frequency lines, and use the frequency corresponding to the discrete frequency line as the input in the next time period; The calculation process of the composite values in the time domain is iteratively performed until the processed composite values are aggregated to obtain the frequency components written into the characteristic sequence.

5. The method for monitoring the health status of a solid-state drive according to claim 3, wherein: The expression of the timing point of writing the characteristic sequence is: ； Where, D Indicates the timing point of writing the feature sequence; A Represents a similar time series; WA Represents the weight of similar time series; Q Indicates writing feature sequence; WQ Represents the weight of writing feature sequence.

6. The method for monitoring the health status of a solid-state drive according to claim 1, wherein: The analysis of the correlation between the heat transfer characteristics and the excessive write load to obtain the abnormal temperature characteristics under the excessive write load includes: Calculating the correlation between the heat transfer feature and the excessive write load using the Pearson correlation coefficient, and comparing the correlation with a preset threshold; If the correlation degree is greater than or equal to the prediction threshold, it means that the temperature rise rate increases abnormally when the write load is excessive, and the load balancing algorithm is used to regulate the temperature parameters of each area built into the solid-state drive. Otherwise, it means that the temperature rise rate is not affected when the write load is excessive.

7. A solid state drive health status monitoring method according to claim 6, characterized in that: The calculation formula of the Pearson correlation coefficient is: ； Where, r represents the Pearson correlation coefficient; represents the mean value of heat transfer characteristics; Indicates excessive write load; Indicates the i heat transfer characteristics; Indicates the i Excessive write load; Indicates the weight value; n Represents the total number of features.

8. A solid state drive health status monitoring system, configured to implement the solid state drive health status monitoring method according to any one of claims 1 to 7, characterized in that: The system includes: A write load feature recognition module is used to collect monitoring data of the solid-state drive, extract write features from the monitoring data and perform fluctuation analysis, and identify excessive write load features of the solid-state drive based on the fluctuation analysis results of the write features; A temperature characteristic analysis module is used to analyze the heat transfer characteristics of the solid-state drive based on the monitoring data, and analyze the correlation between the heat transfer characteristics and excessive write load to obtain abnormal temperature characteristics under excessive write load; The health status assessment module is used to use abnormal temperature characteristics as input to a pre-built solid-state drive health status assessment model, predict the health status of the solid-state drive through the health status assessment model, and formulate a data migration mechanism based on the health status.

Images